1 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2 Alan Hayward <alan.hayward@arm.com>
3 David Sherwood <david.sherwood@arm.com>
5 * fold-const.c (pointer_may_wrap_p): Take the offset as a
6 HOST_WIDE_INT rather than a poly_int64.
8 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
9 Alan Hayward <alan.hayward@arm.com>
10 David Sherwood <david.sherwood@arm.com>
12 * gimple-ssa-store-merging.c (symbolic_number::bytepos): Change from
13 HOST_WIDE_INT to poly_int64_pod.
14 (perform_symbolic_merge): Update accordingly.
15 (bswap_replace): Likewise.
17 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
18 Alan Hayward <alan.hayward@arm.com>
19 David Sherwood <david.sherwood@arm.com>
21 * tree-affine.h (aff_tree::offset): Change from widest_int
23 (wide_int_ext_for_comb): Delete.
24 (aff_combination_const, aff_comb_cannot_overlap_p): Take the
25 constants as poly_widest_int rather than widest_int.
26 (aff_combination_constant_multiple_p): Return the multiplier
28 (aff_combination_zero_p, aff_combination_singleton_var_p): Handle
30 * tree-affine.c (wide_int_ext_for_comb): Make original widest_int
31 version static and add an overload for poly_widest_int.
32 (aff_combination_const, aff_combination_add_cst)
33 (wide_int_constant_multiple_p, aff_comb_cannot_overlap_p): Take
34 the constants as poly_widest_int rather than widest_int.
35 (tree_to_aff_combination): Generalize INTEGER_CST case to
37 (aff_combination_to_tree): Track offsets as poly_widest_ints.
38 (aff_combination_add_product, aff_combination_mult): Handle
40 (aff_combination_constant_multiple_p): Return the multiplier
42 * tree-predcom.c (determine_offset): Return the offset as a
44 (split_data_refs_to_components, suitable_component_p): Update
46 (valid_initializer_p): Update call to
47 aff_combination_constant_multiple_p.
48 * tree-ssa-address.c (addr_to_parts): Handle polynomial offsets.
49 * tree-ssa-loop-ivopts.c (get_address_cost_ainc): Take the step
50 as a poly_int64 rather than a HOST_WIDE_INT.
51 (get_address_cost): Handle polynomial offsets.
52 (iv_elimination_compare_lt): Likewise.
53 (rewrite_use_nonlinear_expr): Likewise.
55 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
56 Alan Hayward <alan.hayward@arm.com>
57 David Sherwood <david.sherwood@arm.com>
59 * tree-dfa.h (get_addr_base_and_unit_offset_1): Return the offset
60 as a poly_int64_pod rather than a HOST_WIDE_INT.
61 (get_addr_base_and_unit_offset): Likewise.
62 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
63 (get_addr_base_and_unit_offset): Likewise.
64 * doc/match-and-simplify.texi: Change off from HOST_WIDE_INT
65 to poly_int64 in example.
66 * fold-const.c (fold_binary_loc): Update call to
67 get_addr_base_and_unit_offset.
68 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
69 (maybe_canonicalize_mem_ref_addr): Likewise.
70 (gimple_fold_stmt_to_constant_1): Likewise.
71 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
73 * ipa-param-manipulation.c (ipa_modify_call_arguments): Likewise.
75 * omp-low.c (lower_omp_target): Likewise.
76 * tree-sra.c (build_ref_for_offset): Likewise.
77 (build_debug_ref_for_model): Likewise.
78 * tree-ssa-address.c (maybe_fold_tmr): Likewise.
79 * tree-ssa-alias.c (ao_ref_init_from_ptr_and_size): Likewise.
80 * tree-ssa-ccp.c (optimize_memcpy): Likewise.
81 * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Likewise.
82 (constant_pointer_difference): Likewise.
83 * tree-ssa-loop-niter.c (expand_simple_operations): Likewise.
84 * tree-ssa-phiopt.c (jump_function_from_stmt): Likewise.
85 * tree-ssa-pre.c (create_component_ref_by_pieces_1): Likewise.
86 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
87 (vn_reference_maybe_forwprop_address, vn_reference_lookup_3): Likewise.
88 (set_ssa_val_to): Likewise.
89 * tree-ssa-strlen.c (get_addr_stridx, addr_stridxptr)
90 (maybe_diag_stxncpy_trunc): Likewise.
91 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
92 * tree.c (build_simple_mem_ref_loc): Likewise.
93 (array_at_struct_end_p): Likewise.
95 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
96 Alan Hayward <alan.hayward@arm.com>
97 David Sherwood <david.sherwood@arm.com>
99 * tree-dfa.h (get_ref_base_and_extent): Return the base, size and
100 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
101 (get_ref_base_and_extent_hwi): Declare.
102 * tree-dfa.c (get_ref_base_and_extent): Return the base, size and
103 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
104 (get_ref_base_and_extent_hwi): New function.
105 * cfgexpand.c (expand_debug_expr): Update call to
106 get_ref_base_and_extent.
107 * dwarf2out.c (add_var_loc_to_decl): Likewise.
108 * gimple-fold.c (get_base_constructor): Return the offset as a
109 poly_int64_pod rather than a HOST_WIDE_INT.
110 (fold_const_aggregate_ref_1): Track polynomial sizes and offsets.
111 * ipa-polymorphic-call.c
112 (ipa_polymorphic_call_context::set_by_invariant)
113 (extr_type_from_vtbl_ptr_store): Track polynomial offsets.
114 (ipa_polymorphic_call_context::ipa_polymorphic_call_context)
115 (check_stmt_for_type_change): Use get_ref_base_and_extent_hwi
116 rather than get_ref_base_and_extent.
117 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
118 * ipa-prop.c (ipa_load_from_parm_agg, compute_complex_assign_jump_func)
119 (get_ancestor_addr_info, determine_locally_known_aggregate_parts):
121 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Update
122 call to get_ref_base_and_extent.
123 * tree-sra.c (create_access, get_access_for_expr): Likewise.
124 * tree-ssa-alias.c (ao_ref_base, aliasing_component_refs_p)
125 (stmt_kills_ref_p): Likewise.
126 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
127 * tree-ssa-scopedtables.c (avail_expr_hash, equal_mem_array_ref_p):
129 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
130 Use get_ref_base_and_extent_hwi rather than get_ref_base_and_extent
131 when calling native_encode_expr.
132 * tree-ssa-structalias.c (get_constraint_for_component_ref): Update
133 call to get_ref_base_and_extent.
134 (do_structure_copy): Use get_ref_base_and_extent_hwi rather than
135 get_ref_base_and_extent.
136 * var-tracking.c (track_expr_p): Likewise.
138 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
139 Alan Hayward <alan.hayward@arm.com>
140 David Sherwood <david.sherwood@arm.com>
142 * ipa-param-manipulation.h (ipa_parm_adjustment::offset): Change from
143 HOST_WIDE_INT to poly_int64_pod.
144 * ipa-param-manipulation.c (ipa_modify_call_arguments): Track
145 polynomail parameter offsets.
147 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
148 Alan Hayward <alan.hayward@arm.com>
149 David Sherwood <david.sherwood@arm.com>
151 * gengtype.c (main): Handle poly_int64_pod.
152 * dwarf2out.h (dw_cfi_oprnd_cfa_loc): New dw_cfi_oprnd_type.
153 (dw_cfi_oprnd::dw_cfi_cfa_loc): New field.
154 (dw_cfa_location::offset, dw_cfa_location::base_offset): Change
155 from HOST_WIDE_INT to poly_int64_pod.
156 * dwarf2cfi.c (queued_reg_save::cfa_offset): Likewise.
157 (copy_cfa): New function.
158 (lookup_cfa_1): Use the cached dw_cfi_cfa_loc, if it exists.
159 (cfi_oprnd_equal_p): Handle dw_cfi_oprnd_cfa_loc.
160 (cfa_equal_p, dwarf2out_frame_debug_adjust_cfa)
161 (dwarf2out_frame_debug_cfa_offset, dwarf2out_frame_debug_expr)
162 (initial_return_save): Treat offsets as poly_ints.
163 (def_cfa_0): Likewise. Cache the CFA in dw_cfi_cfa_loc if either
164 offset is nonconstant.
165 (reg_save): Take the offset as a poly_int64. Fall back to
166 DW_CFA_expression for nonconstant offsets.
167 (queue_reg_save): Take the offset as a poly_int64.
168 * dwarf2out.c (dw_cfi_oprnd2_desc): Handle DW_CFA_def_cfa_expression.
170 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
171 Alan Hayward <alan.hayward@arm.com>
172 David Sherwood <david.sherwood@arm.com>
174 * rtl.h (operand_subword, operand_subword_force): Take the offset
175 as a poly_uint64 an unsigned int.
176 * emit-rtl.c (operand_subword, operand_subword_force): Likewise.
178 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
179 Alan Hayward <alan.hayward@arm.com>
180 David Sherwood <david.sherwood@arm.com>
182 * doc/rtl.texi: Update documentation of SUBREG_BYTE. Document the
183 'p' format code. Use INT_LIST rather than SUBREG as the example of
184 a code with an XINT and an XEXP. Remove the implication that
185 accessing an rtx field using XINT is expected to work.
186 * rtl.def (SUBREG): Change format from "ei" to "ep".
187 * rtl.h (rtunion::rt_subreg): New field.
188 (XCSUBREG): New macro.
189 (SUBREG_BYTE): Use it.
190 (subreg_shape): Change offset from an unsigned int to a poly_uint16.
191 Update constructor accordingly.
192 (subreg_shape::operator ==): Update accordingly.
193 (subreg_shape::unique_id): Return an unsigned HOST_WIDE_INT rather
194 than an unsigned int.
195 (subreg_lsb, subreg_lowpart_offset, subreg_highpart_offset): Return
196 a poly_uint64 rather than an unsigned int.
197 (subreg_lsb_1): Likewise. Take the offset as a poly_uint64 rather
198 than an unsigned int.
199 (subreg_size_offset_from_lsb, subreg_size_lowpart_offset)
200 (subreg_size_highpart_offset): Return a poly_uint64 rather than
201 an unsigned int. Take the sizes as poly_uint64s.
202 (subreg_offset_from_lsb): Return a poly_uint64 rather than
203 an unsigned int. Take the shift as a poly_uint64 rather than
205 (subreg_regno_offset, subreg_offset_representable_p): Take the offset
206 as a poly_uint64 rather than an unsigned int.
207 (simplify_subreg_regno): Likewise.
208 (byte_lowpart_offset): Return the memory offset as a poly_int64
210 (subreg_memory_offset): Likewise. Take the subreg offset as a
211 poly_uint64 rather than an unsigned int.
212 (simplify_subreg, simplify_gen_subreg, subreg_get_info)
213 (gen_rtx_SUBREG, validate_subreg): Take the subreg offset as a
214 poly_uint64 rather than an unsigned int.
215 * rtl.c (rtx_format): Describe 'p' in comment.
216 (copy_rtx, rtx_equal_p_cb, rtx_equal_p): Handle 'p'.
217 * emit-rtl.c (validate_subreg, gen_rtx_SUBREG): Take the subreg
218 offset as a poly_uint64 rather than an unsigned int.
219 (byte_lowpart_offset): Return the memory offset as a poly_int64
221 (subreg_memory_offset): Likewise. Take the subreg offset as a
222 poly_uint64 rather than an unsigned int.
223 (subreg_size_lowpart_offset, subreg_size_highpart_offset): Take the
224 mode sizes as poly_uint64s rather than unsigned ints. Return a
225 poly_uint64 rather than an unsigned int.
226 (subreg_lowpart_p): Treat subreg offsets as poly_ints.
227 (copy_insn_1): Handle 'p'.
228 * rtlanal.c (set_noop_p): Treat subregs offsets as poly_uint64s.
229 (subreg_lsb_1): Take the subreg offset as a poly_uint64 rather than
230 an unsigned int. Return the shift in the same way.
231 (subreg_lsb): Return the shift as a poly_uint64 rather than an
233 (subreg_size_offset_from_lsb): Take the sizes and shift as
234 poly_uint64s rather than unsigned ints. Return the offset as
236 (subreg_get_info, subreg_regno_offset, subreg_offset_representable_p)
237 (simplify_subreg_regno): Take the offset as a poly_uint64 rather than
239 * rtlhash.c (add_rtx): Handle 'p'.
240 * genemit.c (gen_exp): Likewise.
241 * gengenrtl.c (type_from_format, gendef): Likewise.
242 * gensupport.c (subst_pattern_match, get_alternatives_number)
243 (collect_insn_data, alter_predicate_for_insn, alter_constraints)
244 (subst_dup): Likewise.
245 * gengtype.c (adjust_field_rtx_def): Likewise.
246 * genrecog.c (find_operand, find_matching_operand, validate_pattern)
247 (match_pattern_2): Likewise.
248 (rtx_test::SUBREG_FIELD): New rtx_test::kind_enum.
249 (rtx_test::subreg_field): New function.
250 (operator ==, safe_to_hoist_p, transition_parameter_type)
251 (print_nonbool_test, print_test): Handle SUBREG_FIELD.
252 * genattrtab.c (attr_rtx_1): Say that 'p' is deliberately not handled.
253 * genpeep.c (match_rtx): Likewise.
254 * print-rtl.c (print_poly_int): Include if GENERATOR_FILE too.
255 (rtx_writer::print_rtx_operand): Handle 'p'.
256 (print_value): Handle SUBREG.
257 * read-rtl.c (apply_int_iterator): Likewise.
258 (rtx_reader::read_rtx_operand): Handle 'p'.
259 * alias.c (rtx_equal_for_memref_p): Likewise.
260 * cselib.c (rtx_equal_for_cselib_1, cselib_hash_rtx): Likewise.
261 * caller-save.c (replace_reg_with_saved_mem): Treat subreg offsets
263 * calls.c (expand_call): Likewise.
264 * combine.c (combine_simplify_rtx, expand_field_assignment): Likewise.
265 (make_extraction, gen_lowpart_for_combine): Likewise.
266 * loop-invariant.c (hash_invariant_expr_1, invariant_expr_equal_p):
268 * cse.c (remove_invalid_subreg_refs): Take the offset as a poly_uint64
269 rather than an unsigned int. Treat subreg offsets as poly_ints.
270 (exp_equiv_p): Handle 'p'.
271 (hash_rtx_cb): Likewise. Treat subreg offsets as poly_ints.
272 (equiv_constant, cse_insn): Treat subreg offsets as poly_ints.
273 * dse.c (find_shift_sequence): Likewise.
274 * dwarf2out.c (rtl_for_decl_location): Likewise.
275 * expmed.c (extract_low_bits): Likewise.
276 * expr.c (emit_group_store, undefined_operand_subword_p): Likewise.
277 (expand_expr_real_2): Likewise.
278 * final.c (alter_subreg): Likewise.
279 (leaf_renumber_regs_insn): Handle 'p'.
280 * function.c (assign_parm_find_stack_rtl, assign_parm_setup_stack):
281 Treat subreg offsets as poly_ints.
282 * fwprop.c (forward_propagate_and_simplify): Likewise.
283 * ifcvt.c (noce_emit_move_insn, noce_emit_cmove): Likewise.
284 * ira.c (get_subreg_tracking_sizes): Likewise.
285 * ira-conflicts.c (go_through_subreg): Likewise.
286 * ira-lives.c (process_single_reg_class_operands): Likewise.
287 * jump.c (rtx_renumbered_equal_p): Likewise. Handle 'p'.
288 * lower-subreg.c (simplify_subreg_concatn): Take the subreg offset
289 as a poly_uint64 rather than an unsigned int.
290 (simplify_gen_subreg_concatn, resolve_simple_move): Treat
291 subreg offsets as poly_ints.
292 * lra-constraints.c (operands_match_p): Handle 'p'.
293 (match_reload, curr_insn_transform): Treat subreg offsets as poly_ints.
294 * lra-spills.c (assign_mem_slot): Likewise.
295 * postreload.c (move2add_valid_value_p): Likewise.
296 * recog.c (general_operand, indirect_operand): Likewise.
297 * regcprop.c (copy_value, maybe_mode_change): Likewise.
298 (copyprop_hardreg_forward_1): Likewise.
299 * reginfo.c (simplifiable_subregs_hasher::hash, simplifiable_subregs)
300 (record_subregs_of_mode): Likewise.
301 * rtlhooks.c (gen_lowpart_general, gen_lowpart_if_possible): Likewise.
302 * reload.c (operands_match_p): Handle 'p'.
303 (find_reloads_subreg_address): Treat subreg offsets as poly_ints.
304 * reload1.c (alter_reg, choose_reload_regs): Likewise.
305 (compute_reload_subreg_offset): Likewise, and return an poly_int64.
306 * simplify-rtx.c (simplify_truncation, simplify_binary_operation_1):
307 (test_vector_ops_duplicate): Treat subreg offsets as poly_ints.
308 (simplify_const_poly_int_tests<N>::run): Likewise.
309 (simplify_subreg, simplify_gen_subreg): Take the subreg offset as
310 a poly_uint64 rather than an unsigned int.
311 * valtrack.c (debug_lowpart_subreg): Likewise.
312 * var-tracking.c (var_lowpart): Likewise.
313 (loc_cmp): Handle 'p'.
315 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
316 Alan Hayward <alan.hayward@arm.com>
317 David Sherwood <david.sherwood@arm.com>
319 * ira.c (get_subreg_tracking_sizes): New function.
320 (init_live_subregs): Take an integer size rather than a register.
321 (build_insn_chain): Use get_subreg_tracking_sizes. Update calls
322 to init_live_subregs.
324 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
325 Alan Hayward <alan.hayward@arm.com>
326 David Sherwood <david.sherwood@arm.com>
328 * expr.c (store_constructor_field): Change bitsize from a
329 unsigned HOST_WIDE_INT to a poly_uint64 and bitpos from a
330 HOST_WIDE_INT to a poly_int64.
331 (store_constructor): Change size from a HOST_WIDE_INT to
333 (store_field): Likewise bitsize and bitpos.
335 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
336 Alan Hayward <alan.hayward@arm.com>
337 David Sherwood <david.sherwood@arm.com>
339 * expmed.h (store_bit_field): Change bitregion_start and
340 bitregion_end from unsigned HOST_WIDE_INT to poly_uint64.
341 * expmed.c (adjust_bit_field_mem_for_reg, strict_volatile_bitfield_p)
342 (store_bit_field_1, store_integral_bit_field, store_bit_field)
343 (store_fixed_bit_field, store_split_bit_field): Likewise.
344 * expr.c (store_constructor_field, store_field): Likewise.
345 (optimize_bitfield_assignment_op): Likewise. Make the same change
346 to bitsize and bitpos.
347 * machmode.h (bit_field_mode_iterator): Change m_bitregion_start
348 and m_bitregion_end from HOST_WIDE_INT to poly_int64. Make the
349 same change in the constructor arguments.
350 (get_best_mode): Change bitregion_start and bitregion_end from
351 unsigned HOST_WIDE_INT to poly_uint64.
352 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
353 Change bitregion_start and bitregion_end from HOST_WIDE_INT to
355 (bit_field_mode_iterator::next_mode): Update for new types
356 of m_bitregion_start and m_bitregion_end.
357 (get_best_mode): Change bitregion_start and bitregion_end from
358 unsigned HOST_WIDE_INT to poly_uint64.
360 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
361 Alan Hayward <alan.hayward@arm.com>
362 David Sherwood <david.sherwood@arm.com>
364 * rtl.h (simplify_gen_subreg): Add a temporary overload that
365 accepts poly_uint64 offsets.
366 * expmed.h (extract_bit_field): Take bitsize and bitnum as
367 poly_uint64s rather than unsigned HOST_WIDE_INTs.
368 * expmed.c (lowpart_bit_field_p): Likewise.
369 (extract_bit_field_as_subreg): New function, split out from...
370 (extract_bit_field_1): ...here. Take bitsize and bitnum as
371 poly_uint64s rather than unsigned HOST_WIDE_INTs. For vector
372 extractions, check that BITSIZE matches the size of the extracted
373 value and that BITNUM is an exact multiple of that size.
374 If all else fails, try forcing the value into memory if
375 BITNUM is variable, and adjusting the address so that the
376 offset is constant. Split the part that can only handle constant
377 bitsize and bitnum out into...
378 (extract_integral_bit_field): ...this new function.
379 (extract_bit_field): Take bitsize and bitnum as poly_uint64s
380 rather than unsigned HOST_WIDE_INTs.
382 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
383 Alan Hayward <alan.hayward@arm.com>
384 David Sherwood <david.sherwood@arm.com>
386 * expmed.h (store_bit_field): Take bitsize and bitnum as
387 poly_uint64s rather than unsigned HOST_WIDE_INTs.
388 * expmed.c (simple_mem_bitfield_p): Likewise. Add a parameter
389 that returns the byte size.
390 (store_bit_field_1): Take bitsize and bitnum as
391 poly_uint64s rather than unsigned HOST_WIDE_INTs. Update call
392 to simple_mem_bitfield_p. Split the part that can only handle
393 constant bitsize and bitnum out into...
394 (store_integral_bit_field): ...this new function.
395 (store_bit_field): Take bitsize and bitnum as poly_uint64s rather
396 than unsigned HOST_WIDE_INTs.
397 (extract_bit_field_1): Update call to simple_mem_bitfield_p.
399 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
400 Alan Hayward <alan.hayward@arm.com>
401 David Sherwood <david.sherwood@arm.com>
403 * lra-int.h (lra_reg): Change offset from int to poly_int64.
404 (lra_insn_recog_data): Change sp_offset from HOST_WIDE_INT
406 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
407 update_sp_offset from a HOST_WIDE_INT to a poly_int64.
408 (lra_update_reg_val_offset, lra_reg_val_equal_p): Take the
409 offset as a poly_int64 rather than an int.
410 * lra-assigns.c (find_hard_regno_for_1): Handle poly_int64 offsets.
411 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
412 * lra-constraints.c (equiv_address_substitution): Track offsets
414 (emit_inc): Check poly_int_rtx_p instead of CONST_INT_P.
415 (curr_insn_transform): Handle the new form of sp_offset.
416 * lra-eliminations.c (lra_elim_table): Change previous_offset
417 and offset from HOST_WIDE_INT to poly_int64.
418 (print_elim_table, update_reg_eliminate): Update accordingly.
419 (self_elim_offsets): Change from HOST_WIDE_INT to poly_int64_pod.
420 (get_elimination): Update accordingly.
421 (form_sum): Check poly_int_rtx_p instead of CONST_INT_P.
422 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
423 update_sp_offset from a HOST_WIDE_INT to a poly_int64. Handle
424 poly_int64 offsets generally.
425 (curr_sp_change): Change from HOST_WIDE_INT to poly_int64.
426 (mark_not_eliminable, init_elimination): Update accordingly.
427 (remove_reg_equal_offset_note): Return a bool and pass the new
428 offset back by pointer as a poly_int64.
429 * lra-remat.c (change_sp_offset): Take sp_offset as a poly_int64
430 rather than a HOST_WIDE_INT.
431 (do_remat): Track offsets poly_int64s.
432 * lra.c (lra_update_insn_recog_data, setup_sp_offset): Likewise.
434 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
435 Alan Hayward <alan.hayward@arm.com>
436 David Sherwood <david.sherwood@arm.com>
438 * rtl.h (mem_attrs): Add a default constructor. Change size and
439 offset from HOST_WIDE_INT to poly_int64.
440 * emit-rtl.h (set_mem_offset, set_mem_size, adjust_address_1)
441 (adjust_automodify_address_1, set_mem_attributes_minus_bitpos)
442 (widen_memory_access): Take the sizes and offsets as poly_int64s
443 rather than HOST_WIDE_INTs.
444 * alias.c (ao_ref_from_mem): Handle the new form of MEM_OFFSET.
445 (offset_overlap_p): Take poly_int64s rather than HOST_WIDE_INTs
447 (adjust_offset_for_component_ref): Change the offset from a
448 HOST_WIDE_INT to a poly_int64.
449 (nonoverlapping_memrefs_p): Track polynomial offsets and sizes.
450 * cfgcleanup.c (merge_memattrs): Update after mem_attrs changes.
451 * dce.c (find_call_stack_args): Likewise.
452 * dse.c (record_store): Likewise.
453 * dwarf2out.c (tls_mem_loc_descriptor, dw_sra_loc_expr): Likewise.
454 * print-rtl.c (rtx_writer::print_rtx): Likewise.
455 * read-rtl-function.c (test_loading_mem): Likewise.
456 * rtlanal.c (may_trap_p_1): Likewise.
457 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
458 * var-tracking.c (int_mem_offset, track_expr_p): Likewise.
459 * emit-rtl.c (mem_attrs_eq_p, get_mem_align_offset): Likewise.
460 (mem_attrs::mem_attrs): New function.
461 (set_mem_attributes_minus_bitpos): Change bitpos from a
462 HOST_WIDE_INT to poly_int64.
463 (set_mem_alias_set, set_mem_addr_space, set_mem_align, set_mem_expr)
464 (clear_mem_offset, clear_mem_size, change_address)
465 (get_spill_slot_decl, set_mem_attrs_for_spill): Directly
466 initialize mem_attrs.
467 (set_mem_offset, set_mem_size, adjust_address_1)
468 (adjust_automodify_address_1, offset_address, widen_memory_access):
469 Likewise. Take poly_int64s rather than HOST_WIDE_INT.
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 * rtlanal.c (rtx_addr_can_trap_p_1): Take the offset and size
476 as poly_int64s rather than HOST_WIDE_INTs. Use a size of -1
477 rather than 0 to represent an unknown size. Assert that the size
478 is known when the mode isn't BLKmode.
479 (may_trap_p_1): Use -1 for unknown sizes.
480 (rtx_addr_can_trap_p): Likewise. Pass BLKmode rather than VOIDmode.
482 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
483 Alan Hayward <alan.hayward@arm.com>
484 David Sherwood <david.sherwood@arm.com>
486 * dse.c (store_info): Change offset and width from HOST_WIDE_INT
487 to poly_int64. Update commentary for positions_needed.large.
488 (read_info_type): Change offset and width from HOST_WIDE_INT
490 (set_usage_bits): Likewise.
491 (canon_address): Return the offset as a poly_int64 rather than
492 a HOST_WIDE_INT. Use strip_offset_and_add.
493 (set_all_positions_unneeded, any_positions_needed_p): Use
494 positions_needed.large to track stores with non-constant widths.
495 (all_positions_needed_p): Likewise. Take the offset and width
496 as poly_int64s rather than ints. Assert that rhs is nonnull.
497 (record_store): Cope with non-constant offsets and widths.
498 Nullify the rhs of an earlier store if we can't tell which bytes
500 (find_shift_sequence): Take the access_size and shift as poly_int64s
502 (get_stored_val): Take the read_offset and read_width as poly_int64s
503 rather than HOST_WIDE_INTs.
504 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Handle
505 non-constant offsets and widths.
507 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
508 Alan Hayward <alan.hayward@arm.com>
509 David Sherwood <david.sherwood@arm.com>
511 * inchash.h (inchash::hash::add_poly_int): New function.
512 * tree-ssa-alias.h (ao_ref::offset, ao_ref::size, ao_ref::max_size):
513 Use poly_int64 rather than HOST_WIDE_INT.
514 (ao_ref::max_size_known_p): New function.
515 * tree-ssa-sccvn.h (vn_reference_op_struct::off): Use poly_int64_pod
516 rather than HOST_WIDE_INT.
517 * tree-ssa-alias.c (ao_ref_base): Apply get_ref_base_and_extent
518 to temporaries until its interface is adjusted to match.
519 (ao_ref_init_from_ptr_and_size): Handle polynomial offsets and sizes.
520 (aliasing_component_refs_p, decl_refs_may_alias_p)
521 (indirect_ref_may_alias_decl_p, indirect_refs_may_alias_p): Take
522 the offsets and max_sizes as poly_int64s instead of HOST_WIDE_INTs.
523 (refs_may_alias_p_1, stmt_kills_ref_p): Adjust for changes to
525 * alias.c (ao_ref_from_mem): Likewise.
526 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
527 * tree-ssa-dse.c (valid_ao_ref_for_dse, normalize_ref)
528 (clear_bytes_written_by, setup_live_bytes_from_ref, compute_trims)
529 (maybe_trim_complex_store, maybe_trim_constructor_store)
530 (live_bytes_read, dse_classify_store): Likewise.
531 * tree-ssa-sccvn.c (vn_reference_compute_hash, vn_reference_eq):
532 (copy_reference_ops_from_ref, ao_ref_init_from_vn_reference)
533 (fully_constant_vn_reference_p, valueize_refs_1): Likewise.
534 (vn_reference_lookup_3): Likewise.
535 * tree-ssa-uninit.c (warn_uninitialized_vars): Likewise.
537 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
538 Alan Hayward <alan.hayward@arm.com>
539 David Sherwood <david.sherwood@arm.com>
541 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p)
542 (indirect_refs_may_alias_p): Use ranges_may_overlap_p
543 instead of ranges_overlap_p.
545 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
546 Alan Hayward <alan.hayward@arm.com>
547 David Sherwood <david.sherwood@arm.com>
549 * tree-ssa-alias.c (same_addr_size_stores_p): Take the offsets and
550 sizes as poly_int64s rather than HOST_WIDE_INTs.
552 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
553 Alan Hayward <alan.hayward@arm.com>
554 David Sherwood <david.sherwood@arm.com>
556 * gimple-fold.h (fold_ctor_reference): Take the offset and size
557 as poly_uint64 rather than unsigned HOST_WIDE_INT.
558 * gimple-fold.c (fold_ctor_reference): Likewise.
560 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
561 Alan Hayward <alan.hayward@arm.com>
562 David Sherwood <david.sherwood@arm.com>
564 * target.def (dwarf_poly_indeterminate_value): New hook.
565 * targhooks.h (default_dwarf_poly_indeterminate_value): Declare.
566 * targhooks.c (default_dwarf_poly_indeterminate_value): New function.
567 * doc/tm.texi.in (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Document.
568 * doc/tm.texi: Regenerate.
569 * dwarf2out.h (build_cfa_loc, build_cfa_aligned_loc): Take the
570 offset as a poly_int64.
571 * dwarf2out.c (new_reg_loc_descr): Move later in file. Take the
572 offset as a poly_int64.
573 (loc_descr_plus_const, loc_list_plus_const, build_cfa_aligned_loc):
574 Take the offset as a poly_int64.
575 (build_cfa_loc): Likewise. Use loc_descr_plus_const.
576 (frame_pointer_fb_offset): Change to a poly_int64.
577 (int_loc_descriptor): Take the offset as a poly_int64. Use
578 targetm.dwarf_poly_indeterminate_value for polynomial offsets.
579 (based_loc_descr): Take the offset as a poly_int64.
580 Use strip_offset_and_add to handle (plus X (const)).
581 Use new_reg_loc_descr instead of an open-coded version of the
582 previous implementation.
583 (mem_loc_descriptor): Handle CONST_POLY_INT.
584 (compute_frame_pointer_to_fb_displacement): Take the offset as a
585 poly_int64. Use strip_offset_and_add to handle (plus X (const)).
587 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
588 Alan Hayward <alan.hayward@arm.com>
589 David Sherwood <david.sherwood@arm.com>
591 * rtl.h (reg_attrs::offset): Change from HOST_WIDE_INT to poly_int64.
592 (gen_rtx_REG_offset): Take the offset as a poly_int64.
593 * inchash.h (inchash::hash::add_poly_hwi): New function.
594 * gengtype.c (main): Register poly_int64.
595 * emit-rtl.c (reg_attr_hasher::hash): Use inchash. Treat the
596 offset as a poly_int.
597 (reg_attr_hasher::equal): Use must_eq to compare offsets.
598 (get_reg_attrs, update_reg_offset, gen_rtx_REG_offset): Take the
599 offset as a poly_int64.
600 (set_reg_attrs_from_value): Treat the offset as a poly_int64.
601 * print-rtl.c (print_poly_int): New function.
602 (rtx_writer::print_rtx_operand_code_r): Treat REG_OFFSET as
604 * var-tracking.c (track_offset_p, get_tracked_reg_offset): New
606 (var_reg_set, var_reg_delete_and_set, var_reg_delete): Use them.
607 (same_variable_part_p, track_loc_p): Take the offset as a poly_int64.
608 (vt_get_decl_and_offset): Return the offset as a poly_int64.
609 Enforce track_offset_p for parts of a PARALLEL.
610 (vt_add_function_parameter): Use const_offset for the final
611 offset to track. Use get_tracked_reg_offset for the parts
614 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
615 Alan Hayward <alan.hayward@arm.com>
616 David Sherwood <david.sherwood@arm.com>
618 * target.def (truly_noop_truncation): Take poly_uint64s instead of
619 unsigned ints. Change default to hook_bool_puint64_puint64_true.
620 * doc/tm.texi: Regenerate.
621 * hooks.h (hook_bool_uint_uint_true): Delete.
622 (hook_bool_puint64_puint64_true): Declare.
623 * hooks.c (hook_bool_uint_uint_true): Delete.
624 (hook_bool_puint64_puint64_true): New function.
625 * config/mips/mips.c (mips_truly_noop_truncation): Take poly_uint64s
626 instead of unsigned ints.
627 * config/spu/spu.c (spu_truly_noop_truncation): Likewise.
628 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): Likewise.
630 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
631 Alan Hayward <alan.hayward@arm.com>
632 David Sherwood <david.sherwood@arm.com>
634 * optabs.h (expand_operand): Add an int_value field.
635 (create_expand_operand): Add an int_value parameter and use it
636 to initialize the new expand_operand field.
637 (create_integer_operand): Replace with a declaration of a function
638 that accepts poly_int64s. Move the implementation to...
639 * optabs.c (create_integer_operand): ...here.
640 (maybe_legitimize_operand): For EXPAND_INTEGER, check whether
641 the mode preserves the value of int_value, instead of calling
642 const_int_operand on the rtx. Use gen_int_mode to generate
645 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
646 Alan Hayward <alan.hayward@arm.com>
647 David Sherwood <david.sherwood@arm.com>
649 * dumpfile.h (dump_dec): Declare.
650 * dumpfile.c (dump_dec): New function.
651 * pretty-print.h (pp_wide_integer): Turn into a function and
652 declare a poly_int version.
653 * pretty-print.c (pp_wide_integer): New function for poly_ints.
655 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
656 Alan Hayward <alan.hayward@arm.com>
657 David Sherwood <david.sherwood@arm.com>
659 * doc/generic.texi (POLY_INT_CST): Document.
660 * tree.def (POLY_INT_CST): New tree code.
661 * treestruct.def (TS_POLY_INT_CST): New tree layout.
662 * tree-core.h (tree_poly_int_cst): New struct.
663 (tree_node): Add a poly_int_cst field.
664 * tree.h (POLY_INT_CST_P, POLY_INT_CST_COEFF): New macros.
665 (wide_int_to_tree, force_fit_type): Take a poly_wide_int_ref
666 instead of a wide_int_ref.
667 (build_int_cst, build_int_cst_type): Take a poly_int64 instead
669 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
670 instead of an unsigned HOST_WIDE_INT.
671 (build_poly_int_cst, tree_fits_poly_int64_p, tree_fits_poly_uint64_p)
672 (ptrdiff_tree_p): Declare.
673 (tree_to_poly_int64, tree_to_poly_uint64): Likewise. Provide
674 extern inline implementations if the target doesn't use POLY_INT_CST.
675 (poly_int_tree_p): New function.
676 (wi::unextended_tree): New class.
677 (wi::int_traits <unextended_tree>): New override.
678 (wi::extended_tree): Add a default constructor.
679 (wi::extended_tree::get_tree): New function.
680 (wi::widest_extended_tree, wi::offset_extended_tree): New typedefs.
681 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): Use them.
682 (wi::tree_to_poly_widest_ref, wi::tree_to_poly_offset_ref)
683 (wi::tree_to_poly_wide_ref): New typedefs.
684 (wi::ints_for): Provide overloads for extended_tree and
686 (poly_int_cst_value, wi::to_poly_widest, wi::to_poly_offset)
687 (wi::to_wide): New functions.
688 (wi::fits_to_boolean_p, wi::fits_to_tree_p): Handle poly_ints.
689 * tree.c (poly_int_cst_hasher): New struct.
690 (poly_int_cst_hash_table): New variable.
691 (tree_node_structure_for_code, tree_code_size, simple_cst_equal)
692 (valid_constant_size_p, add_expr, drop_tree_overflow): Handle
694 (initialize_tree_contains_struct): Handle TS_POLY_INT_CST.
695 (init_ttree): Initialize poly_int_cst_hash_table.
696 (build_int_cst, build_int_cst_type, build_invariant_address): Take
697 a poly_int64 instead of a HOST_WIDE_INT.
698 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
699 instead of an unsigned HOST_WIDE_INT.
700 (wide_int_to_tree): Rename to...
701 (wide_int_to_tree_1): ...this.
702 (build_new_poly_int_cst, build_poly_int_cst): New functions.
703 (force_fit_type): Take a poly_wide_int_ref instead of a wide_int_ref.
704 (wide_int_to_tree): New function that takes a poly_wide_int_ref.
705 (ptrdiff_tree_p, tree_to_poly_int64, tree_to_poly_uint64)
706 (tree_fits_poly_int64_p, tree_fits_poly_uint64_p): New functions.
707 * lto-streamer-out.c (DFS::DFS_write_tree_body, hash_tree): Handle
709 * tree-streamer-in.c (lto_input_ts_poly_tree_pointers): Likewise.
710 (streamer_read_tree_body): Likewise.
711 * tree-streamer-out.c (write_ts_poly_tree_pointers): Likewise.
712 (streamer_write_tree_body): Likewise.
713 * tree-streamer.c (streamer_check_handled_ts_structures): Likewise.
714 * asan.c (asan_protect_global): Require the size to be an INTEGER_CST.
715 * cfgexpand.c (expand_debug_expr): Handle POLY_INT_CST.
716 * expr.c (expand_expr_real_1, const_vector_from_tree): Likewise.
717 * gimple-expr.h (is_gimple_constant): Likewise.
718 * gimplify.c (maybe_with_size_expr): Likewise.
719 * print-tree.c (print_node): Likewise.
720 * tree-data-ref.c (data_ref_compare_tree): Likewise.
721 * tree-pretty-print.c (dump_generic_node): Likewise.
722 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
723 * tree-vect-data-refs.c (dr_group_sort_cmp): Likewise.
724 * tree-vrp.c (compare_values_warnv): Likewise.
725 * tree-ssa-loop-ivopts.c (determine_base_object, constant_multiple_of)
726 (get_loop_invariant_expr, add_candidate_1, get_computation_aff_1)
727 (force_expr_to_var_cost): Likewise.
728 * tree-ssa-loop.c (for_each_index): Likewise.
729 * fold-const.h (build_invariant_address, size_int_kind): Take a
730 poly_int64 instead of a HOST_WIDE_INT.
731 * fold-const.c (fold_negate_expr_1, const_binop, const_unop)
732 (fold_convert_const, multiple_of_p, fold_negate_const): Handle
734 (size_binop_loc): Likewise. Allow int_const_binop_1 to fail.
735 (int_const_binop_2): New function, split out from...
736 (int_const_binop_1): ...here. Handle POLY_INT_CST.
737 (size_int_kind): Take a poly_int64 instead of a HOST_WIDE_INT.
738 * expmed.c (make_tree): Handle CONST_POLY_INT_P.
739 * gimple-ssa-strength-reduction.c (slsr_process_add)
740 (slsr_process_mul): Check for INTEGER_CSTs before using them
742 * stor-layout.c (bits_from_bytes): New function.
743 (bit_from_pos): Use it.
744 (layout_type): Likewise. For vectors, multiply the TYPE_SIZE_UNIT
745 by BITS_PER_UNIT to get the TYPE_SIZE.
746 * tree-cfg.c (verify_expr, verify_types_in_gimple_reference): Allow
747 MEM_REF and TARGET_MEM_REF offsets to be a POLY_INT_CST.
749 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
750 Alan Hayward <alan.hayward@arm.com>
751 David Sherwood <david.sherwood@arm.com>
753 * doc/rtl.texi (const_poly_int): Document. Also document the
754 rtl sharing behavior.
755 * gengenrtl.c (excluded_rtx): Return true for CONST_POLY_INT.
756 * rtl.h (const_poly_int_def): New struct.
757 (rtx_def::u): Add a cpi field.
758 (CASE_CONST_UNIQUE, CASE_CONST_ANY): Add CONST_POLY_INT.
759 (CONST_POLY_INT_P, CONST_POLY_INT_COEFFS): New macros.
760 (wi::rtx_to_poly_wide_ref): New typedef
761 (const_poly_int_value, wi::to_poly_wide, rtx_to_poly_int64)
762 (poly_int_rtx_p): New functions.
763 (trunc_int_for_mode): Declare a poly_int64 version.
764 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
765 (immed_wide_int_const): Take a poly_wide_int_ref rather than
767 (strip_offset): Declare.
768 (strip_offset_and_add): New function.
769 * rtl.def (CONST_POLY_INT): New rtx code.
770 * rtl.c (rtx_size): Handle CONST_POLY_INT.
771 (shared_const_p): Use poly_int_rtx_p.
772 * emit-rtl.h (gen_int_mode): Take a poly_int64 instead of a
774 (gen_int_shift_amount): Likewise.
775 * emit-rtl.c (const_poly_int_hasher): New class.
776 (const_poly_int_htab): New variable.
777 (init_emit_once): Initialize it when NUM_POLY_INT_COEFFS > 1.
778 (const_poly_int_hasher::hash): New function.
779 (const_poly_int_hasher::equal): Likewise.
780 (gen_int_mode): Take a poly_int64 instead of a HOST_WIDE_INT.
781 (immed_wide_int_const): Rename to...
782 (immed_wide_int_const_1): ...this and make static.
783 (immed_wide_int_const): New function, taking a poly_wide_int_ref
784 instead of a wide_int_ref.
785 (gen_int_shift_amount): Take a poly_int64 instead of a HOST_WIDE_INT.
786 (gen_lowpart_common): Handle CONST_POLY_INT.
787 * cse.c (hash_rtx_cb, equiv_constant): Likewise.
788 * cselib.c (cselib_hash_rtx): Likewise.
789 * dwarf2out.c (const_ok_for_output_1): Likewise.
790 * expr.c (convert_modes): Likewise.
791 * print-rtl.c (rtx_writer::print_rtx, print_value): Likewise.
792 * rtlhash.c (add_rtx): Likewise.
793 * explow.c (trunc_int_for_mode): Add a poly_int64 version.
794 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
795 Handle existing CONST_POLY_INT rtxes.
796 * expmed.h (expand_shift): Take a poly_int64 instead of a
798 * expmed.c (expand_shift): Likewise.
799 * rtlanal.c (strip_offset): New function.
800 (commutative_operand_precedence): Give CONST_POLY_INT the same
801 precedence as CONST_DOUBLE and put CONST_WIDE_INT between that
803 * rtl-tests.c (const_poly_int_tests): New struct.
804 (rtl_tests_c_tests): Use it.
805 * simplify-rtx.c (simplify_const_unary_operation): Handle
807 (simplify_const_binary_operation): Likewise.
808 (simplify_binary_operation_1): Fold additions of symbolic constants
810 (simplify_subreg): Handle extensions and truncations of
812 (simplify_const_poly_int_tests): New struct.
813 (simplify_rtx_c_tests): Use it.
814 * wide-int.h (storage_ref): Add default constructor.
815 (wide_int_ref_storage): Likewise.
816 (trailing_wide_ints): Use GTY((user)).
817 (trailing_wide_ints::operator[]): Add a const version.
818 (trailing_wide_ints::get_precision): New function.
819 (trailing_wide_ints::extra_size): Likewise.
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 * emit-rtl.h (gen_int_shift_amount): Declare.
826 * emit-rtl.c (gen_int_shift_amount): New function.
827 * asan.c (asan_emit_stack_protection): Use gen_int_shift_amount
829 * calls.c (shift_return_value): Likewise.
830 * cse.c (fold_rtx): Likewise.
831 * dse.c (find_shift_sequence): Likewise.
832 * expmed.c (init_expmed_one_mode, store_bit_field_1, expand_shift_1)
833 (expand_shift, expand_smod_pow2): Likewise.
834 * lower-subreg.c (shift_cost): Likewise.
835 * optabs.c (expand_superword_shift, expand_doubleword_mult)
836 (expand_unop, expand_binop, shift_amt_for_vec_perm_mask)
837 (expand_vec_perm_var): Likewise.
838 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
839 (simplify_binary_operation_1): Likewise.
840 * combine.c (try_combine, find_split_point, force_int_to_mode)
841 (simplify_shift_const_1, simplify_shift_const): Likewise.
842 (change_zero_ext): Likewise. Use simplify_gen_binary.
844 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
846 * poly-int.h (multiple_p): Fix handling of two non-poly_ints.
848 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
850 * doc/invoke.texi (ARM Options): Document accepted extension options
851 for -march=armv8.3-a.
853 2017-12-20 Richard Earnshaw <rearnsha@arm.com>
856 * config.gcc (arm*-*-linux*): When configured with --with-float=hard
857 or --with-float=softfp, set the default CPU to arm10e.
859 2017-12-20 Eric Botcazou <ebotcazou@adacore.com>
861 * config/visium/constraints.md (J, K, L): Use IN_RANGE macro.
862 * config/visium/predicates.md (const_shift_operand): Likewise.
863 * config/visium/visium.c (visium_legitimize_address): Fix oversight.
864 (visium_legitimize_reload_address): Likewise.
866 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
869 * config/arm/arm.h (TEST_REGNO): Adjust comment as expected in r255830.
871 2017-12-20 Jakub Jelinek <jakub@redhat.com>
874 * calls.c (compute_argument_addresses): Ignore TYPE_EMPTY_P arguments.
876 2017-12-20 Julia Koval <julia.koval@intel.com>
878 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VPCLMULQDQ_SET,
879 OPTION_MASK_ISA_VPCLMULQDQ_UNSET): New.
880 (ix86_handle_option): Handle -mvpclmulqdq, move cx6 to flags2.
881 * config.gcc: Include vpclmulqdqintrin.h.
882 * config/i386/cpuid.h: Handle bit_VPCLMULQDQ.
883 * config/i386/driver-i386.c (host_detect_local_cpu): Handle
885 * config/i386/i386-builtin.def (__builtin_ia32_vpclmulqdq_v2di,
886 __builtin_ia32_vpclmulqdq_v4di, __builtin_ia32_vpclmulqdq_v8di): New.
887 * config/i386/i386-c.c (__VPCLMULQDQ__): New.
888 * config/i386/i386.c (isa2_opts): Add -mcx16.
889 (isa_opts): Add -mpclmulqdq, remove -mcx16.
890 (ix86_option_override_internal): Move mcx16 to flags2.
891 (ix86_valid_target_attribute_inner_p): Add vpclmulqdq.
892 (ix86_expand_builtin): Handle OPTION_MASK_ISA_VPCLMULQDQ.
893 * config/i386/i386.h (TARGET_VPCLMULQDQ, TARGET_VPCLMULQDQ_P): New.
894 * config/i386/i386.opt: Add mvpclmulqdq, move mcx16 to flags2.
895 * config/i386/immintrin.h: Include vpclmulqdqintrin.h.
896 * config/i386/sse.md (vpclmulqdq_<mode>): New pattern.
897 * config/i386/vpclmulqdqintrin.h (_mm512_clmulepi64_epi128,
898 _mm_clmulepi64_epi128, _mm256_clmulepi64_epi128): New intrinsics.
899 * doc/invoke.texi: Add -mvpclmulqdq.
901 2017-12-20 Tom de Vries <tom@codesourcery.com>
904 * config/i386/i386.c (ix86_static_chain): Move DECL_STATIC_CHAIN test ...
905 * calls.c (rtx_for_static_chain): ... here. New function.
906 * calls.h (rtx_for_static_chain): Declare.
907 * builtins.c (expand_builtin_setjmp_receiver): Use rtx_for_static_chain
908 instead of targetm.calls.static_chain.
909 * df-scan.c (df_get_entry_block_def_set): Same.
911 2017-12-19 Tom de Vries <tom@codesourcery.com>
913 PR tree-optimization/83493
914 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for): Unshare ub
917 2017-12-19 Jakub Jelinek <jakub@redhat.com>
919 * gimple-ssa-sprintf.c (format_directive): Use inform_n instead of
920 inform with hardcoded english plural handling.
922 2017-12-18 Jeff Law <law@redhat.com>
924 PR tree-optimization/83477
925 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis): For
926 a non-virtual PHI, always push a new range.
928 2017-12-19 Martin Sebor <msebor@redhat.com>
931 * builtins.c (compute_objsize): Handle non-constant offsets.
933 2017-12-19 Jakub Jelinek <jakub@redhat.com>
935 PR tree-optimization/83444
936 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): For the
937 character load case, if get_stridx on MEM_REF's operand doesn't
938 look usable, retry with get_addr_stridx.
940 2017-12-19 Alexandre Oliva <aoliva@redhat.com>
943 * var-tracking.c (vt_debug_insns_local): Do not drop markers.
944 (variable_tracking_main_1): Keep markers even when VTA fails.
947 * cfgexpand.c (expand_gimple_basic_block): Expand label first,
948 even if there are markers before it.
949 * cfgrtl.c (rtl_verify_bb_layout): Reject DEBUG_INSNs outside BBs.
951 2017-12-19 Jakub Jelinek <jakub@redhat.com>
953 * read-rtl.c (parse_reg_note_name): Replace Yoda conditions with
954 typical order conditions.
955 * sel-sched.c (extract_new_fences_from): Likewise.
956 * config/visium/constraints.md (J, K, L): Likewise.
957 * config/visium/predicates.md (const_shift_operand): Likewise.
958 * config/visium/visium.c (visium_legitimize_address,
959 visium_legitimize_reload_address): Likewise.
960 * config/m68k/m68k.c (output_reg_adjust, emit_reg_adjust): Likewise.
961 * config/arm/arm.c (arm_block_move_unaligned_straight): Likewise.
962 * config/avr/constraints.md (Y01, Ym1, Y02, Ym2): Likewise.
963 * config/avr/avr-log.c (avr_vdump, avr_log_set_avr_log,
964 SET_DUMP_DETAIL): Likewise.
965 * config/avr/predicates.md (const_8_16_24_operand): Likewise.
966 * config/avr/avr.c (STR_PREFIX_P, avr_popcount_each_byte,
967 avr_is_casesi_sequence, avr_casei_sequence_check_operands,
968 avr_set_core_architecture, avr_set_current_function,
969 avr_legitimize_reload_address, avr_asm_len, avr_print_operand,
970 output_movqi, output_movsisf, avr_out_plus, avr_out_bitop,
971 avr_out_fract, avr_adjust_insn_length, avr_encode_section_info,
972 avr_2word_insn_p, output_reload_in_const, avr_has_nibble_0xf,
973 avr_map_decompose, avr_fold_builtin): Likewise.
974 * config/avr/driver-avr.c (avr_devicespecs_file): Likewise.
975 * config/avr/gen-avr-mmcu-specs.c (str_prefix_p, print_mcu): Likewise.
976 * config/i386/i386.c (ix86_parse_stringop_strategy_string): Likewise.
977 * config/m32c/m32c-pragma.c (m32c_pragma_memregs): Likewise.
978 * config/m32c/m32c.c (m32c_conditional_register_usage,
979 m32c_address_cost): Likewise.
980 * config/m32c/predicates.md (shiftcount_operand,
981 longshiftcount_operand): Likewise.
982 * config/iq2000/iq2000.c (iq2000_expand_prologue): Likewise.
983 * config/nios2/nios2.c (nios2_handle_custom_fpu_insn_option,
984 can_use_cdx_ldstw): Likewise.
985 * config/nios2/nios2.h (CDX_REG_P): Likewise.
986 * config/cr16/cr16.h (RETURN_ADDR_RTX, REGNO_MODE_OK_FOR_BASE_P):
988 * config/cr16/cr16.md (*mov<mode>_double): Likewise.
989 * config/cr16/cr16.c (cr16_create_dwarf_for_multi_push): Likewise.
990 * config/h8300/h8300.c (h8300_rtx_costs, get_shift_alg): Likewise.
991 * config/vax/constraints.md (U06, U08, U16, CN6, S08, S16): Likewise.
992 * config/vax/vax.c (adjacent_operands_p): Likewise.
993 * config/ft32/constraints.md (L, b, KA): Likewise.
994 * config/ft32/ft32.c (ft32_load_immediate, ft32_expand_prologue):
996 * cfgexpand.c (expand_stack_alignment): Likewise.
997 * gcse.c (insert_expr_in_table): Likewise.
998 * print-rtl.c (rtx_writer::print_rtx_operand_codes_E_and_V): Likewise.
999 * cgraphunit.c (cgraph_node::expand): Likewise.
1000 * ira-build.c (setup_min_max_allocno_live_range_point): Likewise.
1001 * emit-rtl.c (add_insn): Likewise.
1002 * input.c (dump_location_info): Likewise.
1003 * passes.c (NEXT_PASS): Likewise.
1004 * read-rtl-function.c (parse_note_insn_name,
1005 function_reader::read_rtx_operand_r, function_reader::parse_mem_expr):
1007 * sched-rgn.c (sched_rgn_init): Likewise.
1008 * diagnostic-show-locus.c (layout::show_ruler): Likewise.
1009 * combine.c (find_split_point, simplify_if_then_else, force_to_mode,
1010 if_then_else_cond, simplify_shift_const_1, simplify_comparison): Likewise.
1011 * explow.c (eliminate_constant_term): Likewise.
1012 * final.c (leaf_renumber_regs_insn): Likewise.
1013 * cfgrtl.c (print_rtl_with_bb): Likewise.
1014 * genhooks.c (emit_init_macros): Likewise.
1015 * poly-int.h (maybe_ne, maybe_le, maybe_lt): Likewise.
1016 * tree-data-ref.c (conflict_fn): Likewise.
1017 * selftest.c (assert_streq): Likewise.
1018 * expr.c (store_constructor_field, expand_expr_real_1): Likewise.
1019 * fold-const.c (fold_range_test, extract_muldiv_1, fold_truth_andor,
1020 fold_binary_loc, multiple_of_p): Likewise.
1021 * reload.c (push_reload, find_equiv_reg): Likewise.
1022 * et-forest.c (et_nca, et_below): Likewise.
1023 * dbxout.c (dbxout_symbol_location): Likewise.
1024 * reorg.c (relax_delay_slots): Likewise.
1025 * dojump.c (do_compare_rtx_and_jump): Likewise.
1026 * gengtype-parse.c (type): Likewise.
1027 * simplify-rtx.c (simplify_gen_ternary, simplify_gen_relational,
1028 simplify_const_relational_operation): Likewise.
1029 * reload1.c (do_output_reload): Likewise.
1030 * dumpfile.c (get_dump_file_info_by_switch): Likewise.
1031 * gengtype.c (type_for_name): Likewise.
1032 * gimple-ssa-sprintf.c (format_directive): Likewise.
1034 2017-12-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1037 * config/arm/arm.h (TEST_REGNO): Check reg_renumber is set before
1038 accessing it. Adjust comment.
1040 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1043 * predict.c (zero_one_minusone): New function.
1044 (apply_return_prediction): Avoid return prediction for functions
1045 returning only -1, 0 and 1 values, unless they only return -1 and 0
1048 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
1050 * config/arc/arc.c (legitimate_scaled_address_p): Clean
1051 fall-through warning.
1052 (arc_compute_frame_size): Remove unused variables.
1053 (arc_print_operand): Fix fprintif format.
1054 (arc_can_follow_jump): Clean fall-through warning.
1056 2017-12-19 Marek Polacek <polacek@redhat.com>
1059 * config/i386/i386.c (init_cumulative_args): Don't check TYPE_EMPTY_P
1062 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
1064 * config/arc/arc.c (overriderregs): New variable.
1065 (arc_override_options): Track fixed/call saved/call options.
1066 (arc_conditional_register_usage): Check against overrideregs
1067 variable whenever we change register properties.
1069 2017-12-19 Nathan Sidwell <nathan@acm.org>
1071 * opts.c (finish_options): Don't prefix dump_base_name if it
1072 already contains directories.
1074 2017-12-19 Martin Liska <mliska@suse.cz>
1076 PR rtl-optimization/82675
1077 * loop-unroll.c (unroll_loop_constant_iterations): Allocate one
1078 more element in sbitmap.
1080 2017-12-19 Paolo Carlini <paolo.carlini@oracle.com>
1082 * gimplify.c (gimplify_expr): Use error_operand_p.
1084 2017-12-19 Sebastian Huber <sebastian.huber@embedded-brains.de>
1087 * config/rs6000/t-rtems (MULTILIB_REQUIRED): Remove 64-bit soft-float
1090 2017-12-19 Daniel Cederman <cederman@gaisler.com>
1092 * config/sparc/sparc.c (sparc_do_work_around_errata): Make sure
1093 the jump is to a label.
1095 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1097 PR tree-optimization/83444
1098 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): Optimize
1103 * ipa-inline.c (flatten_remove_node_hook): New function.
1104 (ipa_inline): Keep only nodes with flatten attribute at the end of
1105 the array in the order from ipa_reverse_postorder, only walk that
1106 portion of array for flattening, if there is more than one such
1107 node, temporarily register a removal hook and ignore removed nodes.
1109 PR tree-optimization/80631
1110 * tree-vect-loop.c (vect_create_epilog_for_reduction): Compare
1111 induc_code against MAX_EXPR or MIN_EXPR instead of reduc_fn against
1112 IFN_REDUC_MAX or IFN_REDUC_MIN.
1114 2017-12-18 Jeff Law <law@redhat.com>
1116 * tree-ssa-dom.c (record_equivalences_from_phis): Do not
1117 record symbolic equivalences from backedges in the CFG.
1120 2017-11-19 Jeff Law <law@redhat.com>
1122 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
1123 of degenerates resulting from ignoring an edge.
1125 2017-12-18 Martin Sebor <msebor@redhat.com>
1128 PR tree-optimization/78450
1129 * tree-ssa-strlen.c (maybe_set_strlen_range): New function.
1130 (handle_builtin_strlen): Call it.
1132 2017-12-18 Segher Boessenkool <segher@kernel.crashing.org>
1134 PR rtl-optimization/83424
1135 * rtlanal.c (dead_or_set_regno_p): Handle CLOBBER just like SET.
1137 2017-12-18 Marek Polacek <polacek@redhat.com>
1140 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
1141 Check if TYPE is INTEGRAL_TYPE_P before accessing its min/max
1144 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
1146 * config/arc/arc.md (maddsidi4, maddsidi4_split): Update pattern.
1147 (umaddsidi4, umaddsidi_split): Likewise.
1149 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
1151 * config/arc/arc.c (arc_legitimate_constant_p): Always check all
1154 2017-12-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1157 * config/s390/s390.c (s390_option_override): Avoid strncpy.
1159 2017-12-18 Richard Biener <rguenther@suse.de>
1161 PR tree-optimization/81877
1162 * tree-ssa-loop-im.c (ref_indep_loop_p): Remove safelen parameters.
1163 (outermost_indep_loop): Adjust.
1164 (ref_indep_loop_p_1): Likewise. Remove safelen handling again.
1165 (can_sm_ref_p): Adjust.
1167 2017-12-18 Richard Biener <rguenther@suse.de>
1170 * tree.c (array_at_struct_end_p): Return true if the underlying
1171 object has space for at least one element in excess of what
1172 the array domain specifies.
1174 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
1176 * doc/extend.texi (x86 Function Attributes): Reformat nocf_check
1177 example to avoid overfull hbox.
1178 * doc/invoke.texi (Option Summary): Add missing @gol.
1179 (C++ Dialect Options): Reformat -Wnoexcept-type example to avoid
1182 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
1183 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1185 * doc/invoke.texi (Option Summary): Add -mverbose-cost-dump
1186 to AArch64 and ARM lists, plus missing -mflip-thumb for ARM.
1187 (AArch64 Options): Document -mverbose-cost-dump.
1188 (ARM Options): Likewise, plus -mflip-thumb.
1190 2017-12-17 Martin Sebor <msebor@redhat.com>
1193 * gimple-ssa-warn-restrict.c
1194 (builtin_memref::offset_out_of_bounds): Correct the handling of
1197 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
1199 * doc/invoke.texi (ARC Options): Add missing -mlra entry.
1201 2017-12-17 John David Anglin <danglin@gcc.gnu.org>
1203 * config/pa/pa.c (pa_som_asm_init_sections): Fix comment.
1205 2017-12-17 Markus Trippelsdorf <markus@trippelsdorf.de>
1207 * x86-tune-costs.h (skylake_cost, core_cost): Decrease r64 multiply
1210 2017-12-16 Sandra Loosemore <sandra@codesourcery.com>
1212 * doc/invoke.texi: Fix some typos.
1214 2017-12-16 Martin Sebor <msebor@redhat.com>
1216 PR tree-optimization/78918
1217 * Makefile.in (OBJS): Add gimple-ssa-warn-restrict.o.
1218 * builtins.c (check_sizes): Rename...
1219 (check_access): ...to this. Rename function arguments for clarity.
1220 (check_memop_sizes): Adjust names.
1221 (expand_builtin_memchr, expand_builtin_memcpy): Same.
1222 (expand_builtin_memmove, expand_builtin_mempcpy): Same.
1223 (expand_builtin_strcat, expand_builtin_stpncpy): Same.
1224 (check_strncat_sizes, expand_builtin_strncat): Same.
1225 (expand_builtin_strncpy, expand_builtin_memset): Same.
1226 (expand_builtin_bzero, expand_builtin_memcmp): Same.
1227 (expand_builtin_memory_chk, maybe_emit_chk_warning): Same.
1228 (maybe_emit_sprintf_chk_warning): Same.
1229 (expand_builtin_strcpy): Adjust.
1230 (expand_builtin_stpcpy): Same.
1231 (expand_builtin_with_bounds): Detect out-of-bounds accesses
1232 in pointer-checking forms of memcpy, memmove, and mempcpy.
1233 (gcall_to_tree_minimal, max_object_size): Define new functions.
1234 * builtins.h (max_object_size): Declare.
1235 * calls.c (alloc_max_size): Call max_object_size instead of
1236 hardcoding ssizetype limit.
1237 (get_size_range): Handle new argument.
1238 * calls.h (get_size_range): Add a new argument.
1239 * cfgexpand.c (expand_call_stmt): Propagate no-warning bit.
1240 * doc/invoke.texi (-Wrestrict): Adjust, add example.
1241 * gimple-fold.c (gimple_fold_builtin_memory_op): Detect overlapping
1243 (gimple_fold_builtin_memory_chk): Same.
1244 (gimple_fold_builtin_stxcpy_chk): New function.
1245 * gimple-ssa-warn-restrict.c: New source.
1246 * gimple-ssa-warn-restrict.h: New header.
1247 * gimple.c (gimple_build_call_from_tree): Propagate location.
1248 * passes.def (pass_warn_restrict): Add new pass.
1249 * tree-pass.h (make_pass_warn_restrict): Declare.
1250 * tree-ssa-strlen.c (handle_builtin_strcpy): Detect overlapping
1252 (handle_builtin_strcat): Same.
1253 (strlen_optimize_stmt): Rename...
1254 (strlen_check_and_optimize_stmt): ...to this. Handle strncat,
1255 stpncpy, strncpy, and their checking forms.
1257 2017-12-16 Jan Hubicka <hubicka@ucw.cz>
1259 PR rtl-optimization/82849
1260 * modulo-sched.c (sms_schedule): Use get_estimated_loop_iterations_int
1261 and get_max_loop_iterations_int.
1263 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1264 Alan Hayward <alan.hayward@arm.com>
1265 David Sherwood <david.sherwood@arm.com>
1267 * machmode.h (mode_for_size, int_mode_for_size, float_mode_for_size)
1268 (smallest_mode_for_size, smallest_int_mode_for_size): Take the mode
1269 size as a poly_uint64.
1270 (mode_for_vector, mode_for_int_vector): Take the number of vector
1271 elements as a poly_uint64.
1272 * stor-layout.c (mode_for_size, smallest_mode_for_size): Take the mode
1273 size as a poly_uint64.
1274 (mode_for_vector, mode_for_int_vector): Take the number of vector
1275 elements as a poly_uint64.
1277 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1278 Alan Hayward <alan.hayward@arm.com>
1279 David Sherwood <david.sherwood@arm.com>
1281 * machmode.h (MACRO_MODE): New macro.
1282 * addresses.h (base_reg_class, ok_for_base_p_1): Use it.
1283 * targhooks.c (default_libcall_value, default_secondary_reload)
1284 (default_memory_move_cost, default_register_move_cost)
1285 (default_class_max_nregs): Likewise.
1287 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1288 Alan Hayward <alan.hayward@arm.com>
1289 David Sherwood <david.sherwood@arm.com>
1291 * doc/sourcebuild.texi: Document IN_TARGET_CODE.
1292 * genattrtab.c (write_header): Define IN_TARGET_CODE to 1 in the
1294 * genautomata.c (main): Likewise.
1295 * genconditions.c (write_header): Likewise.
1296 * genemit.c (main): Likewise.
1297 * genextract.c (print_header): Likewise.
1298 * genopinit.c (main): Likewise.
1299 * genoutput.c (output_prologue): Likewise.
1300 * genpeep.c (main): Likewise.
1301 * genpreds.c (write_insn_preds_c): Likewise.
1302 * genrecog.c (writer_header): Likewise.
1303 * config/aarch64/aarch64-builtins.c (IN_TARGET_CODE): Define.
1304 * config/aarch64/aarch64-c.c (IN_TARGET_CODE): Likewise.
1305 * config/aarch64/aarch64.c (IN_TARGET_CODE): Likewise.
1306 * config/aarch64/cortex-a57-fma-steering.c (IN_TARGET_CODE): Likewise.
1307 * config/aarch64/driver-aarch64.c (IN_TARGET_CODE): Likewise.
1308 * config/alpha/alpha.c (IN_TARGET_CODE): Likewise.
1309 * config/alpha/driver-alpha.c (IN_TARGET_CODE): Likewise.
1310 * config/arc/arc-c.c (IN_TARGET_CODE): Likewise.
1311 * config/arc/arc.c (IN_TARGET_CODE): Likewise.
1312 * config/arc/driver-arc.c (IN_TARGET_CODE): Likewise.
1313 * config/arm/aarch-common.c (IN_TARGET_CODE): Likewise.
1314 * config/arm/arm-builtins.c (IN_TARGET_CODE): Likewise.
1315 * config/arm/arm-c.c (IN_TARGET_CODE): Likewise.
1316 * config/arm/arm.c (IN_TARGET_CODE): Likewise.
1317 * config/arm/driver-arm.c (IN_TARGET_CODE): Likewise.
1318 * config/avr/avr-c.c (IN_TARGET_CODE): Likewise.
1319 * config/avr/avr-devices.c (IN_TARGET_CODE): Likewise.
1320 * config/avr/avr-log.c (IN_TARGET_CODE): Likewise.
1321 * config/avr/avr.c (IN_TARGET_CODE): Likewise.
1322 * config/avr/driver-avr.c (IN_TARGET_CODE): Likewise.
1323 * config/avr/gen-avr-mmcu-specs.c (IN_TARGET_CODE): Likewise.
1324 * config/bfin/bfin.c (IN_TARGET_CODE): Likewise.
1325 * config/c6x/c6x.c (IN_TARGET_CODE): Likewise.
1326 * config/cr16/cr16.c (IN_TARGET_CODE): Likewise.
1327 * config/cris/cris.c (IN_TARGET_CODE): Likewise.
1328 * config/darwin.c (IN_TARGET_CODE): Likewise.
1329 * config/epiphany/epiphany.c (IN_TARGET_CODE): Likewise.
1330 * config/epiphany/mode-switch-use.c (IN_TARGET_CODE): Likewise.
1331 * config/epiphany/resolve-sw-modes.c (IN_TARGET_CODE): Likewise.
1332 * config/fr30/fr30.c (IN_TARGET_CODE): Likewise.
1333 * config/frv/frv.c (IN_TARGET_CODE): Likewise.
1334 * config/ft32/ft32.c (IN_TARGET_CODE): Likewise.
1335 * config/h8300/h8300.c (IN_TARGET_CODE): Likewise.
1336 * config/i386/djgpp.c (IN_TARGET_CODE): Likewise.
1337 * config/i386/driver-i386.c (IN_TARGET_CODE): Likewise.
1338 * config/i386/driver-mingw32.c (IN_TARGET_CODE): Likewise.
1339 * config/i386/host-cygwin.c (IN_TARGET_CODE): Likewise.
1340 * config/i386/host-i386-darwin.c (IN_TARGET_CODE): Likewise.
1341 * config/i386/host-mingw32.c (IN_TARGET_CODE): Likewise.
1342 * config/i386/i386-c.c (IN_TARGET_CODE): Likewise.
1343 * config/i386/i386.c (IN_TARGET_CODE): Likewise.
1344 * config/i386/intelmic-mkoffload.c (IN_TARGET_CODE): Likewise.
1345 * config/i386/msformat-c.c (IN_TARGET_CODE): Likewise.
1346 * config/i386/winnt-cxx.c (IN_TARGET_CODE): Likewise.
1347 * config/i386/winnt-stubs.c (IN_TARGET_CODE): Likewise.
1348 * config/i386/winnt.c (IN_TARGET_CODE): Likewise.
1349 * config/i386/x86-tune-sched-atom.c (IN_TARGET_CODE): Likewise.
1350 * config/i386/x86-tune-sched-bd.c (IN_TARGET_CODE): Likewise.
1351 * config/i386/x86-tune-sched-core.c (IN_TARGET_CODE): Likewise.
1352 * config/i386/x86-tune-sched.c (IN_TARGET_CODE): Likewise.
1353 * config/ia64/ia64-c.c (IN_TARGET_CODE): Likewise.
1354 * config/ia64/ia64.c (IN_TARGET_CODE): Likewise.
1355 * config/iq2000/iq2000.c (IN_TARGET_CODE): Likewise.
1356 * config/lm32/lm32.c (IN_TARGET_CODE): Likewise.
1357 * config/m32c/m32c-pragma.c (IN_TARGET_CODE): Likewise.
1358 * config/m32c/m32c.c (IN_TARGET_CODE): Likewise.
1359 * config/m32r/m32r.c (IN_TARGET_CODE): Likewise.
1360 * config/m68k/m68k.c (IN_TARGET_CODE): Likewise.
1361 * config/mcore/mcore.c (IN_TARGET_CODE): Likewise.
1362 * config/microblaze/microblaze-c.c (IN_TARGET_CODE): Likewise.
1363 * config/microblaze/microblaze.c (IN_TARGET_CODE): Likewise.
1364 * config/mips/driver-native.c (IN_TARGET_CODE): Likewise.
1365 * config/mips/frame-header-opt.c (IN_TARGET_CODE): Likewise.
1366 * config/mips/mips.c (IN_TARGET_CODE): Likewise.
1367 * config/mmix/mmix.c (IN_TARGET_CODE): Likewise.
1368 * config/mn10300/mn10300.c (IN_TARGET_CODE): Likewise.
1369 * config/moxie/moxie.c (IN_TARGET_CODE): Likewise.
1370 * config/msp430/driver-msp430.c (IN_TARGET_CODE): Likewise.
1371 * config/msp430/msp430-c.c (IN_TARGET_CODE): Likewise.
1372 * config/msp430/msp430.c (IN_TARGET_CODE): Likewise.
1373 * config/nds32/nds32-cost.c (IN_TARGET_CODE): Likewise.
1374 * config/nds32/nds32-fp-as-gp.c (IN_TARGET_CODE): Likewise.
1375 * config/nds32/nds32-intrinsic.c (IN_TARGET_CODE): Likewise.
1376 * config/nds32/nds32-isr.c (IN_TARGET_CODE): Likewise.
1377 * config/nds32/nds32-md-auxiliary.c (IN_TARGET_CODE): Likewise.
1378 * config/nds32/nds32-memory-manipulation.c (IN_TARGET_CODE): Likewise.
1379 * config/nds32/nds32-pipelines-auxiliary.c (IN_TARGET_CODE): Likewise.
1380 * config/nds32/nds32-predicates.c (IN_TARGET_CODE): Likewise.
1381 * config/nds32/nds32.c (IN_TARGET_CODE): Likewise.
1382 * config/nios2/nios2.c (IN_TARGET_CODE): Likewise.
1383 * config/nvptx/mkoffload.c (IN_TARGET_CODE): Likewise.
1384 * config/nvptx/nvptx.c (IN_TARGET_CODE): Likewise.
1385 * config/pa/pa.c (IN_TARGET_CODE): Likewise.
1386 * config/pdp11/pdp11.c (IN_TARGET_CODE): Likewise.
1387 * config/powerpcspe/driver-powerpcspe.c (IN_TARGET_CODE): Likewise.
1388 * config/powerpcspe/host-darwin.c (IN_TARGET_CODE): Likewise.
1389 * config/powerpcspe/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
1390 * config/powerpcspe/powerpcspe-c.c (IN_TARGET_CODE): Likewise.
1391 * config/powerpcspe/powerpcspe-linux.c (IN_TARGET_CODE): Likewise.
1392 * config/powerpcspe/powerpcspe.c (IN_TARGET_CODE): Likewise.
1393 * config/riscv/riscv-builtins.c (IN_TARGET_CODE): Likewise.
1394 * config/riscv/riscv-c.c (IN_TARGET_CODE): Likewise.
1395 * config/riscv/riscv.c (IN_TARGET_CODE): Likewise.
1396 * config/rl78/rl78-c.c (IN_TARGET_CODE): Likewise.
1397 * config/rl78/rl78.c (IN_TARGET_CODE): Likewise.
1398 * config/rs6000/driver-rs6000.c (IN_TARGET_CODE): Likewise.
1399 * config/rs6000/host-darwin.c (IN_TARGET_CODE): Likewise.
1400 * config/rs6000/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
1401 * config/rs6000/rs6000-c.c (IN_TARGET_CODE): Likewise.
1402 * config/rs6000/rs6000-linux.c (IN_TARGET_CODE): Likewise.
1403 * config/rs6000/rs6000-p8swap.c (IN_TARGET_CODE): Likewise.
1404 * config/rs6000/rs6000-string.c (IN_TARGET_CODE): Likewise.
1405 * config/rs6000/rs6000.c (IN_TARGET_CODE): Likewise.
1406 * config/rx/rx.c (IN_TARGET_CODE): Likewise.
1407 * config/s390/driver-native.c (IN_TARGET_CODE): Likewise.
1408 * config/s390/s390-c.c (IN_TARGET_CODE): Likewise.
1409 * config/s390/s390.c (IN_TARGET_CODE): Likewise.
1410 * config/sh/sh-c.c (IN_TARGET_CODE): Likewise.
1411 * config/sh/sh-mem.cc (IN_TARGET_CODE): Likewise.
1412 * config/sh/sh.c (IN_TARGET_CODE): Likewise.
1413 * config/sh/sh_optimize_sett_clrt.cc (IN_TARGET_CODE): Likewise.
1414 * config/sh/sh_treg_combine.cc (IN_TARGET_CODE): Likewise.
1415 * config/sparc/driver-sparc.c (IN_TARGET_CODE): Likewise.
1416 * config/sparc/sparc-c.c (IN_TARGET_CODE): Likewise.
1417 * config/sparc/sparc.c (IN_TARGET_CODE): Likewise.
1418 * config/spu/spu-c.c (IN_TARGET_CODE): Likewise.
1419 * config/spu/spu.c (IN_TARGET_CODE): Likewise.
1420 * config/stormy16/stormy16.c (IN_TARGET_CODE): Likewise.
1421 * config/tilegx/mul-tables.c (IN_TARGET_CODE): Likewise.
1422 * config/tilegx/tilegx-c.c (IN_TARGET_CODE): Likewise.
1423 * config/tilegx/tilegx.c (IN_TARGET_CODE): Likewise.
1424 * config/tilepro/mul-tables.c (IN_TARGET_CODE): Likewise.
1425 * config/tilepro/tilepro-c.c (IN_TARGET_CODE): Likewise.
1426 * config/tilepro/tilepro.c (IN_TARGET_CODE): Likewise.
1427 * config/v850/v850-c.c (IN_TARGET_CODE): Likewise.
1428 * config/v850/v850.c (IN_TARGET_CODE): Likewise.
1429 * config/vax/vax.c (IN_TARGET_CODE): Likewise.
1430 * config/visium/visium.c (IN_TARGET_CODE): Likewise.
1431 * config/vms/vms-c.c (IN_TARGET_CODE): Likewise.
1432 * config/vms/vms-f.c (IN_TARGET_CODE): Likewise.
1433 * config/vms/vms.c (IN_TARGET_CODE): Likewise.
1434 * config/xtensa/xtensa.c (IN_TARGET_CODE): Likewise.
1436 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1437 Alan Hayward <alan.hayward@arm.com>
1438 David Sherwood <david.sherwood@arm.com>
1440 * tree-ssa-sccvn.c (vn_reference_lookup_3): Avoid repeated
1443 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1444 Alan Hayward <alan.hayward@arm.com>
1445 David Sherwood <david.sherwood@arm.com>
1447 * doc/generic.texi (VEC_SERIES_EXPR): Document.
1448 * doc/md.texi (vec_series@var{m}): Document.
1449 * tree.def (VEC_SERIES_EXPR): New tree code.
1450 * tree.h (build_vec_series): Declare.
1451 * tree.c (build_vec_series): New function.
1452 * cfgexpand.c (expand_debug_expr): Handle VEC_SERIES_EXPR.
1453 * tree-pretty-print.c (dump_generic_node): Likewise.
1454 * gimple-pretty-print.c (dump_binary_rhs): Likewise.
1455 * tree-inline.c (estimate_operator_cost): Likewise.
1456 * expr.c (expand_expr_real_2): Likewise.
1457 * optabs-tree.c (optab_for_tree_code): Likewise.
1458 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
1459 * fold-const.c (const_binop): Fold VEC_SERIES_EXPRs of constants.
1460 * expmed.c (make_tree): Handle VEC_SERIES.
1461 * optabs.def (vec_series_optab): New optab.
1462 * optabs.h (expand_vec_series_expr): Declare.
1463 * optabs.c (expand_vec_series_expr): New function.
1464 * tree-vect-generic.c (expand_vector_operations_1): Check that
1465 the operands also have vector type.
1467 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1468 Alan Hayward <alan.hawyard@arm.com>
1469 David Sherwood <david.sherwood@arm.com>
1471 * doc/generic.texi (VEC_DUPLICATE_EXPR): Document.
1472 (VEC_COND_EXPR): Add missing @tindex.
1473 * doc/md.texi (vec_duplicate@var{m}): Document.
1474 * tree.def (VEC_DUPLICATE_EXPR): New tree codes.
1475 * tree.c (build_vector_from_val): Add stubbed-out handling of
1476 variable-length vectors, using VEC_DUPLICATE_EXPR.
1477 (uniform_vector_p): Handle VEC_DUPLICATE_EXPR.
1478 * cfgexpand.c (expand_debug_expr): Likewise.
1479 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
1480 * tree-inline.c (estimate_operator_cost): Likewise.
1481 * tree-pretty-print.c (dump_generic_node): Likewise.
1482 * tree-vect-generic.c (ssa_uniform_vector_p): Likewise.
1483 * fold-const.c (const_unop): Fold VEC_DUPLICATE_EXPRs of a constant.
1484 (test_vec_duplicate_folding): New function.
1485 (fold_const_c_tests): Call it.
1486 * optabs.def (vec_duplicate_optab): New optab.
1487 * optabs-tree.c (optab_for_tree_code): Handle VEC_DUPLICATE_EXPR.
1488 * optabs.h (expand_vector_broadcast): Declare.
1489 * optabs.c (expand_vector_broadcast): Make non-static. Try using
1490 vec_duplicate_optab.
1491 * expr.c (store_constructor): Try using vec_duplicate_optab for
1493 (expand_expr_real_2): Handle VEC_DUPLICATE_EXPR.
1495 2017-12-15 Markus Trippelsdorf <markus@trippelsdorf.de>
1498 * config/i386/x86-tune-costs.h (skylake_cost, core_cost): Increase
1499 div/mod latencies a bit.
1501 2017-12-15 Jeff Law <law@redhat.com>
1503 PR tree-optimization/36550
1504 * tree-ssa-threadupdate.c (count_stmts_and_phis_in_block): New.
1505 (mark_threaded_blocks): Rewrite code to avoid block copying when
1506 optimizing for size. Don't pessimize blocks which will be
1507 copied, but all the statements will be dead.
1509 2017-12-15 Alexandre Oliva <aoliva@redhat.com>
1511 PR tree-optimization/81165
1512 * tree-ssa-threadupdate.c (uses_in_bb): New.
1513 (estimate_threading_killed_stmts): New.
1514 * tree-ssa-threadupdate.h (estimate_threading_killed_stmts): Prototype.
1515 * tree-ssa-threadedge.c
1516 (record_temporary_equivalences_from_stmts_at_dest): Expand limit
1519 2017-12-15 Jeff Law <law@redhat.com>
1521 PR tree-optimization/83410
1522 * tree-ssa-threadupdate.c (thread_block_1): Avoid certain jump
1523 threads when parallelizing loops.
1525 2017-12-15 Jakub Jelinek <jakub@redhat.com>
1527 * tree-core.h (struct attribute_spec): Swap affects_type_identity and
1529 * config/alpha/alpha.c (vms_attribute_table): Swap
1530 affects_type_identity and handler fields, adjust comments.
1531 * config/mips/mips.c (mips_attribute_table): Likewise.
1532 * config/visium/visium.c (visium_attribute_table): Likewise.
1533 * config/epiphany/epiphany.c (epiphany_attribute_table): Likewise.
1534 * config/microblaze/microblaze.c (microblaze_attribute_table):
1536 * config/spu/spu.c (spu_attribute_table): Likewise.
1537 * config/mcore/mcore.c (mcore_attribute_table): Likewise.
1538 * config/arc/arc.c (arc_attribute_table): Likewise.
1539 * config/m68k/m68k.c (m68k_attribute_table): Likewise.
1540 * config/v850/v850.c (v850_handle_interrupt_attribute,
1541 v850_handle_data_area_attribute): Formatting fixes.
1542 (v850_attribute_table): Swap affects_type_identity and handler
1543 fields, adjust comments.
1544 * config/m32r/m32r.c (m32r_attribute_table): Likewise.
1545 * config/arm/arm.c (arm_attribute_table): Likewise.
1546 * config/avr/avr.c (avr_attribute_table): Likewise.
1547 * config/s390/s390.c (s390_attribute_table): Likewise.
1548 * config/sh/sh.c (sh_attribute_table): Likewise.
1549 * config/i386/i386.c (ix86_handle_cconv_attribute,
1550 ix86_handle_callee_pop_aggregate_return): Formatting fixes.
1551 (ix86_attribute_table): Swap affects_type_identity and handler
1552 fields, adjust comments.
1553 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
1554 * config/sparc/sparc.c (sparc_attribute_table): Likewise.
1555 * config/m32c/m32c.c (m32c_attribute_table): Likewise.
1556 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Likewise.
1557 * config/ia64/ia64.c (ia64_attribute_table): Likewise.
1558 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
1559 * config/rx/rx.c (rx_attribute_table): Likewise.
1560 * config/cr16/cr16.c (cr16_attribute_table): Likewise.
1561 * config/h8300/h8300.c (h8300_attribute_table): Likewise.
1562 * config/nvptx/nvptx.c (nvptx_attribute_table): Likewise.
1563 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Likewise.
1564 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
1565 * config/stormy16/stormy16.c (xstormy16_attribute_table): Likewise.
1566 * config/bfin/bfin.c (bfin_attribute_table): Likewise.
1567 * config/rs6000/rs6000.c (rs6000_attribute_table): Likewise.
1568 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
1569 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
1570 * doc/plugins.texi (user_attr): Likewise. Add NULL for
1572 * attribs.c (empty_attribute_table): Swap affects_type_identity and
1574 (register_scoped_attributes, decl_attributes): Formatting fixes.
1576 PR tree-optimization/83269
1577 * fold-const.c (fold_binary_loc): Perform (-A) - B -> (-B) - A
1578 subtraction in arg0's type if type is signed and arg0 is unsigned.
1582 * match.pd ((T)(P + A) - (T)P -> (T) A): Use @@0 instead of @0 and
1583 convert? on @0 instead of convert. Check type of @1, not @0.
1584 ((T)P - (T)(P + A) -> -(T) A): Use @@0 instead of @0 and
1585 convert? on @0 instead of convert. Check type of @1, not @0.
1586 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Use @@0 instead of @0,
1587 only optimize if either both @1 and @2 types are narrower
1588 precision, or both are wider or equal precision, and in the former
1589 case only if both have undefined overflow.
1591 2017-12-15 Richard Biener <rguenther@suse.de>
1594 * internal-fn.def (IFN_NOP): Add.
1595 * internal-fn.c (expand_NOP): Do nothing.
1596 * lto-streamer-in.c (input_function): Instead of removing
1597 sanitizer calls replace them with IFN_NOP calls.
1599 2017-12-15 Richard Sandiford <richard.sandiford@linaro.org>
1600 Alan Hayward <alan.hayward@arm.com>
1601 David Sherwood <david.sherwood@arm.com>
1603 * dse.c (store_info, read_info_type): Replace begin and end with
1605 (print_range): New function.
1606 (set_all_positions_unneeded, any_positions_needed_p)
1607 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Update
1609 (record_store): Likewise. Optimize the case in which all positions
1611 (get_stored_val): Replace read_begin and read_end with read_offset
1613 (replace_read): Update call accordingly.
1615 2017-12-15 Bin Cheng <bin.cheng@arm.com>
1617 * gimple-loop-interchange.cc (STMT_COST_RATIO): New macro.
1618 (loop_cand::m_num_stmts, loop_cand::m_const_init_reduc): New members.
1619 (loop_cand::loop_cand): Initialize above members.
1620 (loop_cand::supported_operations): Delete.
1621 (loop_cand::can_interchange_p): Inline above function.
1622 (loop_cand::classify_simple_reduction): Record number of constant
1623 initialized simple reductions.
1624 (should_interchange_loops): New parameters. Check stmt cost of loops
1626 (tree_loop_interchange::interchange): Prepare stmt cost of outer loop.
1627 Update call to should_interchange_loops.
1628 (should_interchange_loop_nest): Update call to
1629 should_interchange_loops.
1631 2017-12-15 Eric Botcazou <ebotcazou@adacore.com>
1634 * ggc-page.c (HOST_BITS_PER_PTR): Do not define here...
1635 * hwint.h (HOST_BITS_PER_PTR): ...but here instead.
1636 * config/i386/xm-mingw32.h (HOST_BITS_PER_PTR): Delete.
1638 2017-12-15 Julia Koval <julia.koval@intel.com>
1640 * config/i386/i386-builtin.def (__builtin_ia32_vaesenclast_v16qi,
1641 __builtin_ia32_vaesenclast_v32qi, __builtin_ia32_vaesenclast_v64qi): New.
1642 * config/i386/sse.md (vaesenclast_<mode>): New pattern.
1643 * config/i386/vaesintrin.h (_mm256_aesenclast_epi128,
1644 _mm512_aesenclast_epi128, _mm_aesenclast_epi128): New intrinsics.
1646 2017-12-15 Julia Koval <julia.koval@intel.com>
1648 * config/i386/i386-builtin.def (__builtin_ia32_vaesenc_v16qi,
1649 __builtin_ia32_vaesenc_v32qi, __builtin_ia32_vaesenc_v64qi): New.
1650 * config/i386/sse.md (vaesenc_<mode>): New pattern.
1651 * config/i386/vaesintrin.h (_mm256_aesenc_epi128, _mm512_aesenc_epi128,
1652 _mm_aesenc_epi128): New intrinsics.
1654 2017-12-15 Julia Koval <julia.koval@intel.com>
1656 * config/i386/i386-builtin.def (__builtin_ia32_vaesdeclast_v16qi,
1657 __builtin_ia32_vaesdeclast_v32qi, __builtin_ia32_vaesdeclast_v64qi): New.
1658 * config/i386/sse.md (vaesdeclast_<mode>): New pattern.
1659 * config/i386/vaesintrin.h (_mm256_aesdeclast_epi128,
1660 _mm512_aesdeclast_epi128, _mm_aesdeclast_epi128): New intrinsics.
1662 2017-12-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1664 * gimple-ssa-strength-reduction.c (analyze_increments):
1665 Distinguish replacement costs for constant strides from those for
1668 2017-12-14 Jakub Jelinek <jakub@redhat.com>
1670 * var-tracking.c (variable_tracking_main_1): Formatting fix.
1672 2017-12-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
1674 * doc/invoke.texi: Document -Wcast-function-type.
1675 * recog.h (stored_funcptr): Change signature.
1676 * tree-dump.c (dump_node): Avoid warning.
1677 * typed-splay-tree.h (typed_splay_tree): Avoid warning.
1679 2017-12-14 Qing Zhao <qing.zhao@oracle.com>
1682 * gimple-fold.c (get_range_strlen): Add the handling of non-member
1685 2017-12-14 David Malcolm <dmalcolm@redhat.com>
1687 PR tree-optimization/83312
1688 * domwalk.h (dom_walker::dom_walker): Fix typo in comment.
1689 * tree-cfg.c (find_taken_edge): Update to handle NULL_TREE for
1690 "val" param, and to cope with arbitrary basic blocks.
1691 (find_taken_edge_cond_expr): Add "cond_stmt" param and use it to
1692 handle NULL_TREE for "val", dropping "bb" param.
1693 (find_taken_edge_switch_expr): Make "switch_stmt" param const and
1694 drop "bb" param. Handle NULL_TREE for "val".
1695 (find_case_label_for_value): Make "switch_stmt" param const.
1696 * tree-vrp.c (class check_array_bounds_dom_walker): New subclass
1698 (vrp_prop::check_all_array_refs): Reimplement as...
1699 (check_array_bounds_dom_walker::before_dom_children): ...this new
1700 vfunc. Replace linear search through BB block list, excluding
1701 those with non-executable in-edges via dominator walk.
1703 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1705 * config/arm/arm.opt (mverbose-cost-dump): New option.
1706 * config/arm/arm.c (arm_rtx_costs): Use it.
1708 2017-12-14 Andreas Schwab <schwab@linux-m68k.org>
1711 * reload1.c (emit_input_reload_insns): Skip debug markers.
1713 2017-12-14 Alexandre Oliva <aoliva@redhat.com>
1715 * config/i386/i386.c (rest_of_insert_endbranch): Use call loc
1719 * config/arc/arc.c (hwloop_optimize): Skip debug insns.
1720 * config/sh/sh-protos.h (sh_find_set_of_reg): Adjust.
1721 * config/sh/sh.c: Skip debug insns besides notes.
1722 * config/sh/sh.md: Likewise.
1723 * config/sh/sh_treg_combine.cc: Likewise.
1724 * config/sh/sync.md: Likewise.
1726 2017-12-14 Tom de Vries <tom@codesourcery.com>
1728 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
1729 Add item for weak_undefined.
1731 2017-12-14 Richard Biener <rguenther@suse.de>
1733 PR tree-optimization/67842
1734 * sese.h (bb_in_region): Remove #if 0'ed code.
1736 2017-12-14 Richard Biener <rguenther@suse.de>
1738 PR tree-optimization/83326
1739 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
1740 may_be_zero parameter and handle it by not marking the first
1741 peeled copy as not exiting the loop.
1742 (try_peel_loop): Likewise.
1743 (canonicalize_loop_induction_variables): Use number_of_iterations_exit
1744 to handle the case of constant or zero iterations and perform
1745 loop header copying on-the-fly.
1747 2017-12-14 Richard Biener <rguenther@suse.de>
1749 PR tree-optimization/83418
1750 * vr-values.c (vr_values::extract_range_for_var_from_comparison_expr):
1751 Instead of asserting we don't get unfolded comparisons deal with
1754 2017-12-14 Jakub Jelinek <jakub@redhat.com>
1757 * var-tracking.c (vt_initialize): Ignore non-DEBUG_INSNs outside of
1758 basic blocks. Assert debug bind insns don't appear outside of bbs,
1759 don't reset them. Assert insns without BLOCK_FOR_INSN are outside of
1762 PR tree-optimization/83198
1763 * gimple-ssa-sprintf.c (format_floating): Set type solely based on
1764 dir.modifier, regardless of TREE_TYPE (arg). Assume non-REAL_CST
1765 value if arg is a REAL_CST with incompatible type.
1767 2017-12-14 Sudakshina Das <sudi.das@arm.com>
1768 Bin Cheng <bin.cheng@arm.com>
1771 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Move LTGT
1773 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Add
1776 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1778 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75,
1779 cortex-a75.cortex-a55): Specify AARCH64_FL_F16 in the arch features.
1781 2017-12-14 Richard Sandiford <richard.sandiford@linaro.org>
1782 Alan Hayward <alan.hayward@arm.com>
1783 David Sherwood <david.sherwood@arm.com>
1785 * poly-int.h: New file.
1786 * poly-int-types.h: Likewise.
1787 * coretypes.h: Include them.
1788 (POLY_INT_CONVERSION): Define.
1789 * target.def (estimated_poly_value): New hook.
1790 * doc/tm.texi.in (TARGET_ESTIMATED_POLY_VALUE): New hook.
1791 * doc/tm.texi: Regenerate.
1792 * doc/poly-int.texi: New file.
1793 * doc/gccint.texi: Include it.
1794 * doc/rtl.texi: Describe restrictions on subreg modes.
1795 * Makefile.in (TEXI_GCCINT_FILES): Add poly-int.texi.
1796 * genmodes.c (NUM_POLY_INT_COEFFS): Provide a default definition.
1797 (emit_insn_modes_h): Emit a definition of NUM_POLY_INT_COEFFS.
1798 * targhooks.h (default_estimated_poly_value): Declare.
1799 * targhooks.c (default_estimated_poly_value): New function.
1800 * target.h (estimated_poly_value): Likewise.
1801 * wide-int.h (WI_UNARY_RESULT): Use wi::binary_traits.
1802 (wi::unary_traits): Delete.
1803 (wi::binary_traits::signed_shift_result_type): Define for
1804 offset_int << HOST_WIDE_INT, etc.
1805 (generic_wide_int::operator <<=): Define for all types and use
1806 wi::lshift instead of <<.
1807 (wi::hwi_with_prec): Add a default constructor.
1808 (wi::ints_for): New class.
1809 (operator <<): Define for all wide-int types.
1810 (operator /): New function.
1811 (operator %): Likewise.
1812 * selftest.h (ASSERT_KNOWN_EQ, ASSERT_KNOWN_EQ_AT, ASSERT_MAYBE_NE)
1813 (ASSERT_MAYBE_NE_AT): New macros.
1815 2017-12-13 Eric Botcazou <ebotcazou@adacore.com>
1816 Dominik Vogt <vogt@linux.vnet.ibm.com>
1819 * emit-rtl.c (init_emit): Remove ??? comment.
1820 * explow.c (get_dynamic_stack_size): Take known alignment of stack
1821 pointer + STACK_DYNAMIC_OFFSET into account in lieu of STACK_BOUNDARY.
1822 * config/sparc/sparc.h (INIT_EXPANDERS): In 32-bit mode, lower the
1823 alignment of 3 virtual registers to BITS_PER_WORD.
1825 * config/sparc/sparc.c (sparc_compute_frame_size): Simplify.
1827 2017-12-13 Peter Bergner <bergner@vnet.ibm.com>
1829 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_HTM_NO_SUSPEND): New define.
1830 * config/rs6000/rs6000.c (cpu_supports_info): Use it.
1832 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
1835 * reload1.c (eliminate_regs_in_insn): Skip debug markers.
1837 2017-12-13 Jakub Jelinek <jakub@redhat.com>
1839 * tree-cfg.c (verify_gimple_in_cfg): Verify no non-label stmts
1840 with the exception of debug begin stmt markers appear before
1844 * final.c (rest_of_handle_final): Call variable_tracking_main only
1845 if !flag_var_tracking.
1847 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
1848 Jakub Jelinek <jakub@redhat.com>
1852 * tree-cfgcleanup.c (remove_forwarder_block): Keep after
1853 labels debug stmts that can only appear after labels.
1855 2017-12-13 Alexander Monakov <amonakov@ispras.ru>
1857 PR rtl-optimization/82398
1858 * sel-sched.c (sel_rank_for_schedule): Fix check for zero
1859 EXPR_USEFULNESS in priority comparison.
1861 2017-12-13 Segher Boessenkool <segher@kernel.crashing.org>
1863 PR rtl-optimization/83393
1864 * combine.c (move_deaths): If reg_stat points to a too new insn in
1865 last_death, do not use it: find the proper insn instead.
1867 2017-12-12 Jeff Law <law@redhat.com>
1869 PR tree-optimization/83298
1870 PR tree-optimization/83362
1871 PR tree-optimization/83383
1872 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Make
1873 push_value_range a public interface. Add new argument to
1874 record_ranges_from_stmt.
1875 * gimple-ssa-evrp-analyze.c
1876 (evrp_range_analyzer::record_ranges_from_stmt): Add new argument.
1877 Update comments. Handle recording temporary equivalences.
1878 * tree-ssa-dom.c (dom_opt_opt_walker::before_dom_children): Add
1879 new argument to call to evrp_range_analyzer::record_ranges_from_stmt.
1880 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Likewise.
1881 * tree-ssa-threadedge.c: Include alloc-pool.h, vr-values.h and
1882 gimple-ssa-evrp-analyze.h.
1883 (record_temporary_equivalences_from_phis): Add new argument. When
1884 the PHI arg is an SSA_NAME, set the result's range to the range
1886 (record_temporary_equivalences_from_stmts_at_dest): Record ranges
1887 from statements too.
1888 (thread_through_normal_block): Accept new argument, evrp_range_analyzer.
1889 Pass it down to children as needed.
1890 (thread_outgoing_edges): Likewise.
1891 (thread_across_edge): Likewise. Push/pop range state as needed.
1892 * tree-ssa-threadedge.h (thread_outgoing_edges): Update prototype.
1894 2017-12-12 Julia Koval <julia.koval@intel.com>
1896 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
1897 (PTA_CANNONLAKE): Remove PTA_CLWB.
1899 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1902 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
1903 (vcondv2div2df): Likewise.
1904 (vconduv2dfv2di): Likewise.
1906 2017-12-12 Jakub Jelinek <jakub@redhat.com>
1908 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
1909 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
1910 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
1911 ATTR_MATHFN_FPROUNDING_ERRNO.
1913 2017-12-12 Richard Biener <rguenther@suse.de>
1915 PR tree-optimization/83385
1916 * graphite-scop-detection.c (get_order, order): Remove.
1917 (bb_to_rpo): New global.
1919 (build_scops): Sort pbbs in RPO order.
1921 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
1923 * combine.c (simplify_set): Do not transform subregs to zero_extends
1924 if the destination mode is a vector mode.
1926 2017-12-12 Jakub Jelinek <jakub@redhat.com>
1928 PR tree-optimization/83359
1929 * tree-cfg.h (fold_loop_internal_call): Declare.
1930 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
1931 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
1932 (find_loop_dist_alias): New function.
1933 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
1934 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
1937 PR tree-optimization/80631
1938 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
1939 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
1940 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
1941 hardcoding zero as the value if COND_EXPR is never true. For
1942 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
1943 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
1944 hardcoding MAX_EXPR as the reduction operation.
1945 (is_nonwrapping_integer_induction): Allow negative step.
1946 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
1947 vect_create_epilog_for_reduction, if no value is suitable, don't
1948 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
1950 2017-12-12 Richard Biener <rguenther@suse.de>
1952 PR tree-optimization/81889
1953 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
1954 range info from the non-wrapping IV instead of just the range
1957 2017-12-12 Julia Koval <julia.koval@intel.com>
1959 * config.gcc: Add vaesintrin.h.
1960 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
1962 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
1963 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
1965 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
1966 * config/i386/immintrin.h: Include vaesintrin.h.
1967 * config/i386/sse.md (vaesdec_<mode>): New pattern.
1968 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
1969 _mm_aesdec_epi128): New intrinsics.
1971 2017-12-12 Julia Koval <julia.koval@intel.com>
1973 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
1974 OPTION_MASK_ISA_VAES_UNSET): New.
1975 (ix86_handle_option): Handle -mvaes.
1976 * config/i386/cpuid.h: Define bit_VAES.
1977 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
1978 * config/i386/i386-c.c (__VAES__): New.
1979 * config/i386/i386.c (ix86_target_string): Add -mvaes.
1980 (ix86_valid_target_attribute_inner_p): Ditto.
1981 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
1982 * config/i386/i386.opt: Add -mvaes.
1983 * doc/invoke.texi: Ditto.
1985 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
1987 * debug.h (gcc_debug_hooks): Add inline_entry.
1988 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
1989 * debug.c (do_nothing_debug_hooks): Likewise.
1990 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
1991 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
1992 (dwarf2_lineno_debug_hooks): Likewise.
1994 * common.opt (gstatement-frontiers): New, setting
1995 debug_nonbind_markers_p.
1996 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
1997 * toplev.c (process_options): Autodetect value for debug statement
1999 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
2000 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
2002 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
2003 markers. Integrate source bind into debug stmt expand loop.
2004 (pass_expand::execute): Check debug marker limit. Avoid deep
2005 TER and expand debug locations for debug bind insns only.
2006 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
2008 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
2009 * final.c (reemit_insn_block_notes): Take current block from
2010 nonbind markers. Declare note where it's first set.
2011 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
2012 begin stmt markers if enabled.
2013 (notice_source_line): Handle nonbind markers. Fail if their
2014 location is unknown or that of builtins.
2015 (rest_of_handle_final): Convert begin stmt markers to notes if
2016 var-tracking didn't run.
2017 (rest_of_clean_state): Skip begin stmt markers.
2018 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
2020 * function.c (allocate_struct_function): Set begin_stmt_markers.
2021 * function.h (struct function): Add debug_marker_count counter
2022 and debug_nonbind_markers flag.
2023 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
2024 * gimple-low.c (lower_function_body): Adjust
2025 debug_nonbind_markers.
2026 (lower_stmt): Drop or skip gimple debug stmts.
2027 (lower_try_catch): Skip debug stmts.
2028 * gimple.c (gimple_build_debug_begin_stmt): New.
2029 (gimple_copy): Increment debug_marker_count if copying one.
2030 * gimple.h (gimple_build_debug_begin_stmt): Declare.
2031 * gimplify.c (rexpr_location): New.
2032 (rexpr_has_location): New.
2033 (warn_switch_unreachable_r): Handle gimple debug stmts.
2034 (shortcut_cond_r): Call expr_location.
2036 (find_goto_label): New.
2037 (shortcut_cond_expr): Call expr_has_location, expr_location, and
2039 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
2041 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
2042 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
2043 (LANG_HOOKS_INITIALIZER): ... this.
2044 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
2045 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
2047 (update_ebb_live_info): Skip debug insn markers.
2048 * lra.c (debug_insn_static_data): Rename to...
2049 (debug_bind_static_data): ... this.
2050 (debug_marker_static_data): New.
2051 (lra_set_insn_recog_data): Select one of the above depending
2053 (lra_update_isn_regno_info): Don't assume debug insns have
2055 (push_insns): Skip debug insns.
2056 * lto-streamer-in.c (input_function): Drop debug stmts
2057 depending on active options. Adjust debug_nonbind_markers.
2058 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
2059 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
2060 begin stmt marker notes.
2061 (print_insn): Likewise.
2062 * recog.c (extract_insn): Recognize rtl for debug markers.
2063 * rtl.def (DEBUG_MARKER): New.
2064 * tree-inline.c: Include params.h.
2065 (remap_gimple_stmt): Handle nonbind markers.
2066 (maybe_move_debug_stmts_to_successors): Likewise.
2067 (copy_debug_stmt): Likewise.
2068 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
2069 markers regardless of no side effects.
2070 (tsi_link_before): Don't update container's side effects when adding
2071 a begin stmt marker.
2072 (tsi_link_after): Likewise.
2073 (expr_first): Skip begin stmt markers.
2074 (expr_last): Likewise.
2075 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
2076 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
2077 Disregard nonbind markers.
2078 * tree.c (make_node_stat): Don't set side effects for begin stmt
2080 (build1_stat): Likewise.
2081 * tree.def (DEBUG_BEGIN_STMT): New.
2082 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
2083 * var-tracking.c (delete_debug_insns): Renamed to...
2084 (delete_vta_debug_insns): ... this.
2085 (reemit_marker_as_note): New.
2086 (vt_initialize): Reemit markers.
2087 (delete_vta_debug_insns): Likewise.
2088 (vt_debug_insns_local): Reemit or delete markers.
2089 (variable_tracking_main_1): Likewise.
2090 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
2091 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
2092 (gimple_debug_nonbind_marker_p): New.
2093 (gimple_build_debug_bind): Adjust.
2094 (gimple_build_debug_begin_stmt): New.
2095 * doc/invoke.texi (max-debug-marker-count): New param.
2096 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
2097 (debug_parameter_ref, debug_marker): New.
2098 (NOTE_INSN_BEGIN_STMT): New.
2099 (DEBUG_INSN): Describe begin stmt markers.
2101 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
2102 without debug insns, we wouldn't, but clean up debug insns
2103 after a control flow insn nevertheless.
2105 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
2106 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
2107 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
2108 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
2109 (rtl_tidy_fallthru_edge): Likewise.
2110 (rtl_verify_fallthru): Likewise.
2111 (rtl_verify_bb_layout): Likewise.
2112 (skip_insns_after_block): Likewise.
2113 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
2114 * dwarf2out.c: Include print-rtl.h.
2115 (dwarf2out_next_real_insn): New.
2116 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
2117 Dump debug binds in asm comments.
2118 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
2119 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
2120 callers to use gsi_start_nondebug_bb instead.
2121 (gsi_after_labels): Skip gimple debug stmts.
2122 (gsi_start_nondebug): New.
2123 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
2124 (proper_loop_form_for_interchange): Adjust.
2125 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
2126 * gimple.h (gimple_seq_last_nondebug_stmt): New.
2127 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
2128 (collect_fallthrough_labels): Likewise.
2129 (should_warn_for_implicit_fallthrough): Likewise.
2130 (warn_implicit_fallthrough_r): Likewise.
2131 (expand_FALLTHROUGH_r): Likewise.
2132 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
2133 (graphite_copy_stmts_from_block): Skip nonbind markers.
2134 * haifa-sched.c (sched_extend_bb): Skip debug insns.
2135 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
2136 * jump.c (clean_barriers): Skip debug insns.
2137 * omp-expand.c (expand_parallel_call): Skip debug insns.
2138 (expand_task_call): Likewise.
2139 (remove_exit_barrier): Likewise.
2140 (expand_omp_taskreg): Likewise.
2141 (expand_omp_for_init_counts): Likewise.
2142 (expand_omp_for_generic): Likewise.
2143 (expand_omp_for_static_nochunk): Likewise.
2144 (expand_omp_for_static_chunk): Likewise.
2145 (expand_omp_simd): Likewise.
2146 (expand_omp_taskloop_for_outer): Likewise.
2147 (expand_omp_taskloop_for_inner): Likewise.
2148 (expand_oacc_for): Likewise.
2149 (expand_omp_sections): Likewise.
2150 (expand_omp_single): Likewise.
2151 (expand_omp_synch): Likewise.
2152 (expand_omp_atomic_load): Likewise.
2153 (expand_omp_atomic_store): Likewise.
2154 (expand_omp_atomic_fetch_op): Likewise.
2155 (expand_omp_atomic_pipeline): Likewise.
2156 (expand_omp_atomic_mutex): Likewise.
2157 (expand_omp_target): Likewise.
2158 (grid_expand_omp_for_loop): Likewise.
2159 (grid_expand_target_grid_body): Likewise.
2160 (build_omp_regions_1): Likewise.
2161 * omp-low.c (check_combined_parallel): Skip debug stmts.
2162 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
2163 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
2164 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
2166 * tree-cfg.c (make_blobs_1): Skip debug stmts.
2167 (make_edges): Likewise.
2168 (cleanup_dead_labels): Likewise.
2169 (gimple_can_merge_blocks_p): Likewise.
2170 (stmt_starts_bb_p): Likewise.
2171 (gimple_block_label): Likewise.
2172 (gimple_redirect_edge_and_branch): Likewise.
2173 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
2175 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
2177 * tree-pretty-print (print_declaration): Omit initializer in slim
2179 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
2181 (eliminate_unnecessary_stmts): Stabilize block removal order.
2182 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
2183 * var-tracking.c (get_first_insn): New.
2184 (vt_emit_notes): Call it.
2185 (vt_initialize): Walk any insns before the first BB.
2186 (delete_debug_insns): Likewise.
2188 * gimple.h (enum gimple_debug_subcode): Add
2189 GIMPLE_DEBUG_BEGIN_STMT.
2190 (gimple_debug_begin_stmt_p): New.
2191 (gimple_debug_nonbind_marker_p): New.
2192 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
2193 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
2194 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
2195 * insn-notes.def (BEGIN_STMT): New.
2196 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
2197 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
2198 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
2199 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
2200 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
2201 (INSN_DEBUG_MARKER_KIND): New.
2202 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
2203 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
2204 (INSN_VAR_LOCATION_PTR): New.
2205 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
2207 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
2208 deps only if debug bind insns are enabled.
2209 (pass_expand::execute): Avoid deep TER and expand
2210 debug locations for debug bind insns only.
2211 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
2212 debug stmts special handling down to debug bind stmts.
2213 * combine.c (try_combine): Narrow debug insns special handling
2214 down to debug bind insns.
2215 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
2216 Narrow debug insns preexisting special handling down to debug
2218 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
2219 handling down to debug bind insns.
2220 * function.c (instantiate_virtual_regs): Skip debug markers,
2221 adjust handling of debug binds.
2222 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
2223 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
2224 * haifa-sched.c (schedule_insn): Narrow special handling of debug
2225 insns to debug bind insns.
2226 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
2227 special handling of debug stmts to debug bind stmts.
2228 * ipa-split.c (split_function): Likewise.
2229 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
2230 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
2232 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
2233 * regrename.c (build_def_use): Likewise.
2234 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
2235 (pass_cprop_hardreg): Narrow special casing of debug insns to
2237 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
2238 * reload1.c (reload): Likewise.
2239 * sese.c (sese_insert_phis_for_liveouts): Narrow special
2240 casing of debug stmts to debug bind stmts.
2241 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
2242 * ssa-iterators.h (num_imm_uses): Likewise.
2243 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
2244 debug stmts to debug bind stmts.
2245 * tree-inline.c (tree_function_versioning): Narrow special casing
2246 of debug stmts to debug bind stmts.
2247 * tree-loop-distribution.c (generate_loops_for_partition):
2248 Narrow special casing of debug stmts to debug bind stmts.
2249 * tree-sra.c (analyze_access_subtree): Narrow special casing
2250 of debug stmts to debug bind stmts.
2251 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
2252 stmts to debug bind stmts.
2253 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
2254 casing of debug stmts to debug bind stmts.
2255 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
2256 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
2257 casing of debug stmts to debug bind stmts.
2258 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
2260 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
2261 debug stmts to debug bind stmts.
2262 (gimple_replace_ssa_lhs): Likewise.
2263 (insert_debug_temp_for_var_def): Likewise.
2264 (insert_debug_temps_for_defs): Likewise.
2265 (reset_debug_uses): Likewise.
2266 * tree-ssanames.c (release_ssa_name_fn): Likewise.
2267 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
2268 (adjust_debug_stmts): Likewise.
2269 (adjust_phi_and_debug_stmts): Likewise.
2270 (vect_do_peeling): Likewise.
2271 * tree-vect-loop.c (vect_transform_loop): Likewise.
2272 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
2273 * var-tracking.c (adjust_mems): Narrow special casing of debug
2274 insns to debug bind insns.
2275 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
2276 (compute_bb_dataflow, vt_find_locations): Likewise.
2277 (vt_expand_loc, emit_notes_for_changes): Likewise.
2278 (vt_init_cfa_base): Likewise.
2279 (vt_emit_notes): Likewise.
2280 (vt_initialize): Likewise.
2281 (vt_finalize): Likewise.
2283 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
2284 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
2285 (next_nonnote_nondebug_insn_bb): New.
2286 (prev_nonnote_nondebug_insn_bb): New.
2287 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
2288 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
2289 (prev_nonnote_nondebug_insn_bb): Declare.
2290 (next_nonnote_nondebug_insn_bb): Declare.
2291 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
2292 * cfgrtl.c (get_last_bb_insn): Likewise.
2293 * lra.c (push_insns): Likewise.
2295 2017-12-11 David Malcolm <dmalcolm@redhat.com>
2298 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
2300 * selftest.c (selftest::test_runner::test_runner): New ctor.
2301 (selftest::test_runner::~test_runner): New dtor.
2302 * selftest.h (class selftest::test_runner): New class.
2304 2017-12-11 Carl Love <cel@us.ibm.com>
2306 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
2307 vec_extract_fp32_from_shortl]): Add #defines.
2308 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
2309 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
2310 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
2311 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
2312 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
2313 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
2314 * doc/extend.texi: Add documentation for the added builtins.
2316 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
2318 PR rtl-optimization/80693
2319 PR rtl-optimization/81019
2320 PR rtl-optimization/81020
2321 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
2322 are not mentioned in i3. Place the REG_UNUSED note on i2,
2323 possibly modified to REG_DEAD, if it did not originate in i3.
2325 2017-12-11 Jakub Jelinek <jakub@redhat.com>
2327 * recog.c (store_data_bypass_p_1): New function.
2328 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
2329 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
2331 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
2333 PR rtl-optimization/83361
2334 * ifcvt.c (if_convert): Call fixup_partitions.
2336 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
2338 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
2339 early folding of splat_u{8,16,32}.
2341 2017-12-11 Jakub Jelinek <jakub@redhat.com>
2343 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
2344 output_operand_lossage first argument with capital letter.
2345 (aarch64_override_options): Don't start error and sorry first argument
2346 with capital letter.
2348 2017-12-11 Andi Kleen <ak@linux.intel.com>
2350 PR gcov-profile/83355
2351 * auto-profile.c (string_table::get_index_by_decl): Don't
2352 recurse when abstract origin points to itself.
2354 2017-12-11 Bin Cheng <bin.cheng@arm.com>
2356 PR tree-optimization/83320
2357 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
2358 (prune_datarefs_not_in_loop): Ditto.
2360 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
2362 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
2364 2017-12-10 Jakub Jelinek <jakub@redhat.com>
2366 PR tree-optimization/83337
2367 * gimple-loop-interchange.cc (compute_access_stride): Handle
2368 bitfield DRs properly.
2370 2017-12-09 Jakub Jelinek <jakub@redhat.com>
2372 PR tree-optimization/83338
2373 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
2374 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
2375 vector of unsigned integers to vector of signed integers.
2377 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
2379 PR rtl-optimization/83317
2380 * lra-constraints.c (process_address_1): Add insn code check.
2382 2017-12-08 Michael Matz <matz@suse.de>
2384 Fix PR tree-optimization/83323
2385 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
2386 head-controlled loops and loop BBs.
2387 * common.opt (funroll-and-jam): Remove, instead ...
2388 (floop-unroll-and-jam): ... reuse this option.
2389 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
2390 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
2391 (-floop-unroll-and-jam): ... this option.
2393 2017-12-08 Jakub Jelinek <jakub@redhat.com>
2395 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
2396 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
2397 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
2399 2017-12-08 Richard Biener <rguenther@suse.de>
2402 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
2403 handle accesses outside of zero-sized vars.
2405 2017-12-08 Martin Jambor <mjambor@suse.cz>
2407 PR tree-optimization/83141
2408 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
2409 test for MEM_REFs implicitely changing types with padding. Remove
2411 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
2413 2017-12-08 Jakub Jelinek <jakub@redhat.com>
2415 * config/arc/arc.c (arc_attribute_table): Add exclusions to
2417 * config/avr/avr.c (avr_attribute_table): Likewise.
2418 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
2419 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
2420 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
2421 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
2422 of struct attribute_spec.
2423 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
2425 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
2428 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
2430 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
2433 * profile-count.c (profile_count::from_gcov_type): Move from
2434 profile-count.h; handle overflow.
2435 * profile-count.h (profile_count::from_gcov_type): Move offline.
2437 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
2439 PR rtl-optimization/83304
2440 * combine.c (move_deaths): If we do not know where a register died,
2443 2017-12-08 Richard Biener <rguenther@suse.de>
2445 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
2446 Provide -fopt-info-loop feedback when we interchange in a nest.
2448 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
2450 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
2451 for armv6 ARM CPU IDs.
2453 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
2455 * common/config/arm/arm-common.c: Include <algorithm>.
2456 (INCLUDE_VECTOR): Define.
2457 (compare_opt_names): New function.
2458 (arm_rewrite_selected_arch): Only strip out extensions that can be
2459 expressed through -mfpu. Sort the remaining extensions
2462 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
2464 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
2465 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
2466 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
2467 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
2469 2017-06-08 Tristan Gingold <gindold@adacore.com>
2472 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
2473 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
2475 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2477 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
2478 of struct attribute_spec.
2480 2017-12-08 Julia Koval <julia.koval@intel.com>
2482 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
2483 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
2484 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
2485 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
2486 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
2487 _mm_maskz_dpwssds_epi32): Ditto.
2489 2017-12-08 Richard Biener <rguenther@suse.de>
2491 PR tree-optimization/81303
2492 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
2493 conditions try to create a comparison vector type matching
2494 the data vector type.
2495 (vectorizable_condition): Adjust.
2496 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
2497 Leave invariant conditions alone in case we can vectorize those.
2499 2017-12-08 Julia Koval <julia.koval@intel.com>
2501 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
2502 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
2503 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
2504 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
2505 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
2507 2017-12-08 Julia Koval <julia.koval@intel.com>
2509 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
2510 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
2511 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
2512 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
2513 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
2514 _mm_maskz_dpbusds_epi32): New intrinsics.
2516 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
2518 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
2520 (ld<bh>io_signed): Likewise.
2521 (st<bhw>io): Likewise.
2522 * config/nios2/predicates.md (ldstio_memory_operand): Allow
2523 SMALL_INT12 constant integer operand.
2525 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
2527 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
2529 (nios2_large_constant_allowed): ...this. Adjust uses.
2530 (nios2_plus_symbolic_constant_p): Rename to...
2531 (nios2_plus_large_constant_p): ...this. Adjust uses.
2532 (nios2_legitimate_address_p): Correct CONST_INT handling.
2533 (nios2_symbolic_memory_operand_p): Rename to...
2534 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
2535 (nios2_large_constant_p): Check for large constant integers too.
2536 (nios2_split_large_constant): Handle constant integers.
2537 (nios2_split_symbolic_memory_operand): Rename to...
2538 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
2539 (nios2_legitimize_constant_address): Handle constant integers.
2540 (r0rel_constant_p): Handle small constant integers.
2541 (nios2_print_operand_address): Handle r0-relative integer addresses.
2542 * config/nios2/nios2-protos.h: Adjust for renamed functions.
2543 * config/nios2/nios2.md: Adjust for renamed functions.
2545 2017-12-07 Andrew Waterman <andrew@sifive.com>
2547 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
2548 (TARGET_HAVE_SRODATA_SECTION): New define.
2549 (riscv_select_section): New function.
2551 2017-12-08 Joseph Myers <joseph@codesourcery.com>
2552 Alexander Monakov <amonakov@ispras.ru>
2553 Jakub Jelinek <jakub@redhat.com>
2556 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
2558 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2560 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
2561 a bool success value. Don't call output_operand_lossage here.
2562 (aarch64_print_ldpstp_address): Return a bool success value.
2563 (aarch64_print_operand_address): Call output_addr_const if
2564 aarch64_print_address_internal fails.
2565 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
2566 'y'; call output_operand_lossage instead. Call output_operand_lossage
2567 if aarch64_print_ldpstp_address fails.
2569 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2571 * tree-vector-builder.h
2572 (tree_vector_builder::binary_encoded_nelts): Declare.
2573 * tree-vector-builder.c
2574 (tree_vector_builder::binary_encoded_nelts): New function.
2575 * fold-const.c (negate_expr_p): Likewise.
2576 (operand_equal_p, fold_checksum_tree): Likewise.
2577 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
2578 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
2579 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
2580 (uniform_vector_p): Likewise.
2581 * varasm.c (const_hash_1, compare_constant): Likewise.
2582 * tree-ssa-ccp.c: Include tree-vector-builder.h.
2583 (valid_lattice_transition): Operate directly on the VECTOR_CST
2585 * ipa-icf.c: Include tree-vector-builder.h.
2586 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
2587 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
2589 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2591 * tree.c (build_vector): Delete.
2592 * tree.h (build_vector): Make static and move into the self-testing
2595 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2597 * vector-builder.h (vector_builder::derived): New const overload.
2598 (vector_builder::elt): New function.
2599 * tree-vector-builder.h (tree_vector_builder::type): New function.
2600 (tree_vector_builder::apply_step): Declare.
2601 * tree-vector-builder.c (tree_vector_builder::apply_step): New
2603 * gimple-fold.h (tree_vector_builder): Declare.
2604 (gimple_build_vector): Take a tree_vector_builder instead of a
2605 type and vector of elements.
2606 * gimple-fold.c (gimple_build_vector): Likewise.
2607 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
2609 (get_initial_defs_for_reduction): Likewise.
2610 (vectorizable_induction): Likewise.
2612 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2614 * tree-vector-builder.h
2615 (tree_vector_builder::new_binary_operation): Declare.
2616 * tree-vector-builder.c
2617 (tree_vector_builder::new_binary_operation): New function.
2618 * fold-const.c (fold_relational_const): Use it.
2619 (const_binop): Likewise. Check that both input vectors have
2620 the same number of elements, thus excluding things like WIDEN_SUM.
2621 Check whether it is possible to operate directly on the encodings
2624 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2626 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
2627 new_unary_operation, operating only on the encoded elements.
2628 (const_unop): Likewise.
2629 (exact_inverse): Likewise.
2630 (distributes_over_addition_p): New function.
2631 (const_binop): Use tree_vector_builder and new_unary_operation
2632 for combinations of VECTOR_CST and INTEGER_CST. Operate only
2633 on the encoded elements unless the encoding is strided and the
2634 operation does not distribute over addition.
2635 (fold_convert_const): Use tree_vector_builder and
2636 new_unary_operation. Operate only on the encoded elements
2637 for truncating integer conversions, or for non-stepped encodings.
2639 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2641 * config/sparc/sparc.c: Include tree-vector-builder.h.
2642 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
2643 * expmed.c: Include tree-vector-builder.h.
2644 (make_tree): Use tree_vector_builder instead of build_vector.
2645 * fold-const.c: Include tree-vector-builder.h.
2646 (const_binop): Use tree_vector_builder instead of build_vector.
2647 (const_unop): Likewise.
2648 (native_interpret_vector): Likewise.
2649 (fold_vec_perm): Likewise.
2650 (fold_ternary_loc): Likewise.
2651 * gimple-fold.c: Include tree-vector-builder.h.
2652 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
2654 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
2655 (simplify_vector_constructor): Use tree_vector_builder instead
2657 * tree-vect-generic.c: Include tree-vector-builder.h.
2658 (add_rshift): Use tree_vector_builder instead of build_vector.
2659 (expand_vector_divmod): Likewise.
2660 (optimize_vector_constructor): Likewise.
2661 * tree-vect-loop.c: Include tree-vector-builder.h.
2662 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
2663 of build_vector. Explicitly use a stepped encoding for
2665 * tree-vect-slp.c: Include tree-vector-builder.h.
2666 (vect_get_constant_vectors): Use tree_vector_builder instead
2668 (vect_transform_slp_perm_load): Likewise.
2669 (vect_schedule_slp_instance): Likewise.
2670 * tree-vect-stmts.c: Include tree-vector-builder.h.
2671 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
2672 (vect_gen_perm_mask_any): Likewise.
2673 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
2674 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
2676 (build_vector_from_val): Likewise. Explicitly use a duplicate
2679 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
2681 * doc/generic.texi (VECTOR_CST): Describe new representation of
2683 * vector-builder.h: New file.
2684 * tree-vector-builder.h: Likewise.
2685 * tree-vector-builder.c: Likewise.
2686 * Makefile.in (OBJS): Add tree-vector-builder.o.
2687 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
2688 * tree-core.h (tree_base): Add a vector_cst field to the u union.
2689 (tree_vector): Change the number of elements to
2690 vector_cst_encoded_nelts.
2691 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
2692 (VECTOR_CST_ELTS): Delete.
2693 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
2694 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
2695 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
2696 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
2697 (VECTOR_CST_ENCODED_ELT): Likewise.
2698 (vector_cst_encoded_nelts): New function.
2699 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
2700 VECTOR_CST_NELTS_PER_PATTERN as arguments.
2701 (vector_cst_int_elt, vector_cst_elt): Declare.
2702 * tree.c: Include tree-vector-builder.h.
2703 (tree_code_size): Abort if passed VECTOR_CST.
2704 (tree_size): Update for new VECTOR_CST layout.
2705 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
2706 VECTOR_CST_NELTS_PER_PATTERN as arguments.
2707 (build_vector): Use tree_vector_builder.
2708 (vector_cst_int_elt, vector_cst_elt): New functions.
2709 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
2710 encoded elements and then create the vector in the canonical form.
2711 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
2712 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
2713 (tree_c_tests): Call test_vector_cst_patterns.
2714 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
2716 (hash_tree): Likewise.
2717 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
2718 (streamer_write_tree_header): Likewise.
2719 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
2720 (streamer_alloc_tree): Likewise. Update call to make_vector.
2721 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
2723 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2725 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
2726 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
2727 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
2729 2017-12-07 Bin Cheng <bin.cheng@arm.com>
2730 Richard Biener <rguenther@suse.de>
2732 PR tree-optimization/81303
2733 * Makefile.in (gimple-loop-interchange.o): New object file.
2734 * common.opt (floop-interchange): Reuse the option from graphite.
2735 * doc/invoke.texi (-floop-interchange): Ditto. New document for
2736 -floop-interchange and mention it for -O3.
2737 * opts.c (default_options_table): Enable -floop-interchange at -O3.
2738 * gimple-loop-interchange.cc: New file.
2739 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
2740 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
2741 * passes.def (pass_linterchange): New pass.
2742 * timevar.def (TV_LINTERCHANGE): New time var.
2743 * tree-pass.h (make_pass_linterchange): New declaration.
2744 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
2745 interchange. Record IV before/after increment in new parameters.
2746 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
2747 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
2749 (check_reduction_path): ...New function here.
2750 * tree-vectorizer.h (check_reduction_path): New declaration.
2752 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
2755 PR rtl-optimization/80818
2756 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
2757 always early clobbered.
2758 * lra-lives.c (process_bb_lives): Check input hard regs for early
2759 clobbered non-operand hard reg.
2761 2017-12-07 Jakub Jelinek <jakub@redhat.com>
2764 * tree-cfg.c (verify_gimple_assign_binary): Don't require
2765 types_compatible_p, just that TYPE_MODE is the same.
2767 2017-12-07 Martin Sebor <msebor@redhat.com>
2770 * attribs.c (empty_attribute_table): Initialize new member of
2771 struct attribute_spec.
2772 (decl_attributes): Add argument. Handle mutually exclusive
2773 combinations of attributes.
2774 (selftests::test_attribute_exclusions): New function.
2775 (selftests::attribute_c_tests): Ditto.
2776 * attribs.h (decl_attributes): Add default argument.
2777 * selftest.h (attribute_c_tests): Declare.
2778 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
2779 * tree-core.h (attribute_spec::exclusions, exclude): New type and
2781 * doc/extend.texi (Common Function Attributes): Update const and pure.
2782 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
2783 of struct attribute_spec.
2784 * config/arc/arc.c (arc_attribute_table): Same.
2785 * config/arm/arm.c (arm_attribute_table): Same.
2786 * config/avr/avr.c ( avr_attribute_table): Same.
2787 * config/bfin/bfin.c (bfin_attribute_table): Same.
2788 * config/cr16/cr16.c (cr16_attribute_table): Same.
2789 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
2790 * config/h8300/h8300.c (h8300_attribute_table): Same.
2791 * config/i386/i386.c (ix86_attribute_table): Same.
2792 * config/ia64/ia64.c (ia64_attribute_table): Same.
2793 * config/m32c/m32c.c (m32c_attribute_table): Same.
2794 * config/m32r/m32r.c (m32r_attribute_table): Same.
2795 * config/m68k/m68k.c (m68k_attribute_table): Same.
2796 * config/mcore/mcore.c (mcore_attribute_table): Same.
2797 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
2798 * config/mips/mips.c (mips_attribute_table): Same.
2799 * config/msp430/msp430.c (msp430_attribute_table): Same.
2800 * config/nds32/nds32.c (nds32_attribute_table): Same.
2801 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
2802 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
2803 * config/rl78/rl78.c (rl78__attribute_table): Same.
2804 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
2805 * onfig/rx/rx.c (rx_attribute_table): Same.
2806 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
2807 * config/sh/sh.c (sh_attribute_table): Same.
2808 * config/sparc/sparc.c (sparc_attribute_table): Same.
2809 * config/spu/spu.c (spu_attribute_table): Same.
2810 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
2811 * config/v850/v850.c (v850_attribute_table): Same.
2812 * config/visium/visium.c (visium_attribute_table): Same.
2814 2017-12-07 Tamar Christina <tamar.christina@arm.com>
2817 * config/arm/arm.c (INCLUDE_STRING): Define.
2818 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
2819 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
2822 2017-12-07 Michael Matz <matz@suse.de>
2824 Add unroll and jam pass
2826 * gimple-loop-jam.c: New file.
2827 * Makefile.in (OBJS): Add gimple-loop-jam.o.
2828 * common.opt (funroll-and-jam): New option.
2829 * opts.c (default_options_table): Add unroll-and-jam at -O3.
2830 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
2831 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
2832 * passes.def: Add pass_loop_jam.
2833 * timevar.def (TV_LOOP_JAM): Add.
2834 * tree-pass.h (make_pass_loop_jam): Declare.
2835 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
2836 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
2837 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
2838 to flow_loop_tree_node_add.
2839 (duplicate_subloops, copy_loops_to): Append to sibling list.
2840 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
2841 * doc/invoke.texi (-funroll-and-jam): Document new option.
2842 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
2844 2017-12-07 Richard Biener <rguenther@suse.de>
2846 PR tree-optimization/83296
2847 PR tree-optimization/67769
2848 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
2849 flow sensitive info in an unrelated BB.
2850 (value_replacement): Use reset_flow_sensitive_info.
2851 (minmax_replacement): Reset flow sensitive info on the def
2852 we move. Do not reset flow sensitive info in the whole BB
2853 we move the stmt to.
2854 (abs_replacement): Likewise.
2856 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
2859 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
2860 rs6000_cpu to the given -mcpu=, or to the default processor.
2862 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
2864 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
2865 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
2866 instead of rs6000_cpu_attr.
2867 (rs6000_adjust_cost): Ditto.
2868 (is_microcoded_insn): Ditto.
2869 (rs6000_adjust_priority): Ditto.
2870 (rs6000_issue_rate): Ditto.
2871 (rs6000_use_sched_lookahead): Ditto.
2872 (rs6000_use_sched_lookahead_guard): Ditto.
2873 (rs6000_sched_reorder): Ditto.
2874 (force_new_group): Ditto.
2875 * config/rs6000/rs6000.md (cpu attribute): Ditto.
2876 (group_ending_nop): Ditto.
2878 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
2880 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
2881 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
2882 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
2883 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
2885 (rs6000_reassociation_width): Ditto.
2886 (rs6000_emit_epilogue): Ditto.
2887 (rs6000_adjust_cost): Ditto.
2888 (is_microcoded_insn): Ditto.
2889 (is_cracked_insn): Ditto.
2890 (rs6000_adjust_priority): Ditto.
2891 (rs6000_sched_reorder): Ditto.
2892 (rs6000_sched_reorder2): Ditto.
2893 (insn_must_be_first_in_group): Ditto.
2894 (insn_must_be_last_in_group): Ditto.
2895 (rs6000_register_move_cost): Ditto.
2896 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
2899 2017-12-07 Julia Koval <julia.koval@intel.com>
2901 * config.gcc: Add vaesintrin.h.
2902 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
2904 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
2905 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
2907 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
2908 * config/i386/immintrin.h: Include vaesintrin.h.
2909 * config/i386/sse.md (vaesdec_<mode>): New pattern.
2910 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
2911 _mm_aesdec_epi128): New intrinsics.
2913 2017-12-06 David Malcolm <dmalcolm@redhat.com>
2915 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
2916 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
2917 c-family/c-spellcheck.cc.
2918 (best_macro_match::best_macro_match): Likewise.
2920 (struct edit_distance_traits<cpp_hashnode *>): Move to
2921 c-family/c-spellcheck.h.
2922 (class best_macro_match): Likewise.
2924 2017-12-06 Jakub Jelinek <jakub@redhat.com>
2926 PR tree-optimization/83293
2927 * gimple-ssa-strength-reduction.c (insert_initializers): Use
2928 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
2929 might insert into empty bb.
2932 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
2933 simplify for plus with :c added, and pointer_plus without that.
2934 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
2935 with undefined overflow and the conversion is not widening,
2936 perform negation in utype and only convert to type afterwards.
2937 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
2938 simplify for plus with :c added, and pointer_plus without that.
2939 If type is integral with undefined overflow and the conversion is
2940 not widening, perform minus in utype and only convert to type
2941 afterwards. Move the last pointer_diff_expr simplify into the
2944 2017-12-06 Martin Sebor <msebor@redhat.com>
2946 PR tree-optimization/82646
2947 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
2948 strncpy, not maxlen.
2950 2017-12-06 Martin Sebor <msebor@redhat.com>
2952 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
2955 PR tree-optimization/83075
2956 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
2957 strncat/strncpy don't change length of source string.
2959 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
2962 2017-11-29 Martin Aberg <maberg@gaisler.com>
2964 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
2965 to prevent b2bst errata sequence.
2966 (sqrtdf2_fix): Likewise.
2968 2017-12-06 Jakub Jelinek <jakub@redhat.com>
2970 PR tree-optimization/81945
2971 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
2972 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
2973 to dest_cfun has orig_loop_num set, either remap it to the new
2974 loop number if the loop got moved too, or clear it.
2976 2017-12-05 Steve Ellcey <sellcey@cavium.com>
2978 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
2980 (thunderx2t99_nothing): New insn reservation.
2981 (thunderx2t99_mrs): New insn reservation.
2982 (thunderx2t99_multiple): New insn reservation.
2983 (thunderx2t99_alu_basi): Add bfx to reservation.
2984 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
2986 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
2989 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
2991 2017-12-05 Bin Cheng <bin.cheng@arm.com>
2993 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
2994 tree-ssa-pre.c::remove_dead_inserted_code.
2995 * tree-ssa-dce.h: New file.
2996 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
2997 (remove_dead_inserted_code): Move and rename to function
2998 tree-ssa-dce.c::simple_dce_from_worklist.
2999 (pass_pre::execute): Update use.
3001 2017-12-05 Richard Biener <rguenther@suse.de>
3003 PR tree-optimization/83277
3004 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
3005 to code-gen liveout vars.
3007 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
3009 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
3010 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
3011 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
3014 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
3016 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
3018 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
3019 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
3022 2017-12-05 Richard Biener <rguenther@suse.de>
3024 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
3025 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
3026 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
3027 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
3029 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3031 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
3033 2017-12-05 Martin Liska <mliska@suse.cz>
3034 Jakub Jelinek <jakub@redhat.com>
3036 * doc/invoke.texi: Document the options.
3037 * flag-types.h (enum sanitize_code): Add
3038 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
3039 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
3040 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
3041 * opts.c: Define new sanitizer options.
3042 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
3043 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
3045 2017-12-05 Julia Koval <julia.koval@intel.com>
3047 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
3048 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
3049 (ix86_handle_option): Handle -mavx512vnni.
3050 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
3051 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
3052 * config/i386/i386-c (__AVX512VNNI__): New.
3053 * config/i386/i386.c (ix86_target_string): Handle new option.
3054 (ix86_valid_target_attribute_inner_p): Handle new option.
3055 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
3056 * config/i386/i386.opt (mavx512vnni): New option.
3058 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
3061 * athlon.md: Disable for generic.
3062 * haswell.md: Enable for generic.
3063 * i386.c (ix86_sched_init_global): Add core hooks for generic.
3064 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
3066 (ix86_adjust_cost): Move generic to haswell path.
3068 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
3070 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
3071 instead of MEM_P in a couple more places. Fix formatting issues.
3073 2017-12-04 Jim Wilson <jimw@sifive.com>
3075 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
3076 instead of GP_REG_LAST-1.
3077 (riscv_adjust_libcall_cfi_prologue): Likewise.
3078 (riscv_adjust_libcall_cri_epilogue): Likewise.
3079 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
3082 2017-12-04 Luis Machado <luis.machado@linaro.org>
3084 * ipa-pure-const.c (check_decl): Add missing newline.
3085 (state_from_flags): Likewise.
3087 2017-12-04 Jeff Law <law@redhat.com>
3089 PR tree-optimizatin/78496
3090 * gimple-ssa-evrp-analyze.h
3091 (evrp_range_analyzer::get_vr_values): Simplify.
3092 * gimple-ssa-evrp-analyze.c: Corresponding changes.
3093 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
3094 and gimple-ssa-evrp-analyze.h.
3095 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
3096 (simplify_stmt_for_jump_threading): Copy a blob of code from
3097 tree-vrp.c to use ranges to simplify statements.
3098 (dom_opt_dom_walker::before_dom_children): Call
3099 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
3100 (dom_opt_dom_walker::after_dom_children): Similarly for
3101 evrp_range_analyzer::leave.
3102 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
3105 * gimple-ssa-evrp-analyze.c
3106 (evrp_range_analyzer::extract_range_from_stmt): Always use
3107 vr_values::update_value_range so preexisting range info is
3108 medged with new range info, even if the new range is VR_VARYING.
3110 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
3112 * combine.c: Adjust comment.
3113 (use_crosses_set_p): Delete.
3114 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
3115 (try_combine): Ditto.
3117 2017-12-04 Richard Biener <rguenther@suse.de>
3119 PR tree-optimization/83255
3120 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
3121 Re-add zero-iteration check.
3123 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
3125 PR rtl-optimization/83245
3126 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
3127 hard registers as earlyclobber, also if not in an asm.
3129 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
3133 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3136 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
3137 rs6000_cpu based on cpu_index, not tune_index.
3139 2017-12-04 Richard Biener <rguenther@suse.de>
3141 PR tree-optimization/83238
3142 * graphite-scop-detection.c (scop_detection::merge_sese): Make
3143 code match comment, rejecting invalid SESE regions.
3145 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
3147 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
3148 require base operand is a REG_POINTER prior to reload on targets
3149 with non-equivalent space registers.
3151 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
3153 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
3154 (update_specialized_profile): Fix updating of counts.
3155 (perhaps_add_new_callers): Likewise.
3157 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
3160 * x86-tune.def: Remove obsolette FIXMEs.
3161 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
3162 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
3163 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
3165 (X86_TUNE_PAD_RETURNS): Disable for generic.
3167 2017-12-02 Jakub Jelinek <jakub@redhat.com>
3169 PR tree-optimization/83170
3170 PR tree-optimization/83241
3171 * gimple-ssa-store-merging.c
3172 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
3173 gimple_vuse (ins_stmt) in case it has changed.
3174 (imm_store_chain_info::output_merged_store): Likewise.
3176 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
3180 * tree-cfg.c (pass_warn_function_return::execute): Handle
3181 __builtin_ubsan_handle_missing_return like __builtin_unreachable
3182 with BUILTINS_LOCATION.
3186 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
3187 is BLKmode for vector field with vector raw mode, use TYPE_MODE
3188 instead of DECL_MODE.
3190 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
3191 last argument to rtx pointer.
3192 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
3193 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
3194 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
3195 depending on the chosen ISAs.
3196 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
3197 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
3198 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
3200 * config/i386/sse.md (mov<mode>_internal): Likewise.
3201 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
3203 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3205 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
3207 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3209 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
3210 parameters from prototype.
3211 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
3212 parameters. Don't print a comment.
3213 (emit_fusion_gpr_load): Adjust.
3214 (emit_fusion_load_store): Adjust.
3215 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
3216 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
3217 comment on the second line.
3219 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3222 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
3223 rs6000_cpu based on cpu_index, not tune_index.
3225 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3227 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
3228 which_alternative instead of which_alternative + 1.
3229 (output_asm_insn): Print an extra tab if the template is short.
3231 2017-12-01 Jim Wilson <jimw@sifive.com>
3233 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
3235 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
3236 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
3238 * doc/tm.texi: Regenerate.
3240 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
3243 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
3244 whether we can allocate pseudos before trying to fix an address.
3245 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
3246 memory address is indexed or indirect.
3247 (floatuns_<mode>si2_hw2): Likewise.
3249 2017-12-01 Jason Merrill <jason@redhat.com>
3251 * Makefile.in (TAGS): Add c-family/*.cc.
3253 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
3255 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
3256 (call_value_insn): Likewise.
3257 (sibcall_insn): Likewise.
3258 (sibcall_value_insn): Likewise.
3259 (movsi_aarch64): Likewise.
3260 (movdi_aarch64): Likewise.
3261 (add_losym_): Likewise.
3262 (ldr_got_small_): Likewise.
3263 (ldr_got_small_sidi): Likewise.
3264 (ldr_got_small_28k_): Likewise.
3265 (ldr_got_small_28k_sidi): Likewise.
3266 * config/aarch64/aarch64.c (aarch64_print_address_internal):
3267 Move output_addr_const to symbolic case. Add error check.
3269 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3271 * config/s390/predicates.md (plus16_Q_operand): New predicate.
3272 * config/s390/s390.md: Disable MVC merging peephole if it would
3273 disable operand forwarding.
3274 (new peephole2): Split MVCs if it would turn them into up to 2
3277 2017-12-01 Richard Biener <rguenther@suse.de>
3279 PR tree-optimization/83232
3280 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
3281 detection of same access. Instead of breaking the group here
3282 do not consider the duplicate. Add comment explaining real fix.
3284 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
3286 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
3288 2017-12-01 Sudakshina Das <sudi.das@arm.com>
3290 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
3292 2017-12-01 Jakub Jelinek <jakub@redhat.com>
3294 * function.h (struct function): Remove cilk_frame_decl,
3295 is_cilk_function and calls_cilk_spawn fields.
3296 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
3298 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
3299 cilk_elemental field.
3300 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
3301 * target.def: Adjust comment.
3302 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
3303 Don't test cilk_elemental.
3305 PR tree-optimization/83233
3306 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
3307 bswap_stat name for the struct.
3310 * tree.h (SWITCH_BREAK_LABEL_P): Define.
3311 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
3312 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
3313 SWITCH_BREAK_LABEL_P set on the label.
3314 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
3315 added for default case if it was missing and not all cases covered.
3316 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
3317 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
3319 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
3320 to avoid -Wimplicit-fallthrough warning.
3321 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
3322 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
3324 PR tree-optimization/83221
3325 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
3327 (init_reassoc): Formatting fix.
3330 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
3331 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
3335 * tree-cfg.c: Include asan.h.
3336 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
3338 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
3340 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
3342 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
3344 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
3345 vpcmpeqd instruction.
3347 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
3349 * config/i386/i386.c (standard_sse_constant_opcode): Fix
3350 registers type for 128bit mode.
3352 2017-11-30 Jakub Jelinek <jakub@redhat.com>
3354 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
3356 * gengtype-state.c (read_state_pair): Likewise.
3357 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
3358 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
3359 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
3360 * ifcvt.c (noce_try_store_flag_constants): Likewise.
3361 * tree-ssa-ccp.c (ccp_finalize): Likewise.
3362 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
3363 * builtins.c (fold_builtin_3): Likewise.
3364 * graphite-scop-detection.c
3365 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
3366 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
3369 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
3372 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
3373 parameter. Return true if ignore_decl_rtl_set_p is true and other
3374 conditions are satisfied.
3375 * asan.h (asan_protect_global): Add new parameter.
3376 * varasm.c (categorize_decl_for_section): Pass true as second parameter
3377 to asan_protect_global calls.
3379 2017-11-30 Jim Wilson <jimw@sifive.com>
3381 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
3382 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
3383 -mstrict-align, add info on default value. Delete redundant lines for
3384 -mabi. Add missing -mexplicit-relocs docs.
3386 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
3388 * config/arc/arc.md (trap): New pattern.
3390 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
3392 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
3393 instruction to end into a delay slot.
3394 * config/arc/arc.md (cond_delay_insn): Check if the instruction
3395 can be placed into a delay slot against reg_note.
3397 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
3399 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
3400 labels number of usages.
3402 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
3404 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
3406 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
3408 2017-11-30 Jakub Jelinek <jakub@redhat.com>
3411 * internal-fn.c (expand_mul_overflow): Optimize unsigned
3412 multiplication by power of 2 constant into two shifts + comparison.
3414 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
3417 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
3419 2017-11-30 Richard Biener <rguenther@suse.de>
3421 PR tree-optimization/83202
3422 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
3423 allow_peel argument and guard peeling.
3424 (canonicalize_loop_induction_variables): Likewise.
3425 (canonicalize_induction_variables): Pass false.
3426 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
3427 peeling from cunrolli.
3429 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
3431 * combine.c (try_combine): Print a message to dump file whenever
3432 I0, I1, or I2 cannot be combined into I3.
3434 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
3436 PR rtl-optimization/83156
3437 PR rtl-optimization/82621
3438 * combine.c (try_combine): Don't split an I2 if one of the dests is
3439 set again before I3. Allow unused dests.
3441 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
3443 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
3445 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
3447 PR rtl-optimization/80818
3448 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
3449 recursively. Use insn code for clobber.
3450 (lra_set_insn_recog_data): Pass the new arg to
3451 collect_non_operand_hard_regs.
3452 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
3454 (lra_update_insn_regno_info): Pass insn to
3455 add_regs_to_insn_regno_info.
3457 2017-11-29 Jim Wilson <jimw@sifive.com>
3458 Andrew Waterman <andrew@sifive.com>
3460 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
3461 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
3462 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
3463 use SINGLE_SHIFT_COST.
3464 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
3465 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
3467 2017-11-29 Julia Koval <julia.koval@intel.com>
3469 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
3470 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
3471 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
3472 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
3473 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
3474 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
3475 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
3476 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
3477 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
3478 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
3479 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
3480 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
3481 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
3482 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
3483 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
3484 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
3485 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
3486 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
3487 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
3488 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
3489 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
3490 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
3491 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
3492 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
3493 __builtin_ia32_vpshldv_v2di_mask,
3494 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
3495 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
3496 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
3498 2017-11-29 Julia Koval <julia.koval@intel.com>
3500 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
3501 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
3502 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
3503 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
3504 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
3505 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
3506 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
3507 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
3508 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
3509 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
3510 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
3511 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
3512 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
3513 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
3514 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
3515 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
3516 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
3517 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
3518 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
3519 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
3520 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
3522 2017-11-29 Daniel Cederman <cederman@gaisler.com>
3524 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
3525 movsi_pic_gotdata_op instruction as a load for the UT699 errata
3528 2017-11-29 Martin Aberg <maberg@gaisler.com>
3530 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
3531 to prevent b2bst errata sequence.
3532 (sqrtdf2_fix): Likewise.
3534 2017-11-29 Daniel Cederman <cederman@gaisler.com>
3536 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
3537 (div_sqrt_insn_p): New function.
3538 (sparc_do_work_around_errata): Insert NOP instructions to
3539 prevent sequences that could trigger the TN-0013 errata for
3540 certain LEON3 processors.
3541 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
3542 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
3543 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
3544 (in_branch_delay): Prevent div and sqrt in delay slot if
3546 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
3548 2017-11-29 Daniel Cederman <cederman@gaisler.com>
3550 * config/sparc/sparc.c (atomic_insn_p): New function.
3551 (sparc_do_work_around_errata): Insert NOP instructions to
3552 prevent sequences that could trigger the TN-0010 errata for
3554 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
3555 instruction referable in atomic_insns_p.
3557 2017-11-29 Daniel Cederman <cederman@gaisler.com>
3559 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
3560 (atomic_compare_and_swap_leon3_1): Likewise.
3563 2017-11-29 Daniel Cederman <cederman@gaisler.com>
3565 * config/sparc/sparc.c (fpop_insn_p): New function.
3566 (sparc_do_work_around_errata): Insert NOP instructions to
3567 prevent sequences that could trigger the TN-0012 errata for
3569 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
3570 * config/sparc/sparc.md (fix_gr712rc): New attribute.
3571 (in_branch_annul_delay): Prevent floating-point instructions
3572 in delay slot of annulled integer branch.
3574 2017-11-29 Richard Biener <rguenther@suse.de>
3576 PR tree-optimization/83202
3577 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
3579 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
3580 nodes vectorized to the same stmts multiple times.
3581 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
3582 (vect_analyze_slp_instance): Adjust.
3583 (scalar_stmts_to_slp_tree_map_t): New typedef.
3584 (vect_schedule_slp_instance): Add a map recording the SLP node
3585 representing the vectorized stmts for a set of scalar stmts.
3586 Avoid code-generating redundancies.
3587 (vect_schedule_slp): Allocate map and pass it down.
3589 2017-11-29 Nathan Sidwell <nathan@acm.org>
3592 * tree.c (build_complex_type): Fix canonicalization. Only fill in
3595 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
3597 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
3598 cases for printing LDP/STP memory addresses.
3599 (aarch64_print_address_internal): Renamed from
3600 aarch64_print_operand_address, added parameter, add Pmode check.
3601 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
3602 (aarch64_print_operand_address): Indirect to
3603 aarch64_print_address_internal.
3604 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
3605 'y' operand output specifier.
3607 2017-11-29 Jakub Jelinek <jakub@redhat.com>
3610 * tree.c (build_simple_mem_ref_loc): Handle
3611 get_addr_base_and_unit_offset returning a MEM_REF.
3614 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
3617 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
3620 2017-11-28 David Malcolm <dmalcolm@redhat.com>
3622 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
3624 (layout::move_to_column): Likewise.
3626 2017-11-28 Jakub Jelinek <jakub@redhat.com>
3629 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
3630 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
3632 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3633 Martin Jambor <mjambor@suse.cz>
3636 * tree.h (expr_type_first_operand_type_p): Declare
3637 * tree.c (expr_type_first_operand_type_p): New function.
3638 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
3639 (ipa_value_from_jfunc): Adjust declaration.
3640 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
3641 Use it as result type for arithmetics, unless it is NULL in which case
3642 be more conservative.
3643 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
3644 ipa_get_jf_pass_through_result.
3645 (propagate_vals_across_pass_through): Likewise.
3646 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
3647 is to propagate_vals_across_pass_through.
3648 (propagate_constants_across_call): Pass PARM_TYPE to
3649 propagate_scalar_across_jump_function.
3650 (find_more_scalar_values_for_callers_subset): Pass parameter type to
3651 ipa_value_from_jfunc.
3652 (cgraph_edge_brings_all_scalars_for_node): Likewise.
3653 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
3654 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
3655 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
3656 target_type, pass it to ipa_value_from_jfunc.
3657 (update_indirect_edges_after_inlining): Pass parameter type to
3658 try_make_edge_direct_simple_call.
3660 2017-11-28 Jeff Law <law@redhat.com>
3662 * gimple-ssa-evrp-analyze.c
3663 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
3664 refine ranges if scev_initialized_p returns true.
3665 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
3667 2017-11-28 Julia Koval <julia.koval@intel.com>
3669 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
3670 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
3671 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
3672 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
3673 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
3674 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
3675 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
3676 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
3677 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
3678 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
3679 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
3680 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
3681 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
3682 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
3683 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
3684 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
3685 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
3686 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
3687 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
3688 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
3689 __builtin_ia32_vpshrd_v2di_mask): New builtins.
3690 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
3692 2017-11-28 Julia Koval <julia.koval@intel.com>
3694 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
3695 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
3696 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
3697 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
3698 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
3699 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
3700 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
3701 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
3702 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
3703 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
3704 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
3705 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
3706 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
3707 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
3708 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
3709 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
3710 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
3711 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
3712 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
3713 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
3714 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
3715 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
3716 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
3717 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
3718 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
3719 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
3720 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
3721 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
3722 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
3723 __builtin_ia32_vpshld_v2di_mask): New builtins.
3724 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
3725 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
3727 2017-11-28 Richard Biener <rguenther@suse.de>
3729 PR tree-optimization/80776
3730 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
3732 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
3734 (evrp_range_analyzer::record_ranges_from_incoming_edges):
3735 If the incoming edge is an effective fallthru because the other
3736 edge only reaches a __builtin_unreachable () then record ranges
3737 derived from the controlling condition in SSA info.
3738 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
3739 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
3741 2017-11-28 Olivier Hainque <hainque@adacore.com>
3743 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
3744 on mingw build hosts.
3746 2017-11-28 Jakub Jelinek <jakub@redhat.com>
3749 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
3751 2017-11-28 Richard Biener <rguenther@suse.de>
3754 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
3755 copies generated from memcpy use a character array as reference
3758 2017-11-28 Julia Koval <julia.koval@intel.com>
3759 Sebastian Peryt <sebastian.peryt@intel.com>
3761 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
3762 c-family/c-cilkplus.o, c-family/array-notation-common.o,
3763 cilk-common.o, cilk.h, cilk-common.c): Remove.
3765 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
3766 * builtins.c (is_builtin_name): Remove cilkplus condition.
3767 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
3768 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
3769 cilk-builtins.def, cilkplus.def): Remove.
3770 * cif-code.def (CILK_SPAWN): Remove.
3771 * cilk-builtins.def: Delete.
3772 * cilk-common.c: Ditto.
3774 * cilkplus.def: Ditto.
3775 * config/darwin.h (fcilkplus): Delete.
3776 * cppbuiltin.c: Ditto.
3777 * doc/extend.texi: Remove cilkplus doc.
3778 * doc/generic.texi: Ditto.
3779 * doc/invoke.texi: Ditto.
3780 * doc/passes.texi: Ditto.
3781 * gcc.c (fcilkplus): Remove.
3782 * gengtype.c (cilk.h): Remove.
3783 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
3785 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
3787 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
3789 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
3790 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
3791 cilkplus conditions.
3792 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
3793 inline_read_section): Ditto.
3794 * ipa-inline-analysis.c (cilk.h): Remove.
3795 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
3796 * lto-wrapper.c (merge_and_complain, append_compiler_options,
3797 append_linker_options): Remove condition for fcilkplus.
3798 * lto/lto-lang.c (cilk.h): Remove.
3799 (lto_init): Remove condition for fcilkplus.
3800 * omp-expand.c (expand_cilk_for_call): Delete.
3801 (expand_omp_taskreg, expand_omp_for_static_chunk,
3802 expand_omp_for): Remove cilkplus
3804 (expand_cilk_for): Delete.
3805 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
3806 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
3807 execute_lower_omp, diagnose_sb_0): Ditto.
3808 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
3809 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
3810 * tree-nested.c: Ditto.
3811 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
3812 (dump_generic_node): Ditto.
3813 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
3814 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
3816 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
3818 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
3820 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
3821 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
3822 for vector mode and !TARGET_SIMD.
3824 2017-11-28 Jakub Jelinek <jakub@redhat.com>
3826 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
3828 * tree.h (SWITCH_LABELS): Remove.
3829 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
3830 assert SWITCH_BODY is non-NULL.
3831 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
3833 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
3835 PR tree-optimization/80788
3836 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
3837 has TREE_OVERFLOW set, call drop_tree_overflow.
3839 2017-11-28 Richard Biener <rguenther@suse.de>
3841 PR tree-optimization/83158
3842 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
3844 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
3847 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
3848 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
3850 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
3852 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
3853 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
3855 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
3858 * varasm.c (compare_constant): Take the mode of the constants into
3859 account when comparing floating point constants.
3861 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
3863 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
3864 from explicit instantiation of debug_helper.
3865 * vec.h (DEFINE_DEBUG_VEC): Ditto.
3867 2017-11-27 Richard Biener <rguenther@suse.de>
3869 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
3872 2017-11-27 Richard Biener <rguenther@suse.de>
3874 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
3875 (type_hash_canon): Also clear int_cst_hash_table entry for
3877 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
3879 2017-11-27 Tamar Christina <tamar.christina@arm.com>
3881 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
3882 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
3883 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
3885 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
3887 * hash-map.h (gt_cleare_cache): Avoid UB.
3889 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
3891 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
3892 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
3893 (decide_unroll_runtime_iterations): Likewise.
3894 (decide_unroll_stupid): Likewise.
3896 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3899 * config/i386/i386.md: Add a loop with incssp.
3901 2017-11-27 Martin Jambor <mjambor@suse.cz>
3903 PR tree-optimization/81248
3904 * tree-sra.c (splice_param_accesses): Remove size check.
3905 (decide_one_param_reduction): Fix size check.
3906 * gimple-pretty-print.c (dump_profile): Silence warning.
3907 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
3909 2017-11-27 Jakub Jelinek <jakub@redhat.com>
3912 * dbxout.c (lastlineno): New variable.
3913 (dbx_debug_hooks): Use dbxout_switch_text_section as
3914 switch_text_section debug hook.
3915 (dbxout_function_end): Switch to current_function_section
3916 rather than function_section. If crtl->has_bb_partition,
3917 output just one N_FUN, depending on in_cold_section_p.
3918 (dbxout_source_line): Remember last lineno in lastlineno.
3919 (dbxout_switch_text_section): New function.
3920 (dbxout_function_decl): Adjust dbxout_block caller.
3921 (dbx_block_with_cold_children): New function.
3922 (dbxout_block): Return true if any LBRAC/RBRAC have been
3923 emitted. Use dbx_block_with_cold_children at depth == 0
3924 in second partition. Add PARENT_BLOCKNUM argument, pass
3925 it optionally adjusted to children. Output LBRAC/RBRAC
3926 around recursive call only if the block is in the current
3927 partition, if not and anything was output, emit empty
3929 * final.c (final_scan_insn): Compute cold_function_name
3930 before calling switch_text_section debug hook. Call
3931 that hook even if dwarf2out_do_frame if not emitting
3935 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
3936 TREE_READONLY decls.
3938 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
3940 PR rtl-optimization/82488
3941 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
3943 2017-11-26 Julia Koval <julia.koval@intel.com>
3945 * config/i386/i386.c (processor_target_table): Add skylake_cost for
3947 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
3950 2017-11-26 Julia Koval <julia.koval@intel.com>
3952 * config/i386/driver-i386.c (host_detect_local_cpu):
3953 Detect skylake-avx512.
3955 2017-11-26 Julia Koval <julia.koval@intel.com>
3957 * config.gcc: Add -march=cannonlake.
3958 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
3959 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
3960 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
3961 (PTA_CANNONLAKE): New.
3962 (processor_target_table): Add cannonlake.
3963 (ix86_option_override_internal): Ditto.
3964 (fold_builtin_cpu): Ditto.
3965 (get_builtin_code_for_version): Handle cannonlake.
3966 (M_INTEL_COREI7_CANNONLAKE): New.
3967 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
3968 * doc/invoke.texi: Add -march=cannonlake.
3970 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
3972 * plugin.c (add_new_plugin): Use platform-specific library extensions.
3973 (try_init_one_plugin): Alternative implementation for MinGW.
3974 * Makefile.in (plugin_implib): New.
3975 (gengtype-lex.c): Fix broken AIX workaround.
3976 * configure: Regenerate.
3977 * doc/plugins.texi: Document support for MinGW.
3979 2017-11-25 Jakub Jelinek <jakub@redhat.com>
3981 PR rtl-optimization/81553
3982 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
3983 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
3984 is a shift where C1 has different mode than the whole shift, use C1's
3985 mode for MULT rather than the shift's mode.
3988 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
3989 builtins not enabled in the currently selected ISA.
3991 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
3993 PR tree-optimization/71026
3994 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
3995 (insert_reciprocals): Change to insert reciprocals before a division
3996 by a square and to insert the square of a reciprocal.
3997 (execute_cse_reciprocals_1): Change to consider division by a square.
3998 (register_division_in): Add importance parameter.
4000 2017-11-24 Richard Biener <rguenther@suse.de>
4002 PR tree-optimization/82402
4003 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
4004 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
4006 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
4008 * match.pd (0-ptr): New transformation.
4010 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
4013 * ipa-inline.c (inline_small_functions): Set current badnes correctly
4014 when skipping checking.
4016 2017-11-24 Richard Biener <rguenther@suse.de>
4018 PR tree-optimization/83128
4019 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
4020 (vn_reference_lookup_3): Likewise.
4022 2017-11-24 Jakub Jelinek <jakub@redhat.com>
4025 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
4026 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
4029 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
4030 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
4031 check for the latter with gimple_call_builtin_p. Do not handle
4032 BUILT_IN_STPNCPY_CHK which is not a pass through call.
4034 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
4036 * config/arm/arm_neon.h: Fix pragma GCC push_options before
4039 2017-11-23 Julia Koval <julia.koval@intel.com>
4041 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
4042 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
4043 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
4044 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
4045 _mm512_maskz_expandloadu_epi16): New intrinsics.
4046 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
4047 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
4048 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
4049 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
4050 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
4051 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
4052 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
4053 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
4054 _mm256_maskz_expandloadu_epi8): New intrinsics.
4055 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
4056 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
4057 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
4058 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
4059 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
4060 * config/i386/sse.md (VI248_VLBW): New iterator.
4061 (expand<mode>_mask, expand<mode>_maskz): New patterns.
4063 2017-11-23 Julia Koval <julia.koval@intel.com>
4065 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
4066 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
4067 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
4068 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
4069 _mm512_mask_compressstoreu_epi16): New.
4070 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
4071 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
4072 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
4073 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
4074 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
4075 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
4076 _mm256_mask_compressstoreu_epi8): New.
4077 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
4078 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
4079 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
4080 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
4081 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
4082 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
4083 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
4084 __builtin_ia32_compresshi128_mask,
4085 __builtin_ia32_compressstoreuqi512_mask,
4086 __builtin_ia32_compressstoreuhi512_mask,
4087 __builtin_ia32_compressstoreuqi256_mask,
4088 __builtin_ia32_compressstoreuqi128_mask,
4089 __builtin_ia32_compressstoreuhi256_mask,
4090 __builtin_ia32_compressstoreuhi128_mask): New builtins.
4091 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
4093 (ix86_expand_special_args_builtin): Handle new types.
4094 (s4fma_expand): Handle new builtin array.
4095 * config/i386/immintrin.h: Include new headers.
4096 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
4097 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
4099 2017-11-23 Jakub Jelinek <jakub@redhat.com>
4101 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
4102 fixes. Declare temp and g variables at the top in order to avoid
4103 {} in most of the cases.
4105 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
4107 * match.pd (ptr-0): New transformation.
4109 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
4111 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
4112 (struct addr_mode_cost_table): New.
4113 (struct tune_params): Add field addr_mode_costs.
4114 * config/arm/arm.c (generic_addr_mode_costs): New.
4115 (arm_slowmul_tune): Initialise addr_mode_costs field.
4116 (arm_fastmul_tune): Likewise.
4117 (arm_strongarm_tune): Likewise.
4118 (arm_xscale_tune): Likewise.
4119 (arm_9e_tune): Likewise.
4120 (arm_marvell_pj4_tune): Likewise.
4121 (arm_v6t2_tune): Likewise.
4122 (arm_cortex_tune): Likewise.
4123 (arm_cortex_a8_tune): Likewise.
4124 (arm_cortex_a7_tune): Likewise.
4125 (arm_cortex_a15_tune): Likewise.
4126 (arm_cortex_a35_tune): Likewise.
4127 (arm_cortex_a53_tune): Likewise.
4128 (arm_cortex_a57_tune): Likewise.
4129 (arm_exynosm1_tune): Likewise.
4130 (arm_xgene1_tune): Likewise.
4131 (arm_cortex_a5_tune): Likewise.
4132 (arm_cortex_a9_tune): Likewise.
4133 (arm_cortex_a12_tune): Likewise.
4134 (arm_cortex_a73_tune): Likewise.
4135 (arm_v7m_tune): Likewise.
4136 (arm_cortex_m7_tune): Likewise.
4137 (arm_v6m_tune): Likewise.
4138 (arm_fa726te_tune): Likewise.
4139 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
4141 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
4143 * config/arm/arm.c (arm_mem_costs): New function.
4144 (arm_rtx_costs_internal): Use arm_mem_costs.
4146 2017-11-23 Mark Wielaard <mark@klomp.org>
4148 * dwarf2out.c (init_sections_and_labels): Use generation to create
4149 unique ranges_section_label and ranges_base_label. Return generation.
4150 (output_rnglists): Add generation argument. Use generation to create
4151 unique ranges labels.
4152 (dwarf2out_finish): Get generation from init_sections_and_labels
4153 and pass generation to output_rnglists.
4155 2017-11-23 Mike Stump <mikestump@comcast.net>
4156 Eric Botcazou <ebotcazou@adacore.com>
4158 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
4159 * cfgloop.h (struct loop): Add unroll field.
4160 * function.h (struct function): Add has_unroll bitfield.
4161 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
4162 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
4163 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
4164 (pass_rtl_unroll_loops::gate): Likewise.
4165 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
4166 for which loop->unroll==1.
4167 (decide_unroll_constant_iterations): Use note for consistency and
4168 take loop->unroll into account. Return early if loop->unroll is set.
4169 Fix thinko in existing test.
4170 (decide_unroll_runtime_iterations): Use note for consistency and
4171 take loop->unroll into account.
4172 (decide_unroll_stupid): Likewise.
4173 * lto-streamer-in.c (input_cfg): Read loop->unroll.
4174 * lto-streamer-out.c (output_cfg): Write loop->unroll.
4175 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
4177 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
4178 (print_loop): Print loop->unroll if set.
4179 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
4180 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
4181 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
4183 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
4184 loop->unroll is set and smaller than the trip count. Otherwise bypass
4185 entirely the heuristics if loop->unroll is set. Remove dead note.
4186 Fix off-by-one bug in other note.
4187 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
4188 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
4190 (tree_unroll_loops_completely): Make static.
4191 (pass_complete_unroll::execute): Use correct type for variable.
4192 (pass_complete_unrolli::execute): Fix formatting.
4193 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
4195 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
4197 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
4198 enable when TARGET_PREFER_AVX128 is set.
4200 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
4202 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
4204 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
4205 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
4207 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
4209 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
4210 frequencies. Fix estimation of aggregate parameters.
4212 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
4214 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
4217 2017-11-23 Tom de Vries <tom@codesourcery.com>
4219 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
4221 2017-11-23 Tom de Vries <tom@codesourcery.com>
4223 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
4226 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
4229 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
4230 sibcall_value_pcrel_fdpic): Use local variable instead of
4232 (calli_tbr_rel): Add missing operand 2.
4233 (call_valuei_tbr_rel): Add missing operand 3.
4235 2017-11-23 Jakub Jelinek <jakub@redhat.com>
4238 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
4239 bitpos/bitsize covering the whole destination, use store_expr only if
4240 the complex mode is the same. Otherwise, use expand_normal and if
4241 it returns CONCAT, subreg each part separately instead of trying to
4242 subreg the whole result.
4244 2017-11-23 Richard Biener <rguenther@suse.de>
4246 PR tree-optimization/23094
4247 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
4248 come here from walking over backedges in the first iteration.
4249 (vn_reference_lookup_3): Skip clobbers that store the same value.
4251 2017-11-23 Richard Biener <rguenther@suse.de>
4253 PR tree-optimization/81403
4254 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
4255 a block we need a leader relative to.
4256 (phi_translate_1): For nary processing require a leader from
4257 get_representative_for given we run expression simplification
4258 using match-and-simplify. Remove previous fix.
4260 2017-11-22 Jeff Law <law@redhat.com>
4262 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
4263 Use new method allocate_value_range rather than accessing the
4264 vrp_value_range_pool data member directly.
4265 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
4266 to use extract_range_from_stmt method to avoid need for
4267 extract_range_from_assignment method.
4268 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
4269 method rather than setting values_propgated data member directly.
4270 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
4271 and values propagated data members and extract_range_from_assignment
4272 method. Reorder private data members to conform to standards.
4273 Add new methods set_lattice_propagation_complete and
4274 allocate_value_range.
4276 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
4278 PR rtl-optimization/83030
4279 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
4280 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
4281 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
4282 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
4283 CROSSING_JUMP_P flag is already set before setting it.
4284 * cfgrtl.c (fixup_partition_crossing): Likewise.
4285 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
4288 2017-11-22 Jakub Jelinek <jakub@redhat.com>
4290 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
4291 Handle the case where both arguments are using gen_const_vec_series.
4293 2017-11-22 David Malcolm <dmalcolm@redhat.com>
4296 * pretty-print.c (pp_format): Move quoting implementation to
4297 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
4298 to pass address of "quote" local.
4299 (pp_begin_quote): New function.
4300 (pp_end_quote): New function.
4301 * pretty-print.h (printer_fn): Convert penultimate param from bool
4303 (pp_begin_quote): New decl.
4304 (pp_end_quote): New decl.
4305 * tree-diagnostic.c (default_tree_printer): Convert penultimate
4306 param from bool to bool *.
4307 * tree-diagnostic.h (default_tree_printer): Likewise.
4309 2017-11-22 Jeff Law <law@redhat.com>
4311 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
4312 Initialize vr_values.
4313 (evrp_range_analyzer::try_find_new_range): Call methods attached to
4314 vr_values via vr_values class instance rather than delegators.
4315 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
4316 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
4317 (evrp_range_analyzer::push_value_range): Likewise.
4318 (evrp_range_analyzer::pop_value_range): Likewise.
4319 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
4320 most delegators. Those remaining are exposed as public interfaces.
4321 Make vr_values a pointer and private.
4322 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
4324 (evrp_range_analyzer::get_vr_value): New method.
4325 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
4326 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
4327 (class evrp_dom_walker): Attach evrp_folder class, initialize
4328 it in the ctor. Remove temporary delegators.
4329 (evrp_dom_walker::before_dom_children): Call methods in attached
4330 evrp_range_analyzer class via class instance pointer. Use
4331 free value_range_constant_singleton to remove need for
4332 op_with_constant_singleton_value delegator method. Do not
4333 create a vrp_prop class instance for every call! Narrow
4334 scope of a couple variables.
4335 (evrp_dom_walker::cleanup): Call methods in attached
4336 evrp_range_analyzer class via class instance pointer.
4337 * vr-values.h (class vr_values): Privatize many methods and
4340 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
4342 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
4343 * cfgexpand.c (expand_debug_expr): Remove handling for them.
4344 * expr.c (expand_expr_real_2): Likewise.
4345 * fold-const.c (const_unop): Likewise.
4346 * optabs-tree.c (optab_for_tree_code): Likewise.
4347 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
4348 * tree-inline.c (estimate_operator_cost): Likewise.
4349 * tree-pretty-print.c (dump_generic_node): Likewise.
4350 (op_code_prio): Likewise.
4351 (op_symbol_code): Likewise.
4352 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
4353 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
4354 * internal-fn.c (direct_internal_fn_optab): New function.
4355 (direct_internal_fn_array, direct_internal_fn_supported_p
4356 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
4357 * fold-const-call.c (fold_const_reduction): New function.
4358 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
4360 * tree-vect-loop.c: Include internal-fn.h.
4361 (reduction_code_for_scalar_code): Rename to...
4362 (reduction_fn_for_scalar_code): ...this and return an internal
4364 (vect_model_reduction_cost): Take an internal_fn rather than
4366 (vect_create_epilog_for_reduction): Likewise. Build calls rather
4368 (vectorizable_reduction): Use internal functions rather than tree
4369 codes for the reduction operation. Update calls to the functions
4371 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
4372 Use calls to internal functions rather than REDUC tree codes.
4373 * config/aarch64/aarch64-simd.md: Update comment accordingly.
4375 2017-11-22 Olivier Hainque <hainque@adacore.com>
4377 * config/vxworks.c (vxworks_override_options): Pick default
4378 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
4379 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
4380 DWARF_GNAT_ENCODINGS_DEFAULT.
4381 * config/vxworksae.h: Likewise.
4383 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
4385 PR tree-optimization/83104
4386 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
4389 2017-11-22 Marek Polacek <polacek@redhat.com>
4390 H.J. Lu <hongjiu.lu@intel.com>
4391 Jason Merrill <jason@redhat.com>
4396 * calls.c (initialize_argument_information): Call
4397 warn_parameter_passing_abi target hook.
4398 (store_one_arg): Use 0 for empty record size. Don't push 0 size
4399 argument onto stack.
4400 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
4401 * common.opt: Update -fabi-version description.
4402 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
4403 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
4405 (ix86_is_empty_record): New function.
4406 (ix86_warn_parameter_passing_abi): New function.
4407 (TARGET_EMPTY_RECORD_P): Redefine.
4408 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
4409 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
4410 * doc/tm.texi: Regenerated.
4411 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
4412 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
4413 * dwarf2out.c (get_ultimate_context): Move to tree.c.
4414 * explow.c (hard_function_value): Call arg_int_size_in_bytes
4415 instead of int_size_in_bytes.
4416 * expr.c (copy_blkmode_to_reg): Likewise.
4417 * function.c (aggregate_value_p): Return 0 for empty types.
4418 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
4420 (locate_and_pad_parm): Call arg size_in_bytes instead
4422 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
4423 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
4424 * target.def (empty_record_p, warn_parameter_passing_abi): New target
4426 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
4427 (std_gimplify_va_arg_expr): Skip empty records. Call
4428 arg_size_in_bytes instead size_in_bytes.
4429 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
4430 * tree-core.h (tree_type_common): Add empty_flag.
4431 (tree_decl_common): Update comments.
4432 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
4434 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
4435 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
4437 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
4438 * tree.c (default_is_empty_type): New function.
4439 (default_is_empty_record): New function.
4440 (arg_int_size_in_bytes): New function.
4441 (arg_size_in_bytes): New function.
4442 (get_ultimate_context): New function.
4443 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
4444 TRANSLATION_UNIT_WARN_EMPTY_P.
4445 (default_is_empty_record, arg_int_size_in_bytes,
4446 arg_size_in_bytes, get_ultimate_context): Declare.
4448 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
4450 * config/arm/arm.c (cmse_clear_registers): New function.
4451 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
4452 code by call to cmse_clear_registers.
4453 (cmse_nonsecure_entry_clear_before_return): Likewise.
4455 2017-11-22 Tamar Christina <tamar.christina@arm.com>
4457 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
4458 (vdot_s32, vdotq_s32): New.
4459 (vdot_lane_u32, vdotq_lane_u32): New.
4460 (vdot_lane_s32, vdotq_lane_s32): New.
4463 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
4466 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
4467 for unsigned values with fewer HWIs than the precision.
4468 (test_overflow): New function.
4469 (wide_int_cc_tests): Call it.
4471 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
4473 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
4474 has the same mode class as Pmode.
4476 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
4478 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
4479 padding_bits_to_clear_ptr.
4480 (cmse_nonsecure_entry_clear_before_return): Likewise.
4482 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
4484 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
4485 auto_sbitap instead of integer bitfield to control register needing
4488 2017-11-22 Jakub Jelinek <jakub@redhat.com>
4490 PR tree-optimization/83044
4491 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
4492 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
4493 that need the upper bound. Subtract offset from
4494 get_addr_base_and_unit_offset only if positive and subtract it
4495 before division by eltsize rather than after it.
4498 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
4499 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
4500 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
4504 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
4505 expand_binop with *mul_widen_optab, make sure at least one of the
4506 operands doesn't have VOIDmode.
4509 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
4511 PR rtl-optimization/82044
4512 PR tree-optimization/82042
4513 * dse.c (record_store): Check for overflow.
4514 (check_mem_read_rtx): Properly check for overflow if width == -1, call
4515 add_wild_read instead of clear_rhs_from_active_local_stores on
4516 overflow and log it into dump_file.
4518 2017-11-22 Richard Biener <rguenther@suse.de>
4520 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
4521 fake edges to exit when looking for a place to insert.
4522 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
4524 (insert_into_preds_of_block): Commit edge insertion immediately,
4525 assert that doesn't require new BBs.
4526 (fini_pre): Release expressions.
4527 (pass_pre::execute): Shuffle things around a bit, if the fn
4528 is too large do not compute AVAIL either as this is really the
4531 2017-11-22 Richard Biener <rguenther@suse.de>
4533 PR tree-optimization/83089
4534 * tree-if-conv.c (pass_if_conversion::execute): If anything
4535 changed reset SCEV and free the number of iteration estimates.
4537 2017-11-21 Martin Sebor <msebor@redhat.com>
4539 PR tree-optimization/82945
4540 * calls.h (warn_nonstring_bound): Remove unused function.
4542 2017-11-21 Martin Sebor <msebor@redhat.com>
4544 PR tree-optimization/82945
4545 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
4546 * calls.h (maybe_warn_nonstring_arg): Declare new function.
4547 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
4549 (initialize_argument_information): Call maybe_warn_nonstring_arg.
4550 * calls.h (get_attr_nonstring_decl): Declare new function.
4551 * doc/extend.texi (attribute nonstring): Update.
4552 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
4553 get_attr_nonstring_decl and handle it.
4554 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
4555 detection of nul-termination.
4556 (strlen_to_stridx): Change to a pointer.
4557 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
4558 (pass_strlen::execute): Same.
4560 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
4562 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
4563 for the new option -mprefer-vector-width=[none|128|256|512].
4564 * config/i386/i386.c (ix86_target_string): remove old style options
4565 -mprefer-avx256 and make -mprefer-avx128 as alias.
4566 (ix86_option_override_internal): Apply defaults for the
4567 -mprefer-vector-width=[128|256] option.
4568 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
4569 Implement macros to work with -mprefer-vector-width=.
4570 * config/i386/i386.opt: Implemented option
4571 -mprefer-vector-width=[none|128|256|512].
4572 * doc/invoke.texi: Documentation for
4573 -mprefer-vector-width=[none|128|256|512].
4575 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
4577 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
4579 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
4581 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
4583 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
4585 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
4587 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
4588 * expr.c (expand_expr_real_2): Likewise.
4589 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
4590 fold_binary_loc): Likewise.
4591 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
4592 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
4593 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
4594 MINUS_EXPR transformations.
4595 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
4596 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
4597 * tree-inline.c (estimate_operator_cost): Likewise.
4598 * tree-pretty-print.c (dump_generic_node, op_code_prio,
4599 op_symbol_code): Likewise.
4600 * tree-vect-stmts.c (vectorizable_operation): Likewise.
4601 * vr-values.c (extract_range_from_binary_expr): Likewise.
4602 * varasm.c (initializer_constant_valid_p_1): Likewise.
4603 * tree.def: New tree code POINTER_DIFF_EXPR.
4605 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
4607 * config/i386/i386.md (*bswap<mode>2_movbe): Add
4608 integer suffix to movbe mnemonic.
4609 (*bswaphi2_movbe): Ditto.
4610 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
4612 2017-11-21 Jakub Jelinek <jakub@redhat.com>
4615 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
4616 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
4617 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
4619 2017-11-21 Martin Liska <mliska@suse.cz>
4621 * tree-inline.c (expand_call_inline): Remove not needed
4624 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
4625 Jakub Jelinek <jakub@redhat.com>
4628 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
4629 Remove static keyword from f variable.
4631 2017-11-21 Jakub Jelinek <jakub@redhat.com>
4633 PR tree-optimization/83086
4634 * gimple-ssa-store-merging.c
4635 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
4636 rather than n.base_addr.
4638 2017-11-21 Martin Liska <mliska@suse.cz>
4640 PR rtl-optimization/82044
4641 PR tree-optimization/82042
4642 * dse.c (check_mem_read_rtx): Check for overflow.
4644 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
4646 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
4649 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
4651 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
4652 and its corresponding call arg location note.
4654 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
4655 Andrew Burgess <andrew.burgess@embecosm.com>
4657 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
4659 (arc_return_slot_offset): Likewise.
4660 (arc_eh_return_address_location): New declaration.
4661 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
4662 (MUST_SAVE_REGISTER): Add exception handler case.
4663 (MUST_SAVE_RETURN_ADDR): Likewise.
4664 (arc_frame_pointer_required): Likewise.
4665 (arc_frame_pointer_needed): New function.
4666 (arc_compute_frame_size): Changed.
4667 (arc_expand_prologue): Likewise.
4668 (arc_expand_epilogue): Likewise.
4669 (arc_initial_elimination_offset): Likewise.
4670 (arc_return_slot_offset): Delete.
4671 (arc_eh_return_address_location): New function.
4672 (arc_builtin_setjmp_frame_value): Likewise.
4673 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
4674 (EH_RETURN_STACKADJ_RTX): Define.
4675 (EH_RETURN_HANDLER_RTX): Likewise.
4676 * config/arc/arc.md (eh_return): Delete.
4678 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
4680 * print-tree.h (debug_vec_tree): Remove prototype.
4681 * gdbinit.in (pvt): Remove macro.
4683 2017-11-21 Jakub Jelinek <jakub@redhat.com>
4685 PR tree-optimization/83047
4686 * gimple-ssa-store-merging.c
4687 (imm_store_chain_info::output_merged_store): If the loads with the
4688 same vuse are in different basic blocks, for load_gsi pick a load
4689 location that is dominated by the other loads.
4692 * config/i386/i386.c (ix86_memmodel_check): Start
4693 -Winvalid-memory-model diagnostics with lowercase letter.
4696 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
4697 set in_cold_section_p to the partition containing loc_list->first.
4698 When seeing loc_list->last_before_switch node, update secname and
4699 perform range_across_switch second partition handling only after that.
4702 * run-rtl-passes.c: Include debug.h.
4703 (run_rtl_passes): Call debug_hooks->assembly_start.
4704 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
4708 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
4709 OPTAB_DIRECT in calls to expand_simple_binop.
4711 2017-11-20 David Malcolm <dmalcolm@redhat.com>
4714 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
4716 2017-11-20 David Malcolm <dmalcolm@redhat.com>
4719 * spellcheck.h (best_match::blithely_get_best_candidate): New
4722 2017-11-20 Jakub Jelinek <jakub@redhat.com>
4724 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
4725 with lower case letter.
4727 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
4729 * config/i386/i386.md (bswaphi2): New expander.
4730 (*bswaphi2_movbe): New insn pattern.
4731 (bswaphi -> rorhi pepehole2): New peephole pattern.
4733 2017-11-20 Jeff Law <law@redhat.com>
4735 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
4736 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
4737 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
4738 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
4739 gimple-ssa-evrp-analyze.h.
4741 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
4742 set BB_VISITED here.
4743 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
4745 2017-11-20 Bin Cheng <bin.cheng@arm.com>
4747 * tree-predcom.c: Add general comment on Store-Store chains.
4748 (split_data_refs_to_components): Postpone clearing eliminate_store_p
4750 (get_chain_last_ref_at): Rename into...
4751 (get_chain_last_write_at): ...this.
4752 (get_chain_last_write_before_load): New function.
4753 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
4754 CT_STORE_STORE when write reference is added.
4755 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
4756 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
4757 (initialize_root_vars_store_elim_1): Ditto.
4758 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
4759 definition is created.
4760 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
4761 chain by replacing it with dominant stored value.
4763 2017-11-20 Bin Cheng <bin.cheng@arm.com>
4765 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
4767 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
4769 * vr-values.c (extract_range_from_binary_expr): Use a full range
4772 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
4774 * config/arm/arm.md (R4_REGNUM): Define constant.
4775 (nonsecure_call_internal): Remove r4 clobber.
4776 (nonsecure_call_value_internal): Likewise.
4777 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
4778 clobber and resequence match_operands.
4779 (nonsecure_call_value_reg_thumb1_v5): Likewise.
4780 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
4781 (nonsecure_call_value_reg_thumb2): Likewise.
4783 2017-11-20 Jakub Jelinek <jakub@redhat.com>
4785 PR tree-optimization/78821
4786 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
4787 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
4788 to the address of the base rather than the base itself.
4789 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
4790 (find_bswap_or_nop_finalize): New function.
4791 (find_bswap_or_nop): Use it.
4792 (bswap_replace): Return a tree rather than bool, change first
4793 argument from gimple * to gimple_stmt_iterator, allow inserting
4794 into an empty sequence, allow ins_stmt to be NULL - then emit
4795 all stmts into gsi. Fix up MEM_REF address gimplification.
4796 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
4797 (struct store_immediate_info): Add N and INS_STMT non-static
4799 (store_immediate_info::store_immediate_info): Initialize them
4800 from newly added ctor args.
4801 (merged_store_group::apply_stores): Formatting fixes. Sort by
4803 (stmts_may_clobber_ref_p): For stores call also
4804 refs_anti_dependent_p.
4805 (gather_bswap_load_refs): New function.
4806 (imm_store_chain_info::try_coalesce_bswap): New method.
4807 (imm_store_chain_info::coalesce_immediate_stores): Use it.
4808 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
4809 (imm_store_chain_info::output_merged_store): Fail if number of
4810 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
4811 and NOP_EXPR rhs_code.
4812 (pass_store_merging::process_store): Compute n and ins_stmt, if
4813 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
4814 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
4816 (pass_store_merging::execute): Calculate dominators.
4818 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
4819 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
4820 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
4821 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
4822 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
4823 class pass_optimize_bswap, bswap_replace,
4824 pass_optimize_bswap::execute): Moved to ...
4825 * gimple-ssa-store-merging.c: ... this file.
4826 Include optabs-tree.h.
4827 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
4828 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
4829 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
4830 anonymous namespace, remove static keywords.
4831 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
4832 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
4834 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
4837 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
4839 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
4841 * vec.h (debug_helper): New function.
4842 (DEFINE_DEBUG_VEC): New macro.
4843 * hash-set.h (debug_helper): New function.
4844 (DEFINE_DEBUG_HASH_SET): New macro.
4845 * cfg.c (debug_slim (edge)): New function.
4846 Call DEFINE_DEBUG_VEC for edges.
4847 Call DEFINE_DEBUG_HASH_SET for edges.
4848 * cfghooks.c (debug_slim (basic_block)): New function.
4849 Call DEFINE_DEBUG_VEC for basic blocks.
4850 Call DEFINE_DEBUG_HASH_SET for basic blocks.
4851 * print-tree.c (debug_slim): New function to handle trees.
4852 Call DEFINE_DEBUG_VEC for trees.
4853 Call DEFINE_DEBUG_HASH_SET for trees.
4854 (debug (vec<tree, va_gc>) &): Remove.
4855 (debug (<vec<tree, va_gc>) *): Remove.
4856 * print-rtl.c (debug_slim): New function to handle const_rtx.
4857 Call DEFINE_DEBUG_VEC for rtx_def.
4858 Call DEFINE_DEBUG_VEC for rtx_insn.
4859 Call DEFINE_DEBUG_HASH_SET for rtx_def.
4860 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
4861 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
4862 (debug (vec<rtx_insn *> *ptr): Remove.
4863 (debug_insn_vector): Remove.
4864 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
4866 2017-11-20 Tom de Vries <tom@codesourcery.com>
4868 PR rtl-optimization/82020
4869 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
4870 IF_THEN_ELSE condition.
4872 2017-11-19 Jeff Law <law@redhat.com>
4874 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
4875 of degenerates resulting from ignoring an edge.
4877 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
4880 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
4882 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
4885 * profile-count.c (profile_count::to_sreal_scale): Fix return value
4886 for uninitialied counts.
4888 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
4891 * tree-inline.c (estimate_num_insns): Set to 1 at least.
4893 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
4896 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
4898 2017-11-19 Tom de Vries <tom@codesourcery.com>
4900 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
4903 2017-11-19 Tom de Vries <tom@codesourcery.com>
4905 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
4908 2017-11-19 Tom de Vries <tom@codesourcery.com>
4910 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
4913 2017-11-19 Tom de Vries <tom@codesourcery.com>
4915 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
4917 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
4919 2017-11-19 Tom de Vries <tom@codesourcery.com>
4921 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
4923 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
4924 semicolon after MCORE_EXPORT_NAME call.
4926 2017-11-19 Tom de Vries <tom@codesourcery.com>
4929 * vmsdbgout.c (vmsdbgout_early_finish): New function.
4930 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
4932 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
4934 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
4935 (cgraph_edge::clone): Cleanup updating of profile.
4936 * ipa-cp.c (update_profiling_info): Likewise.
4937 * ipa-inline-transform.c (inline_transform): Likewise.
4938 * ipa-inline.c (inline_small_functions): Add missing space to dump.
4939 * ipa-split.c (execute_split_functions): Do not split when function
4941 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
4942 * profile-count.c (profile_count::dump): Add global0.
4943 (profile_count::to_cgraph_frequency): Do not ICE when entry is
4945 (profile_count::to_sreal_scale): Likewise.
4946 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
4947 (profile_count::combine_with_ipa_count): New function.
4948 * profile-count.h (profile_guessed_global0adjusted): New.
4949 (profile_count::adjusted_zero): New.
4950 (profile_count::global0adjusted): New.
4951 (profile_count::combine_with_ipa_count): New.
4952 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
4953 correct profile of return block of split functions.
4954 (copy_cfg_body): Remove unused profile_count.
4955 (copy_body): Likewise.
4956 (expand_call_inline): Update.
4957 (tree_function_versioning): Update.
4959 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
4961 * hash-set.h (hash_set::empty): New.
4962 * tree-ssa-threadbackward.h: Delete.
4963 * tree-ssa-threadbackward.c (class thread_jumps): New.
4964 Move max_threaded_paths into class.
4965 (fsm_find_thread_path): Remove arguments that are now in class.
4966 (profitable_jump_thread_path): Rename to...
4967 (thread_jumps::profitable_jump_thread_path): ...this.
4968 (convert_and_register_jump_thread_path): Rename to...
4969 (thread_jumps::convert_and_register_current_path): ...this.
4970 (check_subpath_and_update_thread_path): Rename to...
4971 (thread_jumps::check_subpath_and_update_thread_path): ...this.
4972 (register_jump_thread_path_if_profitable): Rename to...
4973 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
4974 (handle_phi): Rename to...
4975 (thread_jumps::handle_phi): ...this.
4976 (handle_assignment): Rename to...
4977 (thread_jumps::handle_assignment): ...this.
4978 (fsm_find_control_statement_thread_paths): Rename to...
4979 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
4980 (find_jump_threads_backwards): Rename to...
4981 (thread_jumps::find_jump_threads_backwards): ...this.
4982 Initialize path local data.
4983 (pass_thread_jumps::execute): Call find_jump_threads_backwards
4984 from within thread_jumps class.
4985 (pass_early_thread_jumps::execute): Same.
4987 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
4989 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
4991 2017-11-17 Steve Ellcey <sellcey@cavium.com>
4993 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
4994 to canonical location.
4996 2017-11-17 Steve Ellcey <sellcey@cavium.com>
4999 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
5001 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
5003 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
5005 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
5006 rcrt1.o%s/grcrt1.o%s for -static-pie.
5008 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5010 * i386.c (ix86_multiplication_cost, ix86_division_cost,
5011 ix86_shift_rotate_cost): Break out from ...
5012 (ix86_rtx_costs): ... here.
5013 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
5016 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5018 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
5019 when entry block was promoted unlikely.
5020 (estimate_bb_frequencies): Increase frequency scale.
5021 * profile-count.h (profile_count): Export precision info.
5023 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5025 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
5026 disturbin profile of entry block.
5028 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5030 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
5033 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5035 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
5036 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
5038 2017-11-17 Jeff Law <law@redhat.com>
5040 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
5041 from evrp_dom_walker class. Various methods moved into new class.
5042 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
5043 (evrp_range_analyzer::enter): New method.
5044 (evrp_range_analyzer::leave): New method.
5045 (evrp_dom_walker): Remove delegators no longer needed by this class.
5046 Replace vr_values data member with evrp_range_analyzer
5048 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
5049 method extracted from evrp_dom_walker::before_dom_children.
5050 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
5051 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
5053 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
5054 Add private copy constructor and move assignment operators.
5055 Privatize methods and class data where trivially possible.
5056 (evrp_dom_walker::cleanup): New function, extracted from
5057 execute_early_vrp. Simplify access to class data.
5059 * vr-values.h (get_output_for_vrp): Prototype.
5060 * vr-values.c (get_output_for_vrp): New function extracted from
5061 vrp_visit_assignment_or_call and extract_range_from_stmt.
5062 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
5064 2017-11-17 Luis Machado <luis.machado@linaro.org>
5066 * config/aarch64/aarch64.c
5067 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
5068 (qdf24xx_tunings) <autoprefetcher_model>: Set to
5069 tune_params::AUTOPREFETCHER_WEAK.
5071 2017-11-17 Tamar Christina <tamar.christina@arm.com>
5074 * config/arm/arm.c (arm_valid_target_attribute_rec):
5075 Parse "arch=" and "+<ext>".
5076 (arm_valid_target_attribute_tree): Re-init global options.
5077 (arm_option_override): Make non-static.
5078 (arm_options_perform_arch_sanity_checks): Make errors fatal.
5079 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
5080 (__ARM_FEATURE_CRC32): Support undef.
5081 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
5082 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
5084 2017-11-17 David Malcolm <dmalcolm@redhat.com>
5086 * gdbinit.in (break-on-diagnostic): New command.
5088 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5090 * config/i386/i386.c (ix86_expand_epilogue): Change simple
5091 return to indirect jump for EH return if control-flow
5092 protection is enabled. Change explicit 'false' argument in
5093 pro_epilogue_adjust_stack with a value of
5095 * config/i386/i386.md (simple_return_indirect_internal):
5096 Remove SImode restriction to support 64-bit.
5098 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
5100 * combine.c (added_notes_insn): New.
5101 (try_combine): Handle added_notes_insn like added_links_insn.
5102 Rewrite return value code.
5103 (distribute_notes): Set added_notes_insn to the earliest insn we added
5106 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
5108 PR rtl-optimization/82621
5109 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
5110 dest of one of those SETs is unused.
5112 2017-11-17 Richard Biener <rguenther@suse.de>
5115 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
5116 * tree-pretty-print.c (dump_generic_node): Handle
5117 annot_expr_parallel_kind.
5118 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
5119 * gimplify.c (gimple_boolify): Likewise.
5121 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5123 * config.gcc (extra_headers): Add cet.h for x86 targets.
5124 * config/i386/cet.h: New file.
5125 * doc/install.texi: Add --enable-cet/--disable-cet.
5127 2017-11-17 Richard Biener <rguenther@suse.de>
5129 PR tree-optimization/83017
5130 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
5131 (gen_parallel_loop): Properly count iterations.
5132 (parallelize_loops): Handle loop->can_be_parallel independent
5133 of flag_loop_parallelize_all. Make static profitability test match
5135 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
5136 * invoke.texi (parloops-min-per-thread): Document.
5138 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
5140 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
5141 upstreaming review comments.
5143 2017-11-17 Tamar Christina <tamar.christina@arm.com>
5145 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
5146 with fast unaligned access.
5147 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
5149 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
5151 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
5152 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
5154 2017-11-17 Richard Biener <rguenther@suse.de>
5156 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
5157 folding of references.
5159 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
5162 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
5163 of replacing call to strncmp with corresponding call to strcmp when
5166 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
5168 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
5169 option prefer-avx256 for skylake-avx512 configuration.
5170 * config/i386/i386.c (ix86_option_override_internal): Ditto.
5171 (get_builtin_code_for_version): Ditto.
5173 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
5174 Monk Chiang <sh.chiang04@gmail.com>
5176 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
5177 (FIXED_REGISTERS): Reserve more register numbers.
5178 (CALL_USED_REGISTERS): Likewise.
5179 (REG_ALLOC_ORDER): Likewise.
5180 (REG_CLASS_CONTENTS): Likewise.
5181 (REGISTER_NAMES): Likewise.
5183 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
5184 Kito Cheng <kito.cheng@gmail.com>
5186 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
5188 * config/nds32/iterators.md: Add vector mode iterators and attributes.
5190 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
5192 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
5193 parameter list for vec_splats.
5195 2017-11-16 Joseph Myers <joseph@codesourcery.com>
5197 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
5199 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
5201 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
5204 * doc/install.texi: Document incompatibility of Perl >=5.6.26
5205 with the required version of automake 1.11.6.
5207 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
5209 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
5211 (DU_C2_3_power9): Correct reservation combinations.
5212 (FP_DIV_power9, VEC_DIV_power9): New.
5213 (power9-alu): Split out rotate/shift...
5214 (power9-rot): ...to here, correct dispatch resource.
5215 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
5217 (power9-fp): Correct latency.
5218 (power9-sdiv): Add div/sqrt resource.
5219 (power9-ddiv): Correct latency, add div/sqrt resource.
5220 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
5221 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
5223 (power9-qpdiv, power9-qpmul): Adjust resource usage.
5225 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
5227 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
5228 switch statement mapping KF built-ins to TF built-ins if we don't
5229 have the proper ISA 3.0 assembler support.
5231 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5233 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
5234 (lower_emutls_function_body): Do not compute it.
5236 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5238 * ipa-split.c (split_bb_info): Turn time to sreal.
5239 (split_point): Likewise.
5240 (dump_split_point): Likewise.
5241 (fine_split_points): Likewise.
5242 (execute_split_functions): Only zero split_bbs; turn time to sreals.
5244 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5246 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
5249 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5251 * predict.c (combine_predictions_for_bb): Preserve zero predicted
5253 (expensive_function_p): Remove useless assert.
5254 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
5256 2017-11-16 Martin Sebor <msebor@redhat.com>
5258 PR tree-optimization/82588
5259 PR tree-optimization/82583
5260 * tree-vrp.c (check_array_ref): Handle flexible array members,
5261 string literals, and inner indices.
5262 (search_for_addr_array): Add detail to diagnostics.
5264 2017-11-16 Nathan Sidwell <nathan@acm.org>
5268 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
5269 (SET_DECL_ASSEMBLER_NAME): Forward to
5270 overwrite_decl_assembler_name.
5271 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
5272 (overwrite_decl_assembler_name): Declare.
5273 * tree.c (overwrite_decl_assembler_name): New.
5274 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
5275 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
5276 (LANG_HOOKS_INITIALIZER): Add it.
5277 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
5278 * langhooks.c (lhd_set_decl_assembler_name): Use
5279 SET_DECL_ASSEMBLER_NAME.
5280 (lhd_overwrite_decl_assembler_name): Default implementation.
5282 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
5283 Jackson Woodruff <jackson.woodruff@arm.com>
5285 PR tree-optimization/71026
5286 * match.pd: Canonicalize constant multiplies in division.
5288 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
5290 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
5291 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
5292 Remove OPT_fomit_frame_pointer entry.
5293 * common/config/arc/arc-common.c: Likewise.
5294 * common/config/arm/arm-common.c: Likewise.
5295 * common/config/avr/avr-common.c: Likewise.
5296 * common/config/c6x/c6x-common.c: Likewise.
5297 * common/config/cr16/cr16-common.c: Likewise.
5298 * common/config/cris/cris-common.c: Likewise.
5299 * common/config/epiphany/epiphany-common.c: Likewise.
5300 * common/config/fr30/fr30-common.c: Likewise.
5301 * common/config/frv/frv-common.c: Likewise.
5302 * common/config/ia64/ia64-common.c: Likewise.
5303 * common/config/iq2000/iq2000-common.c: Likewise.
5304 * common/config/lm32/lm32-common.c: Likewise.
5305 * common/config/m32r/m32r-common.c: Likewise.
5306 * common/config/mcore/mcore-common.c: Likewise.
5307 * common/config/microblaze/microblaze-common.c: Likewise.
5308 * common/config/mips/mips-common.c: Likewise.
5309 * common/config/mmix/mmix-common.c: Likewise.
5310 * common/config/mn10300/mn10300-common.c: Likewise.
5311 * common/config/nios2/nios2-common.c: Likewise.
5312 * common/config/pa/pa-common.c: Likewise.
5313 * common/config/pdp11/pdp11-common.c: Likewise.
5314 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
5315 * common/config/riscv/riscv-common.c: Likewise.
5316 * common/config/rs6000/rs6000-common.c: Likewise.
5317 * common/config/rx/rx-common.c: Likewise.
5318 * common/config/s390/s390-common.c: Likewise.
5319 * common/config/sh/sh-common.c: Likewise.
5320 * common/config/sparc/sparc-common.c: Likewise.
5321 * common/config/tilegx/tilegx-common.c: Likewise.
5322 * common/config/tilepro/tilepro-common.c: Likewise.
5323 * common/config/v850/v850-common.c: Likewise.
5324 * common/config/visium/visium-common.c: Likewise.
5325 * common/config/xstormy16/xstormy16-common.c: Likewise.
5326 * common/config/xtensa/xtensa-common.c: Likewise.
5327 * invoke.texi (-fomit-frame-pointer): Update documentation.
5329 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5331 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
5333 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5335 * cfg.c (scale_bbs_frequencies_int,
5336 cale_bbs_frequencies_gcov_type): Remove.
5337 * cfg.h (scale_bbs_frequencies_int,
5338 cale_bbs_frequencies_gcov_type): Remove.
5340 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5342 * tree-ssa-loop-manip.c
5343 (scale_dominated_blocks_in_loop): Update to profile counts.
5344 (tree_transform_and_unroll_loop): Likewise.
5346 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5348 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
5349 scale_bbs_frequencies_int.
5351 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5353 * final.c (compute_alignments): Use counts rather than frequencies.
5355 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5357 * cfgloopanal.c: Include sreal.h
5358 (average_num_loop_insns): Use counts and sreal for accounting.
5360 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5362 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
5365 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
5366 Kito Cheng <kito.cheng@gmail.com>
5368 * config/nds32/constraints.md: Provide more constraints.
5369 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
5370 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
5371 support constraints usage.
5373 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
5375 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
5377 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
5378 Kito Cheng <kito.cheng@gmail.com>
5380 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
5381 * config/nds32/nds32.opt: Refine the layout.
5382 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
5383 TARGET_EXT_STRING): Support new options.
5384 * config/nds32/nds32.h: Likewise.
5385 * config/nds32/nds32.md: Likewise.
5386 * config/nds32/nds32-predicates.c: Likewise.
5387 * config/nds32/constraints.md: Likewise.
5388 * common/config/nds32/nds32-common.c: Likewise.
5390 2017-11-16 Julia Koval <julia.koval@intel.com>
5393 * config/i386/gfniintrin.h: Add sse check.
5394 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
5396 2017-11-16 Julia Koval <julia.koval@intel.com>
5398 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
5399 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
5400 (ix86_handle_option): Handle -mavx512vbmi2.
5401 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
5402 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
5403 * config/i386/i386-c.c (__AVX512VBMI2__): New.
5404 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
5405 (ix86_valid_target_attribute_inner_p): Ditto.
5406 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
5407 * config/i386/i386.opt (mavx512vbmi2): New option.
5408 * doc/invoke.texi: Add new option.
5410 2017-11-16 Julia Koval <julia.koval@intel.com>
5412 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
5413 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
5414 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
5415 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
5416 _mm512_gf2p8mul_epi8): New intrinsics.
5417 * config/i386/i386-builtin-types.def
5418 (V64QI_FTYPE_V64QI_V64QI): New type.
5419 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
5420 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
5421 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
5422 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
5423 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
5424 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
5426 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
5428 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
5429 explicitly as a stream of bytes.
5431 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5433 * config/rs6000/altivec.h (vec_xst_be): New #define.
5434 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
5435 and externalize from *altivec_vperm_<mode>_internal.
5436 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
5438 (XL_BE_V8HI): Likewise.
5439 (XL_BE_V4SI): Likewise.
5440 (XL_BE_V4SI): Likewise.
5441 (XL_BE_V2DI): Likewise.
5442 (XL_BE_V4SF): Likewise.
5443 (XL_BE_V2DF): Likewise.
5444 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
5445 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
5446 all array entries with these keys: VSX_BUILTIN_VEC_XL,
5447 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
5448 VSX_BUILTIN_VEC_XST_BE.
5449 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
5450 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
5452 (altivec_init_builtins): Replace conditional calls to def_builtin
5453 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
5454 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
5455 with unconditional calls. Remove calls to def_builtin for
5456 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
5457 __builtin_vec_xst_be.
5458 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
5459 to define_expand, and add alternate RTL generation for P8.
5460 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
5461 vsx_ld_elemrev_v8hi.
5462 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
5463 add alternate RTL generation for P8.
5464 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
5465 vsx_ld_elemrev_v16qi.
5466 (vsx_st_elemrev_v8hi): Convert define_insn
5467 to define_expand, and add alternate RTL generation for P8.
5468 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
5469 vsx_st_elemrev_v8hi.
5470 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
5471 add alternate RTL generation for P8.
5472 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
5473 vsx_st_elemrev_v16qi.
5475 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
5478 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
5479 TARGET_AVX512ER check.
5480 (ix86_option_override_internal): Set MASK_VZEROUPPER if
5481 neither -mzeroupper nor -mno-zeroupper is used and
5482 TARGET_EMIT_VZEROUPPER is set.
5483 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
5484 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
5486 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
5488 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
5489 folding of vector compares.
5490 (fold_build_vec_cmp): New helper function.
5491 (fold_compare_helper): New helper function.
5492 (builtin_function_type): Add compare builtins to the list of functions
5493 having unsigned arguments. Cosmetic updates to comment indentation.
5494 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
5495 the not+eq combination.
5497 2017-11-15 Bin Cheng <bin.cheng@arm.com>
5499 PR tree-optimization/82726
5500 PR tree-optimization/70754
5501 * tree-predcom.c (order_drefs_by_pos): New function.
5502 (combine_chains): Move code setting has_max_use_after to...
5503 (try_combine_chains): ...here. New parameter. Sort combined chains
5504 according to position information.
5505 (tree_predictive_commoning_loop): Update call to above function.
5506 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
5508 2017-11-15 Bin Cheng <bin.cheng@arm.com>
5510 PR tree-optimization/82726
5512 2017-01-23 Bin Cheng <bin.cheng@arm.com>
5514 PR tree-optimization/70754
5515 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
5516 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
5517 combined stmt before it if not NULL.
5518 (combine_chains): Process refs reversely and compute dominance point
5522 2017-02-23 Bin Cheng <bin.cheng@arm.com>
5524 PR tree-optimization/79663
5525 * tree-predcom.c (combine_chains): Process refs in reverse order
5526 only for ZERO length chains, and add explaining comment.
5528 2017-11-15 Tamar Christina <tamar.christina@arm.com>
5530 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
5531 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
5532 * doc/invoke.texi (ARM Options): Add armv8.3-a.
5534 2017-11-15 Tamar Christina <tamar.christina@arm.com>
5536 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
5538 2017-11-15 Martin Liska <mliska@suse.cz>
5540 * tree-cfg.c (pass_warn_function_return::execute):
5541 Compare warn_return_type for greater than zero.
5543 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
5547 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
5548 to return true on Xeon and not on Xeon Phi.
5549 (ix86_check_avx256_register): Changed to ...
5550 (ix86_check_avx_upper_register): ... this. Add extra check for
5551 VALID_AVX512F_REG_OR_XI_MODE.
5552 (ix86_avx_u128_mode_needed): Changed
5553 ix86_check_avx256_register to ix86_check_avx_upper_register.
5554 (ix86_check_avx256_stores): Changed to ...
5555 (ix86_check_avx_upper_stores): ... this. Changed
5556 ix86_check_avx256_register to ix86_check_avx_upper_register.
5557 (ix86_avx_u128_mode_after): Changed
5558 avx_reg256_found to avx_upper_reg_found. Changed
5559 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
5560 (ix86_avx_u128_mode_entry): Changed
5561 ix86_check_avx256_register to ix86_check_avx_upper_register.
5562 (ix86_avx_u128_mode_exit): Ditto.
5563 * config/i386/i386.h: (host_detect_local_cpu): New define.
5565 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
5567 * config/arm/xgene1.md (xgene1): Split into automatons
5568 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
5569 (xgene1_f_load): Adjust reservations and/or types.
5570 (xgene1_f_store): Likewise.
5571 (xgene1_load_pair): Likewise.
5572 (xgene1_store_pair): Likewise.
5573 (xgene1_fp_load1): Likewise.
5574 (xgene1_load1): Likewise.
5575 (xgene1_store1): Likewise.
5576 (xgene1_move): Likewise.
5577 (xgene1_alu): Likewise.
5578 (xgene1_simd): Likewise.
5579 (xgene1_bfm): Likewise.
5580 (xgene1_neon_load1): Likewise.
5581 (xgene1_neon_store1): Likewise.
5582 (xgene1_neon_logic): Likewise.
5583 (xgene1_neon_st1): Likewise.
5584 (xgene1_neon_ld1r): Likewise.
5585 (xgene1_alu_cond): Added.
5586 (xgene1_shift_reg): Likwise.
5587 (xgene1_bfx): Likewise.
5588 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
5590 2017-11-15 Jakub Jelinek <jakub@redhat.com>
5593 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
5595 (can_widen_mult_without_libcall): New function.
5596 (expand_mul_overflow): If only checking unsigned mul overflow,
5597 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
5598 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
5599 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
5600 (expand_DIVMOD): Formatting fix.
5601 * expmed.h (expand_mult): Add NO_LIBCALL argument.
5602 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
5603 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
5605 PR tree-optimization/82977
5606 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
5607 constructed temporary to strlen_to_stridx.put.
5609 2017-11-15 Martin Liska <mliska@suse.cz>
5611 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
5612 * configure: Regenerate.
5614 2017-11-15 Martin Liska <mliska@suse.cz>
5617 * config/sh/sh-mem.cc: Use proper probability for
5620 2017-11-14 Jeff Law <law@redhat.com>
5622 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
5623 the red zone for stack_clash_protection_final_dynamic_probe targets
5624 when the total dynamic stack size is zero bytes.
5626 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
5627 blocks is post order.
5629 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
5631 * dumpfile.h (TDF_COMPARE_DEBUG): New.
5632 * final.c (rest_of_clean_state): Set it for the
5633 -fcompare-debug dump.
5634 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
5635 class when TDF_COMPARE_DEBUG is set.
5637 * dwarf2out.c (gen_producer_string): Discard
5640 2017-11-15 Joseph Myers <joseph@codesourcery.com>
5643 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
5644 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
5645 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
5646 (__TGMATH_REAL_2_3): Remove macros.
5647 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
5648 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
5650 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
5652 (remquo): Define using __TGMATH_REAL_3.
5654 2017-11-14 Jeff Law <law@redhat.com>
5656 * vr-values.c: New file with contents extracted from tree-vrp.c.
5657 * Makefile.in (OBJS): Add vr-values.o
5658 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
5659 (set_value_range, set_and_canonicalize_value_range): Likewise.
5660 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
5661 (value_range_constant_singleton, symbolic_range_p): Likewise.
5662 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
5663 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
5664 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
5665 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
5666 (find_case_label_range, find_case_label_index): Likewise.
5667 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
5668 (range_int_cst_singleton_p, value_inside_range): Likewise.
5669 (get_single_symbol): Likewise.
5670 (switch_update): Move structure definition here.
5671 (to_remove_edges, to_update_switch_stmts): Provide externs.
5672 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
5673 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
5674 (vrp_val_is_min, set_value_range): Likewise.
5675 (set_and_canonicalize_value_range, copy_value_range): Likewise.
5676 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
5677 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
5678 (range_is_nonnull, range_int_cst_p): Likewwise.
5679 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
5680 (get_single_symbol, operand_less_p): Likewise
5681 (compare_values_warnv, compare_values): Likewise.
5682 (value_inside_range, value_range_constant_singleton): Likewise.
5683 (zero_nonzero_bitgs_from_vr): Likewise.
5684 (extract_range_from_binary_expr_1): Likewise.
5685 (overflow_comparison_p): Likewise.
5686 (to_remove_edges, to_update_switch_stmts): Likewise.
5687 (find_case_label-index, find_case_label_range): Likewise.
5688 (switch_update, set_value_range_to_nonnegative): Remove.
5689 (set_value_range_to_truthvalue): Likewise.
5690 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
5691 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
5692 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
5693 (find_case_label_ranges, test_for_singularity): Likewise.
5694 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
5695 (x_vr_values): Move to its remaining use site.
5697 2017-11-10 Jeff Law <law@redhat.com>
5699 * vr-values.h (VR_INITIALIZER): Move #define here.
5700 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
5701 * Makefile.in (OBJS): Add tree-evrp.o
5702 * tree-vrp.h (assert_info): Move structure definition here.
5703 (set_value_range_to_varying): Prototype.
5704 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
5705 (infer_value_range, register_edge_assert_for): Likewise.
5706 (stmt_interesting_for_vrp): Likewise.
5707 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
5708 (set_value_range_to_varying): No longer static.
5709 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
5710 (infer_value_range, register_edge_assert_for): Likewise.
5712 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
5714 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
5715 generate the XXBRD instruction.
5717 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
5718 (rs6000_builtin_type_compatible): Treat _Float128 and long double
5719 as being compatible if -mabi=ieeelongdouble.
5720 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
5721 to setup float128 built-ins with hardware support.
5722 (BU_FLOAT128_HW_2): Likewise.
5723 (BU_FLOAT128_HW_3): Likewise.
5724 (BU_FLOAT128_HW_VSX_1): Likewise.
5725 (BU_FLOAT128_HW_VSX_2): Likewise.
5726 (scalar_extract_expq): Change float128 built-in functions to
5727 accommodate having both KFmode and TFmode functions. Use the
5728 KFmode variant as the default.
5729 (scalar_extract_sigq): Likewise.
5730 (scalar_test_neg_qp): Likewise.
5731 (scalar_insert_exp_q): Likewise.
5732 (scalar_insert_exp_qp): Likewise.
5733 (scalar_test_data_class_qp): Likewise.
5734 (sqrtf128_round_to_odd): Delete processing the round to odd
5735 built-in functions as special built-in functions, and define them
5736 as float128 built-ins. Use the KFmode variant as the default.
5737 (truncf128_round_to_odd): Likewise.
5738 (addf128_round_to_odd): Likewise.
5739 (subf128_round_to_odd): Likewise.
5740 (mulf128_round_to_odd): Likewise.
5741 (divf128_round_to_odd): Likewise.
5742 (fmaf128_round_to_odd): Likewise.
5743 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
5744 support for KFmode and TFmode xststdcqp calls.
5745 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
5746 point, switch the built-in handlers for the get/set float128
5747 exponent, get float128 mantissa, float128 test built-ins, and the
5748 float128 round to odd built-in functions. Eliminate creating the
5749 float128 round to odd built-in functions as special built-ins.
5750 (rs6000_init_builtins): Eliminate special creation of the float128
5751 round to odd built-in functions.
5752 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
5753 function insns to support both TFmode and KFmode variants.
5754 (xsxsigqp_<mode>): Likewise.
5755 (xsiexpqpf_<mode>): Likewise.
5756 (xsiexpqp_<mode>): Likewise.
5757 (xststdcqp_<mode>): Likewise.
5758 (xststdcnegqp_<mode>): Likewise.
5759 (xststdcqp_<mode>): Likewise.
5761 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5763 * tree-ssa-threadupdate.c (compute_path_counts): Remove
5764 unused path_in_freq_ptr parameter.
5765 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
5767 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5769 * ipa-inline.c (edge_badness): Dump sreal frequency.
5770 (compute_inlined_call_time): Match natural implementaiton ...
5771 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
5772 forgotten division by CGRAPH_FREQ_BASE.
5774 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5776 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
5777 Solaris 11. Update comment.
5778 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
5780 * config/sol2.h (STARTFILE_SPEC): Likewise.
5781 * configure: Regenerate.
5783 2017-11-14 Carl Love <cel@us.ibm.com>
5785 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
5786 le_ and be_ prefixes to swap* variables. Remove
5787 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
5790 2017-11-14 Jason Merrill <jason@redhat.com>
5792 Support GTY((cache)) on hash_map.
5793 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
5794 (ggc_cache_remove): Override it instead of ggc_mx.
5795 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
5796 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
5797 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
5798 (simple_cache_map_traits): Override maybe_mx.
5799 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
5800 (hash_map): Friend gt_cleare_cache.
5801 (gt_cleare_cache): New.
5802 * tree.h (tree_cache_traits): New hash_map traits class.
5803 (tree_cache_map): New typedef.
5805 2017-11-14 Richard Biener <rguenther@suse.de>
5807 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
5808 paramter and handling.
5809 (cleanup_control_flow_bb): Likewise.
5810 (cleanup_control_flow_pre): New helper performing a DFS walk
5811 to call cleanup_control_flow_bb in PRE order.
5812 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
5813 via cleanup_control_flow_pre.
5815 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
5817 * config/aarch64/aarch64-simd.md
5818 (aarch64_simd_bsl<mode>_internal): Remove DImode.
5819 (*aarch64_simd_bsl<mode>_alt): Likewise.
5820 (aarch64_simd_bsldi_internal): New.
5821 (aarch64_simd_bsldi_alt): Likewise.
5823 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5825 * tracer.c (better_p): Do not compare frequencies.
5826 * reg-stack.c (better_edge): Likewise.
5827 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
5830 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5832 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
5833 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
5834 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
5835 * lto-streamer-in.c (input_function): Use update_max_bb_count.
5836 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
5837 * predict.c (maybe_hot_frequency_p): Inline to ...
5838 (maybe_hot_count_p): ... here; rewrite to counts.
5839 (counts_to_freqs): Rename to ...
5840 (update_max_bb_count): ... this one.
5841 (expensive_function_p): Use counts.
5842 (estimate_bb_frequencies): Update.
5843 (rebuild_frequencies): Update.
5844 * predict.h (counts_to_freqs): Rename to ...
5845 (update_max_bb_count): ... this one.
5846 * profile.c (compute_branch_probabilities): Add debug info
5847 * tree-inline.c (expand_call_inline): Update debug info.
5848 (optimize_inline_calls): Use update_max_bb_count..
5849 (tree_function_versioning): Use update_max_bb_count..
5850 * value-prof.c (gimple_value_profile_transformations):
5851 Do not use update_max_bb_count.
5853 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5855 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
5856 always use frequencies.
5858 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5860 * bb-reorder.c: Remove frequencies from comments.
5861 (better_edge_p): Use profile counts.
5862 (find_traces): Dump profile counts.
5863 (rotate_loop): Use profile counts.
5864 (find_traces_1_round): Likewise.
5865 (connect_better_edge_p): Use counts instead of probabilities for
5867 (copy_bb_p): Drop early check for non-0 frequency.
5868 (sanitize_hot_paths): Update comments.
5870 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5872 * ipa-split.c (struct split_point): Add count.
5873 (consider_split): Do not compute incoming frequency; compute incoming
5874 count and store it to split_point.
5875 (split_function): Set count of the call to split part correctly.
5877 2017-11-13 Carl Love <cel@us.ibm.com>
5879 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
5881 2017-11-13 Tom Tromey <tom@tromey.com>
5883 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
5885 2017-11-13 Carl Love <cel@us.ibm.com>
5887 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
5888 Add support for builtins:
5889 unsigned int vec_first_{,miss}_match_{,or_eos}index,
5890 vector {un,}signed {char,int,short},
5891 vector {un,}signed {char,int,short}) arguments.
5892 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
5893 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
5894 Add BU_P9V_AV_2 expansions for the builtins.
5895 * config/rs6000/altivec.h (vec_first_match_index,
5896 vec_first_mismatch_index, vec_first_match_or_eos_index,
5897 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
5898 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
5899 new extern declaration.
5900 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
5901 * config/rs6000/vsx.md (first_match_index_<mode>,
5902 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
5903 first_mismatch_or_eos_index_<mode>): Add define expand.
5904 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
5905 * doc/extend.texi: Update the built-in documenation file for the new
5908 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
5910 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
5911 into the min/max operations for _Float<N> and _Float<N>X types.
5913 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
5916 * dwarf2out.c (do_eh_frame): New static variable.
5917 (dwarf2out_begin_prologue): Set it.
5918 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
5920 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5922 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
5924 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
5926 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
5929 * debug.h (dwarf2out_do_eh_frame): Declare.
5930 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
5931 (dwarf2out_do_frame): Use it.
5932 (dwarf2out_do_cfi_asm): Likewise.
5933 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
5934 (dwarf2out_assembly_start): Likewise.
5935 (dwarf2out_begin_prologue): Fix comment.
5936 * toplev.c (compile_file): Always call dwarf2out_frame_finish
5937 if the target needs either debug or unwind DWARF2 info.
5938 * lto-opts.c (lto_write_options): Do not save -fexceptions,
5939 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
5940 -ftrapping-math, -ftrapv and -fwrapv.
5942 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5944 * cgraph.c (cgraph_edge::sreal_frequency): New function.
5945 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
5946 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
5947 (estimate_edge_size_and_time): Likewise.
5948 (ipa_merge_fn_summary_after_inlining): Likewise.
5949 * ipa-inline.c (cgraph_freq_base_rec): Remove.
5950 (compute_uninlined_call_time): Use sreal_frequency.
5951 (compute_inlined_call_time): Likewise.
5952 (ipa_inline): Do not initialize cgraph_freq_base_rec.
5953 * profile-count.c: Include sreal.h.
5954 (profile_count::to_sreal_scale): New.
5955 * profile-count.h: Forward declare sreal.
5956 (profile_count::to_sreal_scale): Declare.
5958 2017-11-13 Nathan Sidwell <nathan@acm.org>
5960 * diagnostic.c (maybe_line_and_column): New.
5961 (diagnostic_get_location_text): Use it.
5962 (diagnostic_report_current_module): Likewise.
5963 (test_diagnostic_get_location_text): Add tests.
5965 2017-11-13 Luis Machado <luis.machado@linaro.org>
5967 * doc/md.texi (Specifying processor pipeline description): Fix
5968 incorrect latency for the div instruction example.
5970 2017-11-13 Jakub Jelinek <jakub@redhat.com>
5972 PR tree-optimization/78821
5973 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
5974 that bit_not_p is the same.
5975 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
5976 (split_group): Count precisely bit_not_p bits in each statement.
5977 (invert_op): New function.
5978 (imm_store_chain_info::output_merged_store): Use invert_op to
5979 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
5980 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
5982 2017-11-13 Martin Liska <mliska@suse.cz>
5984 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
5985 (struct source_info): Likewise.
5986 (add_branch_counts): Likewise.
5987 (add_line_counts): Likewise.
5988 (function_summary): Likewise.
5989 (output_intermediate_line): Likewise.
5990 (generate_results): Likewise.
5992 2017-11-13 Martin Liska <mliska@suse.cz>
5994 * gcov.c (struct block_info): Remove typedef for block_t.
5995 (struct line_info): Likewise.
5996 (line_info::has_block): Likewise.
5997 (EXIT_BLOCK): Likewise.
5998 (unblock): Likewise.
5999 (circuit): Likewise.
6000 (get_cycles_count): Likewise.
6001 (process_file): Likewise.
6002 (read_graph_file): Likewise.
6003 (solve_flow_graph): Likewise.
6004 (find_exception_blocks): Likewise.
6005 (add_line_counts): Likewise.
6006 (accumulate_line_info): Likewise.
6007 (output_line_details): Likewise.
6009 2017-11-13 Martin Liska <mliska@suse.cz>
6011 * gcov.c (struct arc_info): Remove typedef for arc_t.
6012 (struct line_info): Likewise.
6013 (add_branch_counts): Likewise.
6014 (output_branch_count): Likewise.
6015 (function_info::~function_info): Likewise.
6016 (circuit): Likewise.
6017 (output_intermediate_line): Likewise.
6018 (read_graph_file): Likewise.
6019 (solve_flow_graph): Likewise.
6020 (find_exception_blocks): Likewise.
6021 (add_line_counts): Likewise.
6022 (accumulate_line_info): Likewise.
6023 (output_line_details): Likewise.
6024 (output_function_details): Likewise.
6026 2017-11-13 Martin Liska <mliska@suse.cz>
6028 * gcov.c (struct function_info): Remove typedef for function_t.
6029 (struct source_info): Likewise.
6030 (source_info::get_functions_at_location): Likewise.
6031 (solve_flow_graph): Likewise.
6032 (find_exception_blocks): Likewise.
6033 (add_line_counts): Likewise.
6034 (output_intermediate_file): Likewise.
6035 (process_file): Likewise.
6036 (generate_results): Likewise.
6037 (release_structures): Likewise.
6038 (read_graph_file): Likewise.
6039 (read_count_file): Likewise.
6040 (accumulate_line_counts): Likewise.
6041 (output_lines): Likewise.
6043 2017-11-13 Martin Liska <mliska@suse.cz>
6045 * gcov.c (function_info::function_info): Remove num_counts
6046 and add vector<gcov_type>.
6047 (function_info::~function_info): Use the vector.
6048 (process_file): Likewise.
6049 (read_graph_file): Likewise.
6050 (read_count_file): Likewise.
6051 (solve_flow_graph): Likewise.
6053 2017-11-13 Martin Liska <mliska@suse.cz>
6055 * gcov.c (function_info::is_artificial): New function.
6056 (process_file): Erase all artificial early.
6057 (generate_results): Skip as all artificial are already
6060 2017-11-13 Martin Liska <mliska@suse.cz>
6062 * gcov.c (read_graph_file): Store to global vector of functions.
6063 (read_count_file): Iterate the vector.
6064 (process_file): Likewise.
6065 (generate_results): Likewise.
6066 (release_structures): Likewise.
6068 2017-11-13 Jakub Jelinek <jakub@redhat.com>
6070 PR tree-optimization/82954
6071 * gimple-ssa-store-merging.c
6072 (imm_store_chain_info::coalesce_immediate_stores): If
6073 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
6075 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
6077 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
6078 Upddate call to ENDIAN_LANE_N.
6079 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
6080 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
6081 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
6082 and use aarch64_endian_lane_rtx.
6083 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
6085 2017-11-12 Tom de Vries <tom@codesourcery.com>
6087 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
6089 2017-11-12 Tom de Vries <tom@codesourcery.com>
6091 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
6093 2017-11-12 Tom de Vries <tom@codesourcery.com>
6095 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
6097 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
6098 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
6099 * defaults.h (ASM_OUTPUT_LABELREF): Same.
6101 2017-11-11 Martin Sebor <msebor@redhat.com>
6104 * doc/extend.texi (attribute nonstring): Remove spurious argument.
6107 * prefic.c (translate_name): Replace strncpy with memcpy to
6108 avoid -Wstringop-truncation.
6110 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6112 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
6114 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6116 * predict.c (maybe_hot_frequency_p): Do not use cfun.
6118 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6120 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
6123 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
6125 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
6126 enable generating XXBRH if the value is in a vector register.
6127 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
6128 value is in a vector register.
6129 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
6130 register to register bswap64's instead of doing the GPR sequence
6131 used on previous machines.
6132 (bswapdi2_xxbrd): New insn.
6133 (bswapdi2_reg): Disallow on ISA 3.0.
6134 (register to register bswap64 splitter): Do not split the insn on
6135 ISA 3.0 systems that use XXBRD.
6137 2017-11-10 Martin Sebor <msebor@redhat.com>
6140 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
6142 (find_subframework_file): Same.
6144 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6146 * auto-profile.c (afdo_indirect_call): Drop frequency.
6147 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
6148 (cgraph_node::create_edge): Drop frequency argument.
6149 (cgraph_node::create_indirect_edge): Drop frequency argument.
6150 (cgraph_edge::make_speculative): Drop frequency arguments.
6151 (cgraph_edge::resolve_speculation): Do not update frequencies
6152 (cgraph_edge::dump_edge_flags): Do not dump frequency.
6153 (cgraph_node::dump): Check consistency in IPA mode.
6154 (cgraph_edge::maybe_hot_p): Use IPA counter.
6155 (cgraph_edge::verify_count_and_frequency): Rename to ...
6156 (cgraph_edge::verify_count): ... this one; drop frequency checking.
6157 (cgraph_node::verify_node): Update.
6158 * cgraph.h (struct cgraph_edge): Drop frequency.
6159 (cgraph_edge::frequency): New function.
6160 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
6162 (cgraph_edge::rebuild_edges): Likewise.
6163 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
6164 (duplicate_thunk_for_node): Do not pass frequency.
6165 (cgraph_node::create_clone): Scale only counts.
6166 (cgraph_node::create_virtual_clone): Do not pass frequency.
6167 (cgraph_node::create_edge_including_clones): Do not pass frequency.
6168 (cgraph_node::create_version_clone): Do not pass frequency.
6169 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
6170 (cgraph_node::expand_thunk): Do not pass frequency.
6171 (cgraph_node::create_wrapper): Do not pass frequency.
6172 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
6174 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
6175 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
6176 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
6177 (gather_caller_stats): Use frequency function.
6178 (ipcp_cloning_candidate_p): Use frequency function.
6179 (ipcp_propagate_stage): Use frequency function.
6180 (get_info_about_necessary_edges): Use frequency function.
6181 (update_profiling_info): Update only IPA profile.
6182 (update_specialized_profile): Use frequency functoin.
6183 (perhaps_add_new_callers): Update only IPA profile.
6184 * ipa-devirt.c (ipa_devirt): Use IPA profile.
6185 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
6186 (dump_ipa_call_summary): Use frequency function.
6187 (estimate_edge_size_and_time): Use frequency function.
6188 (ipa_merge_fn_summary_after_inlining): Use frequency function.
6189 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
6190 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
6191 (update_noncloned_counts): ... ths one; scale counts only.
6192 (clone_inlined_nodes): Do not scale frequency.
6193 (inline_call): Do not pass frequency.
6194 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
6195 (compute_inlined_call_time): Use IPA profile.
6196 (want_inline_small_function_p): Use IPA profile.
6197 (want_inline_self_recursive_call_p): Use IPA profile.
6198 (edge_badness): Use IPA profile.
6199 (lookup_recursive_calls): Use IPA profile.
6200 (recursive_inlining): Do not pass frequency.
6201 (resolve_noninline_speculation): Do not update frequency.
6202 (inline_small_functions): Collect max of IPA profile.
6203 (dump_overall_stats): Dump IPA porfile.
6204 (dump_inline_stats): Dump IPA porfile.
6205 (ipa_inline): Collect IPA stats.
6206 * ipa-inline.h (clone_inlined_nodes): Update prototype.
6207 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
6208 (ipa_propagate_frequency): Use frequency function.
6209 (ipa_profile): Cleanup.
6210 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
6211 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
6212 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
6213 (input_node): Do not stream frequency.
6214 (input_edge): Do not stream frequency.
6215 (merge_profile_summaries): Scale only IPA profiles.
6216 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
6217 * predict.c (drop_profile): Do not recompute frequency.
6218 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
6219 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
6220 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
6221 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
6222 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
6223 * tree-inline.c (copy_bb): Do not scale frequency.
6224 (expand_call_inline): Do not scale frequency.
6225 (tree_function_versioning): Do not scale frequency.
6226 * ubsan.c (ubsan_create_edge): Do not pass frequency.
6228 2017-11-10 Julia Koval <julia.koval@intel.com>
6230 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
6231 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
6232 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
6233 (_mm256_maskz_gf2p8affine_epi64_epi8)
6234 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
6235 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
6236 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
6237 (__builtin_ia32_vgf2p8affineqb_v32qi)
6238 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
6239 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
6241 2017-11-10 Tamar Christina <tamar.christina@arm.com>
6245 (arm_option_override): Refactor.
6246 (arm_option_reconfigure_globals): New.
6247 (arm_options_perform_arch_sanity_checks): New.
6248 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
6250 (arm_options_perform_arch_sanity_checks): Likewise
6252 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
6254 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
6255 (power9-qpmul): New.
6256 * rs6000/rs6000.md ("type" attr): Add qmul.
6257 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
6258 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
6259 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
6261 2017-11-10 Martin Sebor <msebor@redhat.com>
6264 * builtins.c (compute_objsize): Handle arrays that
6265 compute_builtin_object_size likes to fail for. Make extern.
6266 * builtins.h (compute_objsize): Declare.
6267 (check_strncpy_sizes): New function.
6268 (expand_builtin_strncpy): Call check_strncpy_sizes.
6269 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
6270 -Wstringop-truncation.
6271 (gimple_fold_builtin_strncat): Same.
6272 * gimple.c (gimple_build_call_from_tree): Set call location.
6273 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
6274 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
6275 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
6276 (handle_builtin_strlen): Use strlen_to_stridx.
6277 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
6279 Use strlen_to_stridx.
6280 (pass_strlen::execute): Release strlen_to_stridx.
6281 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
6282 (-Wstringop-truncation): Document new option.
6284 2017-11-10 Martin Liska <mliska@suse.cz>
6286 PR gcov-profile/82702
6287 * gcov.c (main): Handle intermediate files in a different
6289 (get_gcov_intermediate_filename): New function.
6290 (output_gcov_file): Remove support of intermediate files.
6291 (generate_results): Allocate intermediate file.
6292 (release_structures): Clean-up properly fn_end.
6293 (output_intermediate_file): Start iterating with line 1.
6295 2017-11-10 Jakub Jelinek <jakub@redhat.com>
6297 PR tree-optimization/82929
6298 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
6299 ops_swapped_p non-static data member.
6300 (store_immediate_info::store_immediate_info): Clear it.
6301 (imm_store_chain_info::coalesce_immediate_stores): If swapping
6302 ops set ops_swapped_p.
6303 (count_multiple_uses): Handle ops_swapped_p.
6305 2017-11-10 Martin Liska <mliska@suse.cz>
6307 * coverage.c (coverage_init): Stream information about
6308 support of has_unexecuted_blocks.
6309 * doc/gcov.texi: Document that.
6310 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
6311 * gcov.c (read_graph_file): Likewise.
6312 (output_line_beginning): Fix a small issue with
6315 2017-11-10 Bin Cheng <bin.cheng@arm.com>
6317 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
6318 reference of trivial component.
6320 2017-11-10 Jakub Jelinek <jakub@redhat.com>
6323 * gimple-ssa-store-merging.c
6324 (pass_store_merging::terminate_all_aliasing_chains): For
6325 gimple_store_p stmts also call refs_output_dependent_p.
6327 PR rtl-optimization/82913
6328 * compare-elim.c (try_merge_compare): Punt if def_insn is not
6331 2017-11-09 Jeff Law <law@redhat.com>
6333 * vr-values.h: New file with vr_values class.
6334 * tree-vrp.c: Include vr-values.h
6335 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
6336 data objects into the vr_values class.
6337 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
6338 (get_value_range): Make it a member function within vr_values class.
6339 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
6340 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
6341 (op_with_constant_singleton_value_range): Likewise.
6342 (extract_range_for_var_from_comparison_expr): Likewise.
6343 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
6344 (extract_range_from_binary_expr): Likewise.
6345 (extract_range_from_unary_expr): Likewise.
6346 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
6347 (check_for_binary_op_overflow, extract_range_basic): Likewise.
6348 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
6349 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
6350 (compare_name_with_value, compare_names): Likewise.
6351 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
6352 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
6353 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
6354 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
6355 (extract_range_from_phi_node): Likewise.
6356 (simplify_truth_ops_using_ranges): Likewise.
6357 (simplify_div_or_mod_using_ranges): Likewise.
6358 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
6359 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
6360 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
6361 (simplify_float_conversion_using_ranges): Likewise.
6362 (simplify_internal_call_using_ranges): Likewise.
6363 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
6364 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
6365 poitner via x_vr_values for calls into gimple folder.
6366 (vrp_initialize_lattice): Make this the vr_values ctor.
6367 (vrp_free_lattice): Make this the vr_values dtor.
6368 (set_vr_value): New function.
6369 (class vrp_prop): Add vr_values data member. Add various member
6370 functions as well as member functions that delegate to vr_values.
6371 (check_array_ref): Make a member function within vrp_prop class.
6372 (search_for_addr_array, vrp_initialize): Likewise.
6373 (vrp_finalize): Likewise. Revamp to avoid direct access to
6374 vr_value, values_propagated, etc.
6375 (check_array_bounds): Extract vrp_prop class instance pointer from
6376 walk info structure. Use it to call member functions.
6377 (check_all_array_refs): Make a member function within vrp_prop class.
6378 Smuggle class instance pointer via walk info structure.
6379 (x_vr_values): New local static.
6380 (vrp_valueize): Use x_vr_values to get class instance.
6381 (vr_valueize_1): Likewise.
6382 (class vrp_folder): Add vr_values data member. Add various member
6383 functions as well as member functions that delegate to vr_values.
6384 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
6385 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
6386 class instance from vr_values. Use it to call member functions.
6387 (vrp_dom_walker): Add vr_values data member.
6388 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
6389 instance via x_vr_values.
6390 (identify_jump_threads): Accept vr_values as argument. Store
6391 it into the walker structure.
6392 (evrp_dom_walker): Add vr_values class data member. Add various
6394 (evrp_dom_walker::try_find_new_range): Use vr_values data
6395 member to access the memory allocator.
6396 (evrp_dom_walker::before_dom_children): Store vr_values class
6397 instance into the vrp_folder class.
6398 (evrp_dom_walker::push_value_range): Rework to avoid direct
6399 access to num_vr_values and vr_value.
6400 (evrp_dom_walker::pop_value_range): Likewise.
6401 (execute_early_vrp): Remove call to vrp_initialize_lattice.
6402 Use vr_values to get to dump_all_value_ranges member function.
6403 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
6404 and simplify_cond_using_ranges_2 via vrp_prop class instance.
6405 Pass vr_values class instance down to identify_jump_threads.
6406 Remove call to vrp_free_lattice.
6407 (debug_all_value_ranges): Remove.
6409 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
6410 (vrp_folder): Likewise.
6412 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
6413 Get it from the existing bitmap instead.
6414 (vrp_intersect_ranges_1): Likewise.
6416 2017-11-09 Jakub Jelinek <jakub@redhat.com>
6418 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
6420 (store_immediate_info::store_immediate_info): Add bitnotp argument,
6421 set bit_not_p to it.
6422 (imm_store_chain_info::coalesce_immediate_stores): Break group
6423 if bit_not_p is different.
6424 (count_multiple_uses, split_group,
6425 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
6426 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
6427 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
6428 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
6429 to store_immediate_info ctor.
6431 2017-11-09 Jim Wilson <jimw@sifive.com>
6433 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
6434 (scan_prog_file): Likewise.
6436 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
6438 * bb-reorder.c (max_entry_frequency): Remove.
6439 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
6440 connect_traces, push_to_next_round_p): Remove prototypes.
6441 (find_traces_1_round): Use counts only.
6442 (push_to_next_round_p): Likewise.
6443 (find_traces): Likewise.
6444 (rotate_loop): Likewise.
6445 (find_traces_1_round): Likewise.
6446 (connect_traces): Likewise.
6447 (edge_order): Likewise.
6449 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
6451 * config/arm/arm.c (output_return_instruction): Add comments to
6452 indicate requirement for cmse_nonsecure_entry return to account
6453 for the size of clearing instruction output here.
6454 (thumb_exit): Likewise.
6455 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
6456 return in hardfloat mode.
6458 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
6460 * config/rs6000/rs6000.c (machine_function): Add a bool,
6461 "toc_is_wrapped_separately".
6462 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
6463 if it wasn't explicitly set or unset, we are optimizing for speed, and
6464 doing separate shrink-wrapping.
6465 (rs6000_get_separate_components): Enable the TOC component if
6466 saving the TOC register in the prologue.
6467 (rs6000_components_for_bb): Handle the TOC component.
6468 (rs6000_emit_prologue_components): Store the TOC register where needed.
6469 (rs6000_set_handled_components): Mark TOC as handled, if handled.
6470 (rs6000_emit_prologue): Don't save the TOC if that is already done.
6472 2017-11-09 Martin Jambor <mjambor@suse.cz>
6474 * ipa-param-manipulation.c: New file.
6475 * ipa-param-manipulation.h: Likewise.
6476 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
6477 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
6478 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
6479 (ipa_parm_adjustment): Likewise.
6480 (ipa_parm_adjustment_vec): Likewise.
6481 (ipa_get_vector_of_formal_parms): Moved declaration to
6482 ipa-param-manipulation.h.
6483 (ipa_get_vector_of_formal_parm_types): Likewise.
6484 (ipa_modify_formal_parameters): Likewise.
6485 (ipa_modify_call_arguments): Likewise.
6486 (ipa_combine_adjustments): Likewise.
6487 (ipa_dump_param_adjustments): Likewise.
6488 (ipa_modify_expr): Likewise.
6489 (ipa_get_adjustment_candidate): Likewise.
6490 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
6491 ipa-param-manipulation.c.
6492 (ipa_get_vector_of_formal_parm_types): Likewise.
6493 (ipa_modify_formal_parameters): Likewise.
6494 (ipa_modify_call_arguments): Likewise.
6495 (ipa_modify_expr): Likewise.
6496 (get_ssa_base_param): Likewise.
6497 (ipa_get_adjustment_candidate): Likewise.
6498 (index_in_adjustments_multiple_times_p): Likewise.
6499 (ipa_combine_adjustments): Likewise.
6500 (ipa_dump_param_adjustments): Likewise.
6501 * tree-sra.c: Also include ipa-param-manipulation.h
6502 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
6505 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6506 Alan Hayward <alan.hayward@arm.com>
6507 David Sherwood <david.sherwood@arm.com>
6509 * doc/sourcebuild.texi (vect_masked_store): Document.
6511 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6512 Alan Hayward <alan.hayward@arm.com>
6513 David Sherwood <david.sherwood@arm.com>
6515 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
6517 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6518 Alan Hayward <alan.hayward@arm.com>
6519 David Sherwood <david.sherwood@arm.com>
6521 * doc/sourcebuild.texi (vect_variable_length): Document.
6523 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6524 Alan Hayward <alan.hayward@arm.com>
6525 David Sherwood <david.sherwood@arm.com>
6527 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
6529 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6530 Alan Hayward <alan.hayward@arm.com>
6531 David Sherwood <david.sherwood@arm.com>
6533 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
6535 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6536 Alan Hayward <alan.hayward@arm.com>
6537 David Sherwood <david.sherwood@arm.com>
6539 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
6540 previously undocumented selectors.
6541 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
6543 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6545 * doc/rtl.texi (const_vector): Say that elements can be
6546 const_wide_ints too.
6547 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
6548 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
6549 (gen_vec_duplicate): Use it instead of CONSTANT_P.
6550 * optabs.c (expand_vector_broadcast): Likewise.
6552 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6553 Alan Hayward <alan.hayward@arm.com>
6554 David Sherwood <david.sherwood@arm.com>
6556 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
6557 scaled index even if the unscaled address was invalid.
6558 Don't increase the complexity of using a scale in that case.
6560 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6561 Alan Hayward <alan.hayward@arm.com>
6562 David Sherwood <david.sherwood@arm.com>
6564 * doc/rtl.texi: Rewrite the subreg rules so that they partition
6565 the inner register into REGMODE_NATURAL_SIZE bytes rather than
6566 UNITS_PER_WORD bytes.
6567 * emit-rtl.c (validate_subreg): Divide subregs into blocks
6568 based on REGMODE_NATURAL_SIZE of the inner mode.
6569 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
6570 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
6571 * expmed.c (lowpart_bit_field_p): Divide the value up into
6572 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
6573 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
6574 whether something is likely to occupy more than one register.
6576 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
6579 * ipa-inline-transform.c (update_noncloned_frequencies): Use
6580 profile_count::adjust_for_ipa_scaling.
6581 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
6582 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
6584 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
6586 2017-11-09 Jakub Jelinek <jakub@redhat.com>
6588 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
6589 (split_group): Add total_orig and total_new arguments, estimate the
6590 number of statements related to the store group without store merging
6591 and with store merging.
6592 (imm_store_chain_info::output_merged_store): Adjust split_group
6593 callers, punt if estimated number of statements with store merging
6594 is not smaller than estimated number of statements without it.
6596 (handled_load): Remove has_single_use checks.
6597 (pass_store_merging::process_store): Likewise.
6599 2017-11-09 Richard Biener <rguenther@suse.de>
6601 PR tree-optimization/82902
6602 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
6604 2017-11-09 Martin Liska <mliska@suse.cz>
6607 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
6610 2017-11-09 Martin Liska <mliska@suse.cz>
6612 PR tree-optimization/82669
6613 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
6615 2017-11-09 Martin Liska <mliska@suse.cz>
6617 PR gcov-profile/48463
6618 * coverage.c (coverage_begin_function): Output also end locus
6619 of a function and information whether the function is
6621 * gcov-dump.c (tag_function): Parse and print the information.
6622 * gcov.c (INCLUDE_MAP): Add include.
6623 (INCLUDE_SET): Likewise.
6624 (struct line_info): Move earlier in the source file because
6625 of vector<line_info> in function_info structure.
6626 (line_info::line_info): Likewise.
6627 (line_info::has_block): Likewise.
6628 (struct source_info): Add new member index.
6629 (source_info::get_functions_at_location): New function.
6630 (function_info::group_line_p): New function.
6631 (output_intermediate_line): New function.
6632 (output_intermediate_file): Use the mentioned function.
6633 (struct function_start): New.
6634 (struct function_start_pair_hash): Likewise.
6635 (process_file): Add code that identifies group functions.
6636 Assign lines either to global or function scope.
6637 (generate_results): Skip artificial functions.
6638 (find_source): Assign index for each source file.
6639 (read_graph_file): Read new flag artificial and end_line.
6640 (add_line_counts): Assign it either to global of function scope.
6641 (accumulate_line_counts): Isolate core of the function to
6642 accumulate_line_info and call it for both function and global
6644 (accumulate_line_info): New function.
6645 (output_line_beginning): Fix GNU coding style.
6646 (print_source_line): New function.
6647 (output_line_details): Likewise.
6648 (output_function_details): Likewise.
6649 (output_lines): Iterate both source (global) scope and function
6651 (struct function_line_start_cmp): New class.
6652 * doc/gcov.texi: Reflect changes in documentation.
6654 2017-11-09 Jakub Jelinek <jakub@redhat.com>
6657 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
6658 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
6659 and similarly for not instead of neg.
6661 2017-11-08 Andi Kleen <ak@linux.intel.com>
6663 * config/i386/i386.opt: Add -mforce-indirect-call.
6664 * config/i386/predicates.md: Check for flag_force_indirect_call.
6665 * doc/invoke.texi: Document -mforce-indirect-call
6667 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
6669 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
6671 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
6672 config/riscv/riscv.c (predict.h): New include.
6673 (riscv_slow_unaligned_access_p): No longer static.
6674 (riscv_block_move_straight): Add require.
6675 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
6677 2017-11-08 Jakub Jelinek <jakub@redhat.com>
6680 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
6681 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
6682 nonimmediate_operand predicate for operand 1 instead of
6685 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6687 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
6689 * config/aarch64/constraints.md (Uml): New constraint.
6690 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
6693 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6695 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
6696 of two vec_duplicates into a vec_concat.
6698 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6700 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
6701 Simplify vec_merge of vec_duplicate and vec_concat.
6702 * config/aarch64/constraints.md (Utq): New constraint.
6703 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
6706 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6708 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
6709 Simplify vec_merge of vec_duplicate and const_vector.
6710 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
6712 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
6713 mode iterator. Update predicate on operand 1 to
6714 handle non-const_vec constants. Delete constraints.
6715 (*aarch64_combinez_be<mode>): Likewise for operand 2.
6717 2017-11-08 Jakub Jelinek <jakub@redhat.com>
6719 PR tree-optimization/78821
6720 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
6722 (store_operand_info::store_operand_info): Initialize it to false.
6723 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
6724 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
6725 store in the group, and if chain_info is non-NULL, to ignore altogether
6727 (compatible_load_p): Fail if bit_not_p does not match.
6728 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
6729 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
6730 (pass_store_merging::process_store): Adjust
6731 terminate_all_aliasing_chains calls to pass NULL in all current spots,
6732 call terminate_all_aliasing_chains newly when adding a store into
6733 a chain with non-NULL chain_info.
6735 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
6737 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
6739 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
6741 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
6743 (aarch64_layout_frame): Initialise emit_frame_chain.
6744 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
6745 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
6747 2017-11-08 Martin Liska <mliska@suse.cz>
6749 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
6750 of gimple_call_internal_p.
6752 2017-11-07 Tom de Vries <tom@codesourcery.com>
6754 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
6756 2017-11-07 Tom de Vries <tom@codesourcery.com>
6758 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
6761 2017-11-08 Martin Liska <mliska@suse.cz>
6764 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
6766 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
6768 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
6770 2017-11-07 Jakub Jelinek <jakub@redhat.com>
6773 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
6774 (*cmp<mode>_ccz_1): New insn with $k alternative.
6777 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
6778 RTX_COMM_COMPARE as commutative as well.
6779 (ix86_binary_operator_ok): Formatting fix.
6780 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
6781 *<code><mode>3<mask_name><round_saeonly_name>,
6782 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
6783 *<s>mul<mode>3_highpart<mask_name>,
6784 *vec_widen_umult_even_v16si<mask_name>,
6785 *vec_widen_umult_even_v8si<mask_name>,
6786 *vec_widen_umult_even_v4si<mask_name>,
6787 *vec_widen_smult_even_v16si<mask_name>,
6788 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
6789 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
6790 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
6791 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
6792 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
6793 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
6794 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
6795 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
6796 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
6797 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
6798 ix86_binary_operator_ok. Formatting fixes.
6799 (*<plusminus_insn><mode>3<mask_name><round_name>,
6800 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
6803 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
6805 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
6806 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
6807 eq and ne if TARGET_ISEL.
6808 (cmp): New code_iterator.
6809 (UNS, UNSU_, UNSIK): New code_attrs.
6810 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
6811 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
6813 ("eq<mode>3"): ... this.
6814 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
6816 ("ne<mode>3"): ... this.
6818 2017-11-07 Julia Koval <julia.koval@intel.com>
6821 * common/config/i386/i386-common.c
6822 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
6823 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
6824 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
6825 * config/i386/i386.opt: Ditto.
6826 * config/i386/i386.c (ix86_target_string): Ditto.
6827 (ix86_option_override_internal): Ditto.
6828 (ix86_init_mpx_builtins): Move MPX to args2.
6829 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
6830 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
6831 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
6832 __builtin_ia32_vgf2p8affineinvqb_v32qi,
6833 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
6834 __builtin_ia32_vgf2p8affineinvqb_v16qi,
6835 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
6837 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
6840 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
6841 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
6842 (zero-extendsidi peephole2): Remove peephole.
6844 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
6847 * stor-layout.c: Include attribs.h.
6848 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
6849 explicit lookup of "aligned" attribute.
6851 2017-11-07 Andrew Waterman <andrew@sifive.com>
6853 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
6854 (riscv_expand_block_move): Likewise.
6855 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
6857 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
6858 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
6859 * config/riscv/riscv.c (riscv_block_move_straight): New function.
6860 (riscv_adjust_block_mem): Likewise.
6861 (riscv_block_move_loop): Likewise.
6862 (riscv_expand_block_move): Likewise.
6863 * config/riscv/riscv.md (movmemsi): New pattern.
6865 2017-11-07 Michael Clark <michaeljclark@mac.com>
6867 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
6868 (MUSL_DYNAMIC_LINKER): Likewise.
6870 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
6872 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
6873 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
6876 2017-11-07 Richard Biener <rguenther@suse.de>
6878 * match.pd: Fix build.
6880 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
6881 Jackson Woodruff <jackson.woodruff@arm.com>
6883 PR tree-optimization/71026
6884 * match.pd: Canonicalize negate in division.
6886 2017-11-07 Sudakshina Das <sudi.das@arm.com>
6889 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
6891 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
6893 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
6894 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
6896 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
6898 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
6899 non-scalar integral types.
6900 * match.pd (negate_expr_p): Handle MINUS_EXPR.
6901 (-(A-B), -(~A)): New transformations.
6903 2017-11-07 Tom de Vries <tom@codesourcery.com>
6905 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
6906 semicolon after "do {} while (0)".
6907 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6908 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6909 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6910 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6911 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6913 2017-11-07 Tom de Vries <tom@codesourcery.com>
6915 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
6916 after "do {} while (0)".
6917 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6918 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6919 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6920 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6921 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6923 2017-11-07 Tom de Vries <tom@codesourcery.com>
6926 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
6928 (arm_rtx_costs_internal): Add missing semicolon after
6929 HANDLE_NARROW_SHIFT_ARITH call.
6931 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
6933 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
6934 disable isel if it was not set explicitly.
6936 2017-11-06 James Bowman <james.bowman@ftdichip.com>
6938 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
6939 (add_type_attribute) likewise.
6941 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
6943 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
6944 of struct ix86_frame.
6945 (ix86_initial_elimination_offset): Likewise.
6946 (ix86_expand_split_stack_prologue): Likewise.
6948 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
6950 * tree-vrp.h (enum value_range_type): Update stale comment.
6952 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
6953 Alan Hayward <alan.hayward@arm.com>
6954 David Sherwood <david.sherwood@arm.com>
6956 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
6957 (aarch64_expand_vec_perm_const): Take the number of units too.
6958 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
6959 (aarch64_expand_vec_perm_const): Likewise.
6960 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
6961 (vec_perm<mode>): Update accordingly.
6963 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
6964 Alan Hayward <alan.hayward@arm.com>
6965 David Sherwood <david.sherwood@arm.com>
6967 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
6968 Take the number of units too.
6969 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
6970 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
6971 but check for a vector mode before rather than after the call.
6972 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
6973 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
6974 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
6975 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
6976 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
6977 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
6978 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
6979 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
6980 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
6981 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
6982 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
6983 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
6984 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
6985 (aarch64_sqdmull2_n<mode>): Update accordingly.
6987 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
6988 Alan Hayward <alan.hayward@arm.com>
6989 David Sherwood <david.sherwood@arm.com>
6991 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
6992 the number of units too.
6993 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
6994 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
6995 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
6996 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
6997 (vec_store_lanesxi<mode>): Update accordingly.
6999 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7000 Alan Hayward <alan.hayward@arm.com>
7001 David Sherwood <david.sherwood@arm.com>
7003 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
7004 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
7005 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
7006 of units rather than the mode.
7007 * config/aarch64/iterators.md (nunits): New mode attribute.
7008 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
7009 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
7010 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
7011 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
7012 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
7013 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
7014 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
7015 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
7016 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
7017 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
7018 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
7019 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
7020 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
7021 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
7022 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
7023 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
7024 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
7025 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
7026 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
7027 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
7028 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
7029 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
7030 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
7031 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
7032 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
7033 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
7034 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
7035 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
7036 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
7037 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
7038 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
7039 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
7040 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
7041 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
7042 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
7043 (aarch64_simd_vec_setv2di): Likewise.
7045 2017-11-06 Carl Love <cel@us.ibm.com>
7047 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
7049 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
7050 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
7052 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
7054 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
7055 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
7056 Add power 8 macro expansions.
7057 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
7058 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
7059 power 8 instructions. (VSX_XXBR): Add iterator.
7061 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
7063 * config/arm/arm.md (predicable_short_it): Change default to "no",
7064 improve documentation, remove uses that are identical to the default.
7065 (enabled_for_depr_it): Rename to enabled_for_short_it.
7066 * gcc/config/arm/arm-fixed.md (predicable_short_it):
7067 Remove default uses.
7068 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
7069 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
7070 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
7071 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
7073 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
7076 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
7077 float128 helper macros, which are no longer used after deleting
7078 the old 'q' built-in functions, and moving the round to odd
7079 built-in functions to being special built-in functions.
7080 (BU_FLOAT128_2): Likewise.
7081 (BU_FLOAT128_1_HW): Likewise.
7082 (BU_FLOAT128_2_HW): Likewise.
7083 (BU_FLOAT128_3_HW): Likewise.
7084 (FABSQ): Delete old 'q' built-in functions.
7085 (COPYSIGNQ): Likewise.
7086 (SQRTF128_ODD): Move round to odd built-in functions to be
7087 special built-in functions, so that we can handle
7088 -mabi=ieeelongdouble.
7089 (TRUNCF128_ODD): Likewise.
7090 (ADDF128_ODD): Likewise.
7091 (SUBF128_ODD): Likewise.
7092 (MULF128_ODD): Likewise.
7093 (DIVF128_ODD): Likewise.
7094 (FMAF128_ODD): Likewise.
7095 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
7096 built-in names to 'f128'.
7097 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
7098 old 'q' built-in functions, as the machine independent code for
7099 'f128' built-in functions handles this.
7100 (rs6000_expand_builtin): Add expansion for float128 round to odd
7101 functions, keying off on -mabi=ieeelongdouble of whether to use
7102 the KFmode or TFmode variant.
7103 (rs6000_init_builtins): Initialize the _Float128 round to odd
7105 * doc/extend.texi (PowerPC Built-in Functions): Document the old
7106 _Float128 'q' built-in functions are now mapped into the new
7107 'f128' built-in functions.
7109 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
7111 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
7112 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
7114 2017-11-06 David Malcolm <dmalcolm@redhat.com>
7117 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
7118 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
7119 * toplev.c: Include "ipa-fnsummary.h".
7120 (toplev::finalize): Call ipa_fnsummary_c_finalize.
7122 2017-11-06 Jakub Jelinek <jakub@redhat.com>
7124 PR tree-optimization/82838
7125 * gimple-ssa-store-merging.c
7126 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
7127 on a separate gimple_seq which is then appended to seq.
7129 2017-11-06 Jeff Law <law@redhat.com>
7132 * config/i386/i386.c (PROBE_INTERVAL): Remove.
7133 (get_probe_interval): New functions.
7134 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
7135 (ix86_adjust_stack_and_probe): Likewise.
7136 (output_adjust_stack_and_probe): Likewise.
7137 (ix86_emit_probe_stack_range): Likewise.
7138 (ix86_expand_prologue): Likewise.
7140 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7142 PR tree-optimization/82816
7143 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
7144 if the modes of the two types are the same.
7145 (convert_plusminus_to_widen): Likewise.
7147 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7149 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
7151 (usadv16qi): New define_expand.
7152 (usadv8hi): New define_expand.
7154 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
7157 * ipa-inline-transform.c (update_noncloned_frequencies): Always
7159 (inline_transform): Likewise.
7160 * predict.c (counts_to_freqs): Remove useless conditional.
7161 * profile-count.h (profile_count::apply_scale): Move sanity check.
7162 * tree-inline.c (copy_bb): Always scale.
7163 (copy_cfg_body): Likewise.
7165 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
7168 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
7170 (*cmp_ite0): Add enabled_for_depr_it attribute.
7171 (*cmp_ite1): Likewise.
7173 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
7175 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
7178 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7180 * tree-vrp.c (vrp_int_const_binop): Return true on success and
7181 return the value by pointer.
7182 (extract_range_from_multiplicative_op_1): Update accordingly.
7183 Return as soon as an operation fails.
7185 2017-11-05 Tom de Vries <tom@codesourcery.com>
7188 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
7189 (DEF_SANITIZER_BUILTIN): ... here.
7190 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
7191 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
7193 2017-11-05 Tom de Vries <tom@codesourcery.com>
7196 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
7198 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
7199 ASM_OUTPUT_BEFORE_CASE_LABEL call.
7200 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
7202 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
7203 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
7204 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
7206 2017-11-05 Tom de Vries <tom@codesourcery.com>
7209 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
7212 2017-11-04 Michael Clark <michaeljclark@mac.com>
7214 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
7215 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
7217 (*addsi3_extended): Likewise.
7218 (*addsi3_extended2): Likewise.
7219 (<optab>si3): Likewise.
7220 (<optab>di3): Likewise.
7221 (<optab><mode>3): Likewise.
7222 (<*optabe>si3_internal): Likewise.
7223 (zero_extendqi<SUPERQI:mode>2): Likewise.
7224 (*add<mode>hi3): Likewise.
7225 (*xor<mode>hi3): Likewise.
7226 (<optab>di3): Likewise.
7227 (*<optab>si3_extend): Likewise.
7228 (*sge<u>_<X:mode><GPR:mode>): Likewise.
7229 (*slt<u>_<X:mode><GPR:mode>): Likewise.
7230 (*sle<u>_<X:mode><GPR:mode>): Likewise.
7232 2017-11-04 Andrew Waterman <andrew@sifive.com>
7234 * config/riscv/riscv.c (riscv_option_override): Conditionally set
7235 TARGET_STRICT_ALIGN based upon -mtune argument.
7237 2017-11-04 Andrew Waterman <andrew@sifive.com>
7239 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
7241 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
7243 * config/i386/i386.c (choose_basereg): Use optional scratch
7244 register and add assertion.
7245 (x86_emit_outlined_ms2sysv_save): Use scratch register when
7246 needed, and don't allocate stack.
7247 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
7248 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
7249 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
7251 2017-11-03 Jeff Law <law@redhat.com>
7253 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
7254 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
7255 to probe at the start of a noreturn function.
7257 2017-11-03 Jakub Jelinek <jakub@redhat.com>
7259 PR tree-optimization/78821
7260 * gimple-ssa-store-merging.c: Update the file comment.
7261 (MAX_STORE_ALIAS_CHECKS): Define.
7262 (struct store_operand_info): New type.
7263 (store_operand_info::store_operand_info): New constructor.
7264 (struct store_immediate_info): Add rhs_code and ops data members.
7265 (store_immediate_info::store_immediate_info): Add rhscode, op0r
7266 and op1r arguments to the ctor, initialize corresponding data members.
7267 (struct merged_store_group): Add load_align_base and load_align
7269 (merged_store_group::merged_store_group): Initialize them.
7270 (merged_store_group::do_merge): Update them.
7271 (merged_store_group::apply_stores): Pick the constant for
7272 encode_tree_to_bitpos from one of the two operands, or skip
7273 encode_tree_to_bitpos if neither operand is a constant.
7274 (class pass_store_merging): Add process_store method decl. Remove
7275 bool argument from terminate_all_aliasing_chains method decl.
7276 (pass_store_merging::terminate_all_aliasing_chains): Remove
7277 var_offset_p argument and corresponding handling.
7278 (stmts_may_clobber_ref_p): New function.
7279 (compatible_load_p): New function.
7280 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
7281 if there is overlap and rhs_code is not INTEGER_CST. For
7282 non-overlapping stores terminate group if rhs is not mergeable.
7283 (get_alias_type_for_stmts): Change first argument from
7284 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
7285 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
7286 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
7288 (get_location_for_stmts): Change first argument from
7289 auto_vec<gimple *> & to vec<gimple *> &.
7290 (struct split_store): Remove orig_stmts data member, add orig_stores.
7291 (split_store::split_store): Create orig_stores rather than orig_stmts.
7292 (find_constituent_stmts): Renamed to ...
7293 (find_constituent_stores): ... this. Change second argument from
7294 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
7295 to info structures rather than the statements.
7296 (split_group): Rename ALLOW_UNALIGNED argument to
7297 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
7298 it. Adjust find_constituent_stores caller.
7299 (imm_store_chain_info::output_merged_store): Handle rhs_code other
7300 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
7301 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
7302 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
7303 (mem_valid_for_store_merging): New function.
7304 (handled_load): New function.
7305 (pass_store_merging::process_store): New method.
7306 (pass_store_merging::execute): Use process_store method. Adjust
7307 terminate_all_aliasing_chains caller.
7309 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
7311 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
7312 Return true for more constants, symbols and label references.
7313 (aarch64_valid_floating_const): Remove unused function.
7315 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
7318 * config/aarch64/aarch64.c (aarch64_layout_frame):
7319 Undo forcing of LR at bottom of frame.
7321 2017-11-03 Jeff Law <law@redhat.com>
7324 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
7325 for int_registers_saved.
7327 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
7328 extracted from tree-ssa-dom.c.
7329 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
7330 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
7331 (record_equivalences_from_incoming_edge): Add additional argument
7332 to single_pred_edge_ignoring_loop_edges call.
7333 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
7334 (uncprop_dom_walker::before_dom_children): Add additional argument
7335 to single_pred_edge_ignoring_loop_edges call.
7336 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
7337 single_pred_edge_ignoring_loop_edges rather than open coding.
7338 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
7340 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
7342 * match.pd (-(-A)): Rewrite.
7344 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
7346 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
7347 (rs6000_emit_int_cmove): New declaration.
7348 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
7349 (rs6000_emit_sISEL): Delete.
7350 (rs6000_emit_int_cmove): Make non-static.
7351 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
7352 instead of rs6000_emit_sISEL.
7354 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
7356 * asan.c (create_cond_insert_point): Maintain profile.
7357 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
7359 * basic-block.h (struct basic_block_def): Remove frequency.
7360 (EDGE_FREQUENCY): Use to_frequency
7361 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
7363 (find_traces): Update to use to_frequency.
7364 (find_traces_1_round): Likewise; use only IPA counts.
7365 (bb_to_key): Likewise.
7366 (connect_traces): Use IPA counts only.
7367 (copy_bb_p): Update to use to_frequency.
7368 (fix_up_crossing_landing_pad): Likewise.
7369 (sanitize_hot_paths): Likewise.
7370 * bt-load.c (basic_block_freq): Likewise.
7371 * cfg.c (init_flow): Set count_max to uninitialized.
7372 (check_bb_profile): Remove frequencies; check counts.
7373 (dump_bb_info): Do not dump frequencies.
7374 (update_bb_profile_for_threading): Update counts only.
7375 (scale_bbs_frequencies_int): Likewise.
7376 (MAX_SAFE_MULTIPLIER): Remove.
7377 (scale_bbs_frequencies_gcov_type): Update counts only.
7378 (scale_bbs_frequencies_profile_count): Update counts only.
7379 (scale_bbs_frequencies): Update counts only.
7380 * cfg.h (struct control_flow_graph): Add count-max.
7381 (update_bb_profile_for_threading): Update prototype.
7382 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
7383 (find_many_sub_basic_blocks): Likewise.
7384 * cfgcleanup.c (try_forward_edges): Likewise.
7385 (try_crossjump_to_edge): Likewise.
7386 * cfgexpand.c (expand_gimple_cond): Likewise.
7387 (expand_gimple_tailcall): Likewise.
7388 (construct_init_block): Likewise.
7389 (construct_exit_block): Likewise.
7390 * cfghooks.c (verify_flow_info): Check consistency of counts.
7391 (dump_bb_for_graph): Do not dump frequencies.
7392 (split_block_1): Do not update frequencies.
7393 (split_edge): Do not update frequencies.
7394 (make_forwarder_block): Do not update frequencies.
7395 (duplicate_block): Do not update frequencies.
7396 (account_profile_record): Do not update frequencies.
7397 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
7398 for global heuristics.
7399 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
7400 (expected_loop_iterations_unbounded): Use counts only.
7401 * cfgloopmanip.c (scale_loop_profile): Simplify.
7402 (create_empty_loop_on_edge): Simplify
7404 (duplicate_loop_to_header_edge): Simplify
7405 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
7406 (update_br_prob_note): Take care of removing note when profile
7408 (relink_block_chain): Do not dump frequency.
7409 (rtl_account_profile_record): Use to_frequency.
7410 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
7411 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
7412 (cgraph_update_edges_for_call_stmt_node): Likewise.
7413 (cgraph_edge::verify_count_and_frequency): Update.
7414 (cgraph_node::verify_node): Temporarily disable frequency verification.
7415 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
7416 to_cgraph_frequency.
7417 (cgraph_edge::rebuild_edges): Convert to ipa counts.
7418 * cgraphunit.c (init_lowered_empty_function): Do not initialize
7420 (cgraph_node::expand_thunk): Update profile.
7421 * except.c (dw2_build_landing_pads): Do not update frequency.
7422 * final.c (compute_alignments): Use to_frequency.
7423 (dump_basic_block_info): Do not dump frequency.
7424 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
7425 (dump_gimple_bb_header): Do not dump frequency.
7426 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
7428 * gimple-streamer-in.c (input_bb): Do not stream frequency.
7429 * gimple-streamer-out.c (output_bb): Do not stream frequency.
7430 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
7431 (init_before_recovery): Do not update frequency.
7432 (sched_create_recovery_edges): Do not update frequency.
7433 * hsa-gen.c (convert_switch_statements): Do not update frequency.
7434 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
7435 (ipa_cp_c_finalize): Set max_count to uninitialized.
7436 * ipa-fnsummary.c (get_minimal_bb): Use counts.
7437 (param_change_prob): Use counts.
7438 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
7440 * ipa-split.c (consider_split): Use to_frequency.
7441 (split_function): Use to_frequency.
7442 * ira-build.c (loop_compare_func): Likewise.
7443 (mark_loops_for_removal): Likewise.
7444 (mark_all_loops_for_removal): Likewise.
7445 * loop-doloop.c (doloop_modify): Do not update frequency.
7446 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
7448 * lto-streamer-in.c (input_function): Update count_max.
7449 * omp-expand.c (expand_omp_taskreg): Update count_max.
7450 * omp-simd-clone.c (simd_clone_adjust): Update profile.
7451 * predict.c (maybe_hot_frequency_p): Use to_frequency.
7452 (maybe_hot_count_p): Use ipa counts only.
7453 (maybe_hot_bb_p): Simplify.
7454 (maybe_hot_edge_p): Simplify.
7455 (probably_never_executed): Do not take frequency argument.
7456 (probably_never_executed_bb_p): Do not pass frequency.
7457 (probably_never_executed_edge_p): Likewise.
7458 (combine_predictions_for_bb): Check that profile is nonzero.
7459 (propagate_freq): Do not set frequency.
7460 (drop_profile): Simplify.
7461 (counts_to_freqs): Simplify.
7462 (expensive_function_p): Use to_frequency.
7463 (propagate_unlikely_bbs_forward): Simplify.
7464 (determine_unlikely_bbs): Simplify.
7465 (estimate_bb_frequencies): Add hack to silence graphite issues.
7466 (compute_function_frequency): Use ipa counts.
7467 (pass_profile::execute): Update.
7468 (rebuild_frequencies): Use counts only.
7469 (force_edge_cold): Use counts only.
7470 * profile-count.c (profile_count::dump): Dump new count types.
7471 (profile_count::differs_from_p): Check compatiblity.
7472 (profile_count::to_frequency): New function.
7473 (profile_count::to_cgraph_frequency): New function.
7474 * profile-count.h (struct function): Declare.
7475 (enum profile_quality): Add profile_guessed_local and
7476 profile_guessed_global0.
7477 (class profile_proability): Decrease number of bits to 29;
7478 update from_reg_br_prob_note and to_reg_br_prob_note.
7479 (class profile_count: Update comment; decrease number of bits
7480 to 61. Check compatibility.
7481 (profile_count::compatible_p): New private member function.
7482 (profile_count::ipa_p): New member function.
7483 (profile_count::operator<): Handle global zero correctly.
7484 (profile_count::operator>): Handle global zero correctly.
7485 (profile_count::operator<=): Handle global zero correctly.
7486 (profile_count::operator>=): Handle global zero correctly.
7487 (profile_count::nonzero_p): New member function.
7488 (profile_count::force_nonzero): New member function.
7489 (profile_count::max): New member function.
7490 (profile_count::apply_scale): Handle IPA scalling.
7491 (profile_count::guessed_local): New member function.
7492 (profile_count::global0): New member function.
7493 (profile_count::ipa): New member function.
7494 (profile_count::to_frequency): Declare.
7495 (profile_count::to_cgraph_frequency): Declare.
7496 * profile.c (OVERLAP_BASE): Delete.
7497 (compute_frequency_overlap): Delete.
7498 (compute_branch_probabilities): Do not use compute_frequency_overlap.
7499 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
7500 * sched-ebb.c (rank): Use counts only.
7501 * shrink-wrap.c (handle_simple_exit): Use counts only.
7502 (try_shrink_wrapping): Use counts only.
7503 (place_prologue_for_one_component): Use counts only.
7504 * tracer.c (find_best_predecessor): Use to_frequency.
7505 (find_trace): Use to_frequency.
7506 (tail_duplicate): Use to_frequency.
7507 * trans-mem.c (expand_transaction): Do not update frequency.
7508 * tree-call-cdce.c: Do not update frequency.
7509 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
7510 (gimple_merge_blocks): Likewise.
7511 (gimple_split_edge): Likewise.
7512 (gimple_duplicate_sese_region): Likewise.
7513 (gimple_duplicate_sese_tail): Likewise.
7514 (move_sese_region_to_fn): Likewise.
7515 (gimple_account_profile_record): Likewise.
7516 (insert_cond_bb): Likewise.
7517 * tree-complex.c (expand_complex_div_wide): Likewise.
7518 * tree-eh.c (lower_resx): Update profile.
7519 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
7521 (initialize_cfun): Do not initialize frequencies
7522 (freqs_to_counts): Delete.
7523 (copy_cfg_body): Ignore count parameter.
7524 (copy_body): Update.
7525 (expand_call_inline): Update count_max.
7526 (optimize_inline_calls): Update count_max.
7527 (tree_function_versioning): Update count_max.
7528 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
7529 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
7531 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
7532 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
7533 (try_peel_loop): Likewise.
7534 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
7536 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
7537 (tree_transform_and_unroll_loop): Do not use frequencies
7538 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
7539 Use reliable prediction only.
7540 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
7541 * tree-ssa-sink.c (select_best_block): Use to_frequency.
7542 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
7543 probability scaling.
7544 * tree-ssa-threadupdate.c (create_block_for_threading): Do
7545 not update frequency
7546 (any_remaining_duplicated_blocks): Likewise.
7547 (update_profile): Likewise.
7548 (estimated_freqs_path): Delete.
7549 (freqs_to_counts_path): Delete.
7550 (clear_counts_path): Delete.
7551 (ssa_fix_duplicate_block_edges): Likewise.
7552 (duplicate_thread_path): Likewise.
7553 * tree-switch-conversion.c (gen_inbound_check): Use counts.
7554 * tree-tailcall.c (decrease_profile): Do not update frequency.
7555 (eliminate_tail_call): Likewise.
7556 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
7557 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
7558 (optimize_mask_stores): Likewise.
7559 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
7560 * ubsan.c (ubsan_expand_null_ifn): Update profile.
7561 (ubsan_expand_ptr_ifn): Update profile.
7562 * value-prof.c (gimple_ic): Simplify.
7563 * value-prof.h (gimple_ic): Update prototype.
7564 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
7565 * ipa-inline.c (compute_uninlined_call_time): Be sure that
7567 (want_inline_self_recursive_call_p): Likewise.
7568 (resolve_noninline_speculation): Only cummulate defined counts.
7569 (inline_small_functions): Use nonzero_p.
7570 (ipa_inline): Do not access freed node.
7572 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
7574 * config/aarch64/aarch64.c (aarch64_override_options_internal):
7575 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
7577 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
7579 * config/riscv/riscv.c (riscv_legitimize_move): Handle
7580 non-legitimate address.
7582 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
7584 * config/rs6000/rs6000.md (*lt0_disi): Delete.
7585 (*lt0_<mode>di, *lt0_<mode>si): New.
7587 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
7589 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
7590 TARGET_PAIRED_FLOAT.
7592 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
7593 Jim Wilson <jim.wilson@linaro.org>
7595 * config/aarch64/aarch64-cores.def (saphira): New CPU.
7596 * config/aarch64/aarch64-tune.md: Regenerated.
7597 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
7598 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
7600 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
7602 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
7603 (arc_expand_prologue): Restore blink for millicode.
7604 * config/arc/linux.h (LINK_EH_SPEC): Defined.
7606 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
7609 * config/i386/i386.c (ix86_vector_duplicate_value): Use
7610 gen_vec_duplicate after forcing the scalar into a register.
7612 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
7614 * combine (try_combine): Print the insns input to try_combine to the
7617 2017-11-02 Steve Ellcey <sellcey@cavium.com>
7620 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
7621 Remove second argument from aarch64_process_target_attr call.
7622 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
7624 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
7626 (aarch64_handle_attr_arch): Remove second argument.
7627 (aarch64_handle_attr_cpu): Ditto.
7628 (aarch64_handle_attr_tune): Ditto.
7629 (aarch64_handle_attr_isa_flags): Ditto.
7630 (aarch64_process_one_target_attr): Ditto.
7631 (aarch64_process_target_attr): Ditto.
7632 (aarch64_option_valid_attribute_p): Remove second argument.
7633 on aarch64_process_target_attr call.
7635 2017-11-02 David Malcolm <dmalcolm@redhat.com>
7637 * diagnostic.c: Include "selftest-diagnostic.h".
7638 (selftest::assert_location_text): New function.
7639 (selftest::test_diagnostic_get_location_text): New function.
7640 (selftest::diagnostic_c_tests): Call it.
7642 2017-11-02 David Malcolm <dmalcolm@redhat.com>
7644 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
7645 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
7646 (class selftest::test_diagnostic_context): Move to...
7647 * selftest-diagnostic.c: New file.
7648 * selftest-diagnostic.h: New file.
7650 2017-11-02 James Bowman <james.bowman@ftdichip.com>
7652 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
7653 offset range for FT32B.
7654 * config/ft32/ft32.h: option "mcompress" enables relaxation.
7655 * config/ft32/ft32.md: Add TARGET_NOPM.
7656 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
7657 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
7659 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
7661 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
7663 2017-11-02 Jeff Law <law@redhat.com>
7665 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
7666 virtual keyword on FINAL OVERRIDE members.
7668 * tree-ssa-propagate.h (ssa_propagation_engine): Group
7669 virtuals together. Add virtual destructor.
7670 (substitute_and_fold_engine): Similarly.
7672 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
7674 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
7676 2017-11-02 Richard Biener <rguenther@suse.de>
7678 PR tree-optimization/82795
7679 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
7681 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7683 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
7685 (gcc_GAS_CHECK_FEATURE): Remove.
7686 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
7687 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
7688 * configure: Regenerate.
7690 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
7692 * config/arc/arc.c (hwloop_optimize): Account for empty
7695 2017-11-02 Richard Biener <rguenther@suse.de>
7698 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
7699 Truncate ARRAY_REF index and element size.
7701 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
7703 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
7705 2017-11-01 Jeff Law <law@redhat.com>
7707 * tree-ssa-ccp.c (ccp_folder): New class derived from
7708 substitute_and_fold_engine.
7709 (ccp_folder::get_value): New member function.
7710 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
7711 (ccp_fold_stmt): Remove prototype.
7712 (ccp_finalize): Call substitute_and_fold from the ccp_class.
7713 * tree-ssa-copy.c (copy_folder): New class derived from
7714 substitute_and_fold_engine.
7715 (copy_folder::get_value): Renamed from get_value.
7716 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
7717 * tree-vrp.c (vrp_folder): New class derived from
7718 substitute_and_fold_engine.
7719 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
7720 (vrp_folder::get_value): New member function.
7721 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
7722 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
7723 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
7724 provide a class interface to folder/substitute routines.
7725 (ssa_prop_fold_stmt_fn): Remove typedef.
7726 (ssa_prop_get_value_fn): Likewise.
7727 (subsitute_and_fold): Remove prototype.
7728 (replace_uses_in): Likewise.
7729 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
7730 Renamed from replace_uses_in. Call the virtual member function
7731 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
7732 (substitute_and_fold_dom_walker): Remove initialization of
7733 data member entries for calbacks. Add substitute_and_fold_engine
7734 member and initialize it.
7735 (substitute_and_fold_dom_walker::before_dom_children0: Use the
7736 member functions for get_value, replace_phi_args_in c
7737 replace_uses_in, and fold_stmt calls.
7738 (substitute_and_fold_engine::substitute_and_fold): Renamed from
7739 substitute_and_fold. Remove assert. Update ctor call.
7741 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
7742 (ssa_prop_visit_phi_fn): Likewise.
7743 (class ssa_propagation_engine): New class to provide an interface
7745 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
7747 (ssa_prop_visit_phi): Likewise.
7748 (ssa_propagation_engine::simulate_stmt): Moved into class.
7749 Call visit_phi/visit_stmt from the class rather than via
7750 file scoped static variables.
7751 (ssa_propagation_engine::simulate_block): Moved into class.
7752 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
7753 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
7754 set file scoped statics for the visit_stmt/visit_phi callbacks.
7755 * tree-complex.c (complex_propagate): New class derived from
7756 ssa_propagation_engine.
7757 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
7758 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
7759 (tree_lower_complex): Call ssa_propagate via the complex_propagate
7761 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
7762 ssa_propagation_engine.
7763 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
7764 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
7765 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
7766 * tree-ssa-copy.c (copy_prop): New class derived from
7767 ssa_propagation_engine.
7768 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
7769 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
7770 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
7771 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
7772 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
7773 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
7774 (execute_vrp): Call ssa_propagate from the vrp_prop class.
7776 2017-11-01 Jakub Jelinek <jakub@redhat.com>
7778 PR rtl-optimization/82778
7779 PR rtl-optimization/82597
7780 * compare-elim.c (struct comparison): Add in_a_setter field.
7781 (find_comparison_dom_walker::before_dom_children): Remove killed
7782 bitmap and df_simulate_find_defs call, instead walk the defs.
7783 Compute last_setter and initialize in_a_setter. Merge definitions
7784 with first initialization for a few variables.
7785 (try_validate_parallel): Use insn_invalid_p instead of
7786 recog_memoized. Return insn rather than just the pattern.
7787 (try_merge_compare): Fix up comment. Don't uselessly test if
7788 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
7790 (execute_compare_elim_after_reload): Remove df_chain_add_problem
7793 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7794 Alan Hayward <alan.hayward@arm.com>
7795 David Sherwood <david.sherwood@arm.com>
7797 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
7798 aarch64_hard_regno_nregs to get the number of registers
7801 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7802 Alan Hayward <alan.hayward@arm.com>
7803 David Sherwood <david.sherwood@arm.com>
7805 * config/aarch64/constraints.md (Upl): Rename to...
7807 * config/aarch64/aarch64.md
7808 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
7811 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7812 Alan Hayward <alan.hayward@arm.com>
7813 David Sherwood <david.sherwood@arm.com>
7815 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
7816 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
7819 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7820 Alan Hayward <alan.hayward@arm.com>
7821 David Sherwood <david.sherwood@arm.com>
7823 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
7824 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
7825 (aarch64_evpc_dup): Generate rtl direcly, rather than using
7827 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
7828 of a single element.
7829 * config/aarch64/iterators.md: Add a comment above the permute
7830 unspecs to say that they are generated directly by
7831 aarch64_expand_vec_perm_const.
7832 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
7834 2017-11-01 Nathan Sidwell <nathan@acm.org>
7836 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
7838 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
7840 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
7841 and medany code models, and describe what they do.
7843 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7845 Revert accidental duplicate:
7847 * combine.c (can_change_dest_mode): Reject changes in
7848 REGMODE_NATURAL_SIZE.
7850 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
7852 PR rtl-optimization/64682
7853 PR rtl-optimization/69567
7854 PR rtl-optimization/69737
7855 PR rtl-optimization/82683
7856 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
7857 register mentioned in the note, drop the note, unless it came from I3,
7858 in which case it should go to I3 again.
7860 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7862 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
7863 and return false if not.
7864 (clear_bytes_written_by, live_bytes_read): Update accordingly.
7866 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7868 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
7869 range is known to be empty.
7871 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7872 Alan Hayward <alan.hayward@arm.com>
7873 David Sherwood <david.sherwood@arm.com>
7875 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
7876 and CONST_VECTOR_NUNITS instead of computing the number of units from
7877 the byte sizes of the vector and element.
7878 (simplify_binary_operation_1): Likewise.
7879 (simplify_const_binary_operation): Likewise.
7880 (simplify_ternary_operation): Likewise.
7882 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7883 Alan Hayward <alan.hayward@arm.com>
7884 David Sherwood <david.sherwood@arm.com>
7886 * var-tracking.c (INT_MEM_OFFSET): Replace with...
7887 (int_mem_offset): ...this new function.
7888 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
7889 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
7890 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
7893 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7894 Alan Hayward <alan.hayward@arm.com>
7895 David Sherwood <david.sherwood@arm.com>
7897 * lower-subreg.c (interesting_mode_p): New function.
7898 (compute_costs, find_decomposable_subregs, decompose_register)
7899 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
7900 (resolve_clobber, dump_choices): Use it.
7902 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7903 Alan Hayward <alan.hayward@arm.com>
7904 David Sherwood <david.sherwood@arm.com>
7906 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
7908 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7909 Alan Hayward <alan.hayward@arm.com>
7910 David Sherwood <david.sherwood@arm.com>
7912 * alias.c (find_base_value, find_base_term): Only process integer
7913 truncations. Check the precision rather than the size.
7915 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7916 Alan Hayward <alan.hayward@arm.com>
7917 David Sherwood <david.sherwood@arm.com>
7919 * machmode.h (is_narrower_int_mode): New function
7920 * optabs.c (expand_float, expand_fix): Use it.
7921 * dwarf2out.c (rotate_loc_descriptor): Likewise.
7923 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7924 Alan Hayward <alan.hayward@arm.com>
7925 David Sherwood <david.sherwood@arm.com>
7927 * rtl.h (narrower_subreg_mode): New function.
7928 * ira-color.c (update_costs_from_allocno): Use it.
7930 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7931 Alan Hayward <alan.hayward@arm.com>
7932 David Sherwood <david.sherwood@arm.com>
7934 * optabs-query.h (convert_optab_p): New function, split out from...
7935 (convert_optab_handler): ...here.
7936 (widening_optab_handler): Delete.
7937 (find_widening_optab_handler): Remove permit_non_widening parameter.
7938 (find_widening_optab_handler_and_mode): Likewise. Provide an
7939 override that operates on mode class wrappers.
7940 * optabs-query.c (widening_optab_handler): Delete.
7941 (find_widening_optab_handler_and_mode): Remove permit_non_widening
7942 parameter. Assert that the two modes are the same class and that
7943 the "from" mode is narrower than the "to" mode. Use
7944 convert_optab_handler instead of widening_optab_handler.
7945 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
7946 instead of widening_optab_handler.
7947 * expr.c (expand_expr_real_2): Update calls to
7948 find_widening_optab_handler.
7949 * optabs.c (expand_widen_pattern_expr): Likewise.
7950 (expand_binop_directly): Take the insn_code as a parameter.
7951 (expand_binop): Only call find_widening_optab_handler for
7952 conversion optabs; use optab_handler otherwise. Update calls
7953 to find_widening_optab_handler and expand_binop_directly.
7954 Use convert_optab_handler instead of widening_optab_handler.
7955 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
7956 find_widening_optab_handler and use scalar_mode rather than
7958 (convert_plusminus_to_widen): Likewise.
7960 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7961 Alan Hayward <alan.hayward@arm.com>
7962 David Sherwood <david.sherwood@arm.com>
7964 * machmode.h (fixed_size_mode): New class.
7965 * rtl.h (get_pool_mode): Return fixed_size_mode.
7966 * gengtype.c (main): Add fixed_size_mode.
7967 * target.def (get_raw_result_mode): Return a fixed_size_mode.
7968 (get_raw_arg_mode): Likewise.
7969 * doc/tm.texi: Regenerate.
7970 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
7971 * targhooks.c (default_get_reg_raw_mode): Likewise.
7972 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
7973 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
7974 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
7975 (msp430_get_raw_result_mode): Likewise.
7976 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
7977 * dbxout.c (dbxout_parms): Require fixed-size modes.
7978 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
7979 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
7980 * omp-low.c (lower_oacc_reductions): Likewise.
7981 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
7982 (simplify_subreg): Update accordingly.
7983 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
7984 (force_const_mem): Update accordingly. Return NULL_RTX for modes
7985 that aren't fixed-size.
7986 (get_pool_mode): Return a fixed_size_mode.
7987 (output_constant_pool_2): Take a fixed_size_mode.
7989 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7990 Alan Hayward <alan.hayward@arm.com>
7991 David Sherwood <david.sherwood@arm.com>
7993 * doc/rtl.texi (vec_series): Document.
7994 (const): Say that the operand can be a vec_series.
7995 * rtl.def (VEC_SERIES): New rtx code.
7996 * rtl.h (const_vec_series_p_1): Declare.
7997 (const_vec_series_p): New function.
7998 * emit-rtl.h (gen_const_vec_series): Declare.
7999 (gen_vec_series): Likewise.
8000 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
8001 (gen_vec_series): Likewise.
8002 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
8003 * simplify-rtx.c (simplify_unary_operation): Handle negations
8005 (simplify_binary_operation_series): New function.
8006 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
8007 (test_vector_ops_series): New function.
8008 (test_vector_ops): Call it.
8009 * config/powerpcspe/altivec.md (altivec_lvsl): Use
8010 gen_const_vec_series.
8011 (altivec_lvsr): Likewise.
8012 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
8014 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8015 Alan Hayward <alan.hayward@arm.com>
8016 David Sherwood <david.sherwood@arm.com>
8018 * doc/rtl.texi (const): Update description of address constants.
8019 Say that vector constants are allowed too.
8020 * common.md (E, F): Use CONSTANT_P instead of checking for
8022 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
8023 checking for CONST_VECTOR.
8024 * expmed.c (make_tree): Use build_vector_from_val for a CONST
8026 * expr.c (expand_expr_real_2): Check for vector modes instead
8027 of checking for CONST_VECTOR.
8028 * rtl.h (const_vec_p): New function.
8029 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
8030 (unwrap_const_vec_duplicate): Handle them here too.
8032 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8033 David Malcolm <dmalcolm@redhat.com>
8034 Alan Hayward <alan.hayward@arm.com>
8035 David Sherwood <david.sherwood@arm.com>
8037 * rtl.h (vec_duplicate_p): New function.
8038 * selftest-rtl.c (assert_rtx_eq_at): New function.
8039 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
8040 (assert_rtx_eq_at): Declare.
8041 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
8042 * selftest-run-tests.c (selftest::run_tests): Call it.
8043 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
8044 (simplify_unary_operation_1): Recursively handle vector duplicates.
8045 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
8047 (simplify_subreg): Handle subregs of vector duplicates.
8048 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
8049 (selftest::simplify_rtx_c_tests): New functions.
8051 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8052 Alan Hayward <alan.hayward@arm.com>
8053 David Sherwood <david.sherwood@arm.com>
8055 * emit-rtl.h (gen_const_vec_duplicate): Declare.
8056 (gen_vec_duplicate): Likewise.
8057 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
8059 (gen_const_vector): ...here.
8060 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
8061 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
8062 whose elements are all equal.
8063 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
8064 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
8065 (simplify_relational_operation): Likewise.
8066 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
8068 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
8069 (aarch64_expand_vector_init): Likewise.
8070 * config/arm/arm.c (neon_vdup_constant): Likewise.
8071 (neon_expand_vector_init): Likewise.
8072 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
8073 (arm_block_set_unaligned_vect): Likewise.
8074 (arm_block_set_aligned_vect): Likewise.
8075 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
8076 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
8077 (expand_vec_perm_even_odd_pack): Likewise.
8078 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
8079 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
8080 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
8081 gen_const_vec_duplicate.
8082 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
8083 * config/mips/mips.c (mips_gen_const_int_vector): Use
8084 gen_const_vec_duplicate.
8085 (mips_expand_vector_init): Use CONST0_RTX.
8086 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
8087 (define_split): Use gen_const_vec_duplicate.
8088 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
8089 (define_split): Use gen_const_vec_duplicate.
8090 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
8091 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
8092 * config/spu/spu.c (spu_const): Likewise.
8094 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8095 Alan Hayward <alan.hayward@arm.com>
8096 David Sherwood <david.sherwood@arm.com>
8098 * combine.c (can_change_dest_mode): Reject changes in
8099 REGMODE_NATURAL_SIZE.
8101 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
8103 * configure.ac (--enable-libssp): New.
8104 (gcc_cv_libc_provides_ssp): Check for explicit setting before
8105 trying to determine target-specific default. Adjust indentation.
8106 * configure: Regenerated.
8107 * doc/install.texi (Configuration): Expand --disable-libssp
8110 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
8112 config/i386/i386.c (ix86_expand_epilogue): Correct stack
8115 2017-10-31 Martin Jambor <mjambor@suse.cz>
8118 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
8120 2017-10-31 David Malcolm <dmalcolm@redhat.com>
8122 * auto-profile.c (autofdo_source_profile::read): Use
8123 UNKNOWN_LOCATION rather than 0.
8124 * diagnostic-core.h (warning_at_rich_loc): Rename to...
8125 (warning_at): ...this overload.
8126 (warning_at_rich_loc_n): Rename to...
8127 (warning_n): ...this overload.
8128 (error_at_rich_loc): Rename to...
8129 (error_at): ...this overload.
8130 (pedwarn_at_rich_loc): Rename to...
8131 (pedwarn): ...this overload.
8132 (permerror_at_rich_loc): Rename to...
8133 (permerror): ...this overload.
8134 (inform_at_rich_loc): Rename to...
8135 (inform): ...this overload.
8136 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
8137 (diagnostic_n_impl_richloc): Rename to...
8138 (diagnostic_n_impl): ...this rich_location *-based decl.
8139 (inform_at_rich_loc): Rename to...
8140 (inform): ...this, and add an assertion.
8141 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
8142 (warning_at_rich_loc): Rename to...
8143 (warning_at): ...this, and add an assertion.
8144 (warning_at_rich_loc_n): Rename to...
8145 (warning_n): ...this, and add an assertion.
8146 (warning_n): Update location_t-based implementation for removal of
8147 location_t-based diagnostic_n_impl.
8148 (pedwarn_at_rich_loc): Rename to...
8149 (pedwarn): ...this, and add an assertion.
8150 (permerror_at_rich_loc): Rename to...
8151 (permerror): ...this, and add an assertion.
8152 (error_n): Update for removal of location_t-based diagnostic_n_impl.
8153 (error_at_rich_loc): Rename to...
8154 (error_at): ...this, and add an assertion.
8155 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
8156 (driver::do_spec_on_infiles): Likewise.
8157 * substring-locations.c (format_warning_va): Update for renaming
8158 of inform_at_rich_loc.
8160 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
8162 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
8163 _Float<N>X built-in functions so that the variant without the
8164 "__builtin_" prefix is only enabled for the GNU C and Objective C
8165 languages when they are in non-strict ANSI/ISO mode.
8166 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
8167 * target.def (floatn_builtin_p): Add a target hook to control
8168 whether _Float<N> and _Float<N>X built-in functions without the
8169 "__builtin_" prefix are enabled, and return true for C and
8170 Objective C in the default hook. Include langhooks.h in
8172 * targhooks.h (default_floatn_builtin_p): Likewise.
8173 * targhooks.c (default_floatn_builtin_p): Likewise.
8174 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
8175 floatn_builtin_p target hook.
8176 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
8178 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
8179 Eric Botcazou <ebotcazou@adacore.com>
8181 PR rtl-optimization/81803
8182 * lra-constraints.c (curr_insn_transform): Also reload the whole
8183 register for a strict subreg no wider than a word if this is for
8184 a WORD_REGISTER_OPERATIONS target.
8186 2017-10-31 Jason Merrill <jason@redhat.com>
8188 * gdbinit.in: Skip over inlines from timevar.h.
8190 2017-10-31 Martin Liska <mliska@suse.cz>
8192 * doc/gcov.texi: Document new option.
8193 * gcov.c (print_usage): Likewise print it.
8194 (process_args): Support the argument.
8195 (format_count): New function.
8196 (format_gcov): Use the function.
8198 2017-10-31 Martin Liska <mliska@suse.cz>
8200 * gcov.c (struct name_map): do not use typedef.
8201 Define operator== and operator<.
8202 (name_search): Remove.
8203 (name_sort): Remove.
8204 (main): Do not allocate names.
8205 (process_file): Add vertical space.
8206 (generate_results): Use std::find.
8207 (release_structures): Do not release memory.
8208 (find_source): Use std::find.
8210 2017-10-31 Martin Liska <mliska@suse.cz>
8212 * gcov.c (struct line_info): Remove it's typedef.
8213 (line_info::line_info): Add proper ctor.
8214 (line_info::has_block): Do not use a typedef.
8215 (struct source_info): Do not use typedef.
8216 (circuit): Likewise.
8217 (get_cycles_count): Likewise.
8218 (output_intermediate_file): Iterate via vector iterator.
8219 (add_line_counts): Use std::vector methods.
8220 (accumulate_line_counts): Likewise.
8221 (output_lines): Likewise.
8223 2017-10-31 Martin Liska <mliska@suse.cz>
8225 * gcov.c (struct source_info): Remove typedef.
8226 (source_info::source_info): Add proper ctor.
8227 (accumulate_line_counts): Use struct, not it's typedef.
8228 (output_gcov_file): Likewise.
8229 (output_lines): Likewise.
8230 (main): Do not allocate an array.
8231 (output_intermediate_file): Use size of vector container.
8232 (process_file): Resize the vector.
8233 (generate_results): Do not preallocate, use newly added vector
8235 (release_structures): Do not release sources.
8236 (find_source): Use vector methods.
8237 (add_line_counts): Do not use typedef.
8239 2017-10-31 Martin Liska <mliska@suse.cz>
8241 * doc/gcov.texi: Document that.
8242 * gcov.c (add_line_counts): Mark lines with a non-executed
8244 (output_line_beginning): Handle such lines.
8245 (output_lines): Pass new argument.
8246 (output_intermediate_file): Print it in intermediate format.
8248 2017-10-31 Martin Liska <mliska@suse.cz>
8250 * color-macros.h: New file.
8251 * diagnostic-color.c: Factor out color related to macros to
8253 * doc/gcov.texi: Document -k option.
8254 * gcov.c (INCLUDE_STRING): Include string.h.
8255 (print_usage): Add -k option.
8256 (process_args): Parse it.
8257 (pad_count_string): New function.
8258 (output_line_beginning): Likewise.
8259 (DEFAULT_LINE_START): New macro.
8260 (output_lines): Support color output.
8262 2017-10-31 Martin Liska <mliska@suse.cz>
8264 PR gcov-profile/82633
8265 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
8266 their interaction with GCOV infrastructure.
8267 * configure.ac: Add -fkeep-{inline,static}-functions to
8269 * configure: Regenerate.
8271 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
8274 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
8276 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
8279 * config/rs6000/rs6000.md (allocate_stack): Force update interval
8280 into a register if it does not fit into an immediate offset field.
8282 2017-10-31 Olivier Hainque <hainque@adacore.com>
8284 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
8286 2017-10-31 Julia Koval <julia.koval@intel.com>
8288 * config.gcc: Add gfniintrin.h.
8289 * config/i386/gfniintrin.h: New.
8290 * config/i386/i386-builtin-types.def
8291 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
8292 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
8293 __builtin_ia32_vgf2p8affineinvqb_v32qi,
8294 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
8295 __builtin_ia32_vgf2p8affineinvqb_v16qi,
8296 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
8297 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
8298 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
8299 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
8300 V64QI_FTYPE_V64QI_V64QI_INT): New types.
8301 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
8302 * config/i386/immintrin.h: Include gfniintrin.h.
8303 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
8305 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
8307 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
8309 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
8311 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
8312 (arm_ashldi3_1bit): Remove pattern.
8313 (ashrdi3): Remove shift by 1 expansion.
8314 (arm_ashrdi3_1bit): Remove pattern.
8315 (lshrdi3): Remove shift by 1 expansion.
8316 (arm_lshrdi3_1bit): Remove pattern.
8317 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
8318 cost of ashldi3 by 1.
8319 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
8320 (<shift>di3_neon): Likewise.
8322 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
8324 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
8325 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
8326 and (*aarch64_simd_mov<VQ:mode>).
8327 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
8328 pattern alternative.
8329 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
8330 attributes to match pattern alternative.
8332 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
8334 * config.gcc (powerpc*-*-*): Add emmintrin.h.
8335 * config/rs6000/emmintrin.h: New file.
8336 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
8338 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
8340 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
8341 * (movdi_vfp_cortexa8): Remove pattern.
8343 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8345 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
8347 (Specific, alpha*-dec-osf5.1): Remove.
8348 (Specific, mips-sgi-irix5): Remove.
8349 (Specific, mips-sgi-irix6): Remove.
8351 2017-10-30 Jakub Jelinek <jakub@redhat.com>
8354 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
8355 arguments to clear_bit_region_be.
8357 2017-10-30 Jim Wilson <wilson@tuliptree.org>
8359 * gimplify.c: Include memmodel.h.
8361 2017-10-30 Martin Jambor <mjambor@suse.cz>
8363 * omp-grid.c (grid_attempt_target_gridification): Also insert a
8364 condition whether loop should be executed at all.
8366 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
8368 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
8369 gimple folding of vec_madd() intrinsics.
8370 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
8371 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
8372 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
8374 2017-10-30 Richard Biener <rguenther@suse.de>
8376 PR tree-optimization/82762
8378 2017-10-23 Richard Biener <rguenther@suse.de>
8380 PR tree-optimization/82129
8382 2017-08-01 Richard Biener <rguenther@suse.de>
8384 PR tree-optimization/81181
8385 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
8386 (compute_antic): ... end of iteration here.
8388 2017-10-30 Joseph Myers <joseph@codesourcery.com>
8390 * doc/invoke.texi (C Dialect Options): Document -std=c17,
8391 -std=iso9899:2017 and -std=gnu17.
8392 * doc/standards.texi (C Language): Document C17 support.
8393 * doc/cpp.texi (Overview): Mention -std=c17.
8394 (Standard Predefined Macros): Document C11 and C17 values of
8395 __STDC_VERSION__. Do not refer to C99 support as incomplete.
8396 * doc/extend.texi (Inline): Do not list individual options for
8397 standards newer than C99.
8398 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
8400 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
8403 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
8405 * asan.c (asan_finish_file): Align asan globals array by shadow
8408 2017-10-30 Jakub Jelinek <jakub@redhat.com>
8411 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
8412 (struct store_immediate_info): Add bitregion_start and bitregion_end
8414 (store_immediate_info::store_immediate_info): Add brs and bre
8415 arguments and initialize bitregion_{start,end} from those.
8416 (struct merged_store_group): Add bitregion_start, bitregion_end,
8417 align_base and mask fields. Drop unnecessary struct keyword from
8418 struct store_immediate_info. Add do_merge method.
8419 (clear_bit_region_be): Use memset instead of loop storing zeros.
8420 (merged_store_group::do_merge): New method.
8421 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
8422 stores as long as the surrounding bitregions have no gaps.
8423 (merged_store_group::merge_overlapping): Use do_merge.
8424 (merged_store_group::apply_stores): Test that bitregion_{start,end}
8425 is byte aligned, rather than requiring that start and width are
8426 byte aligned. Drop unnecessary struct keyword from
8427 struct store_immediate_info. Allocate and populate also mask array.
8428 Make start of the arrays relative to bitregion_start rather than
8429 start and size them according to bitregion_{end,start} difference.
8430 (struct imm_store_chain_info): Drop unnecessary struct keyword from
8431 struct store_immediate_info.
8432 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
8433 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
8434 struct keyword from struct store_immediate_info.
8435 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
8436 between stores as long as the surrounding bitregions have no gaps.
8438 (struct split_store): Add orig non-static data member.
8439 (split_store::split_store): Initialize orig to false.
8440 (find_constituent_stmts): Return store_immediate_info *, non-NULL
8441 if there is exactly a single original stmt. Change stmts argument
8442 to pointer from reference, if NULL, don't push anything to it. Add
8443 first argument, use it to optimize skipping over orig stmts that
8444 are known to be before bitpos already. Simplify.
8445 (split_group): Return unsigned int count how many stores are or
8446 would be needed rather than a bool. Add allow_unaligned argument.
8447 Change split_stores argument from reference to pointer, if NULL,
8448 only do a dry run computing how many stores would be produced.
8449 Rewritten algorithm to use both alignment and misalign if
8450 !allow_unaligned and handle bitfield stores with gaps.
8451 (imm_store_chain_info::output_merged_store): Set start_byte_pos
8452 from bitregion_start instead of start. Compute allow_unaligned
8453 here, if true, do 2 split_group dry runs to compute which one
8454 produces fewer stores and prefer aligned if equal. Punt if
8455 new count is bigger or equal than original before emitting any
8456 statements, rather than during that. Remove no longer needed
8457 new_ssa_names tracking. Replace num_stmts with
8458 split_stores.length (). Use 32-bit stack allocated entries
8459 in split_stores auto_vec. Try to reuse original store lhs/rhs1
8460 if possible. Handle bitfields with gaps.
8461 (pass_store_merging::execute): Ignore bitsize == 0 stores.
8462 Compute bitregion_{start,end} for the stores and construct
8463 store_immediate_info with that. Formatting fixes.
8465 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
8468 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
8469 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
8471 2017-10-29 Jim Wilson <wilson@tuliptree.org>
8473 * gimplify.c: Include tm_p.h.
8475 * common.opt (gcoff): Re-add as ignored option.
8476 (gcoff1, gcoff2, gcoff3): Likewise.
8478 * Makefile.in (OBJS): Delete sdbout.o.
8479 (GTFILES): Delete $(srcdir)/sdbout.c.
8480 * debug.h: Delete sdb_debug_hooks.
8481 * final.c: Delete sdbout.h include.
8482 (final_scan_insn): Delete SDB_DEBUG check.
8483 (rest_of_clean_state): Likewise.
8484 * output.h: Delete sdb_begin_function_line.
8487 * toplev.c: Delete sdbout.h include.
8488 (process_options): Delete SDB_DEBUG check.
8489 * tree-core.h (tree_type_common): Delete pointer field of
8491 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
8492 TYPE_SYMTAB_POINTER.
8493 * tree.h (TYPE_SYMTAB_POINTER): Delete.
8494 (TYPE_SYMTAB_IS_POINTER): Delete.
8495 (TYPE_SYMTAB_IS_DIE): Renumber.
8496 * xcoffout.c: Refer to former sdbout.c file.
8497 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
8499 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
8500 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
8501 (Debugging Options): Delete -gcoff.
8502 (-gstabs): Delete SDB reference.
8504 (-gcoff@var{level}): Delete.
8505 * doc/passes.texi (Debugging information output): Delete SDB and
8506 sdbout.c references.
8507 * doc/tm.texi: Regenerate.
8508 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
8509 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
8511 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
8512 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
8513 (SDB_DEBUGGING_INFO): Delete.
8514 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
8515 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
8516 * target.def (output_source_filename): Delete COFF reference.
8518 * common.opt (gcoff): Delete.
8519 (gxcoff+): Update Negative chain.
8520 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
8522 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
8523 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
8524 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
8526 (expand_function_start): Change sdb reference to past tense.
8527 (expand_function_end): Change sdb reference to past tense.
8528 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
8529 * opts.c (debug_type_names): Delete coff entry.
8530 (common_handle_option): Delete OPT_gcoff case.
8531 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
8533 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
8534 * config/cris/cris.h: Delete SDB reference in comment.
8535 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
8536 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
8537 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
8538 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
8540 (ix86_expand_prologue): Likewise.
8541 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
8542 * config/ia64/ia64.h: Likewise.
8543 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
8544 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
8546 * config/mmix/mmix.h: Likewise.
8547 * config/nds32/nds32.c: Likewise.
8548 * config/stormy/storym16.h: Likewise.
8549 * config/visium/visium.h: Likewise.
8550 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
8552 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
8554 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
8555 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
8556 FRAME_POINTER_REGNUM point at high end of local var area.
8558 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
8560 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
8561 Move comment around. Do not reset best_edge for a copiable
8562 destination if the copy would cause a partition change.
8563 (better_edge_p): Remove redundant check.
8565 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
8567 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
8569 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
8571 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
8572 for math functions that have _Float<N> and _Float<N>X variants.
8573 (mathfn_built_in_2): Add support for math functions that have
8574 _Float<N> and _Float<N>X variants.
8575 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
8576 (expand_builtin_mathfn_ternary): Add support for fma with
8577 _Float<N> and _Float<N>X variants.
8578 (expand_builtin): Likewise.
8579 (fold_builtin_3): Likewise.
8580 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
8581 create math function _Float<N> and _Float<N>X variants as external
8583 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
8584 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
8585 the __builtin_ prefix and if not strict ansi, without the prefix.
8586 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
8587 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
8588 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
8589 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
8590 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
8591 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
8592 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
8593 function signatures for fma _Float<N> and _Float<N>X variants.
8594 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
8595 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
8596 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
8597 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
8598 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
8599 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
8600 * gencfn-macros.c (print_case_cfn): Add support for math functions
8601 that have _Float<N> and _Float<N>X variants.
8602 (print_define_operator_list): Likewise.
8603 (fltfn_suffixes): Likewise.
8605 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
8606 for math functions that have _Float<N> and _Float<N>X variants.
8607 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
8608 and _Float<N>X variants.
8609 (COPYSIGN): Likewise.
8612 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
8613 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
8615 (integer_valued_read_call_p): Likewise.
8616 * fold-const-call.c (fold_const_call_ss): Likewise.
8617 (fold_const_call_sss): Add support for copysign, fmin, and fmax
8618 _Float<N> and _Float<N>X variants.
8619 (fold_const_call_ssss): Add support for fma _Float<N> and
8620 _Float<N>X variants.
8621 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
8622 support for copysign and fma _Float<N> and _Float<N>X variants.
8623 (backprop::process_builtin_call_use): Likewise.
8624 * tree-call-cdce.c (can_test_argument_range); Add support for
8625 sqrt _Float<N> and _Float<N>X variants.
8626 (edom_only_function): Likewise.
8627 (get_no_error_domain): Likewise.
8628 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
8629 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
8630 copysign _Float<N> and _Float<N>X variants.
8631 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
8632 handled by machine independent code.
8633 (FMAF128): Likewise.
8634 * doc/cpp.texi (Common Predefined Macros): Document defining
8635 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
8636 fma _Float<N> and _Float<N>X variants.
8638 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
8641 * config/i386/i386-modes.def (CCFPU): Remove definition.
8642 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
8643 (ix86_cc_modes_compatible): Ditto.
8644 (ix86_expand_carry_flag_compare): Ditto.
8645 (ix86_expand_int_movcc): Ditto.
8646 (ix86_expand_int_addcc): Ditto.
8647 (ix86_reverse_condition): Ditto.
8648 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
8649 Return true/false for unordered/ordered fp comparisons.
8650 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
8651 (ix86_prepare_fp_compare_args): Update for rename.
8652 (ix86_expand_fp_compare): Update for rename. Generate unordered
8653 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
8654 (ix86_expand_sse_compare_and_jump): Ditto.
8655 * config/i386/predicates.md (fcmov_comparison_operator):
8656 Remove CCFPU mode handling.
8657 (ix86_comparison_operator): Ditto.
8658 (ix86_carry_flag_operator): Ditto.
8659 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
8660 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
8661 (*cmpu<mode>_cc_i387): Ditto.
8662 (FPCMP): Remove mode iterator.
8663 (unord): Remove mode attribute.
8664 (unord_subst): New define_subst transformation
8665 (unord): New define_subst attribute.
8667 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
8668 (*cmpi<unord>xf_i387): Ditto.
8669 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
8670 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
8671 using unord_subst transformation.
8672 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
8673 (round_saeonly): Also handle CCFP mode.
8674 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
8675 Remove UNSPEC_SAHF unspec handling.
8677 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
8679 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
8681 2017-10-27 Jeff Law <law@redhat.com>
8683 * gimple-ssa-sprintf.c: Include domwalk.h.
8684 (class sprintf_dom_walker): New class, derived from dom_walker.
8685 (sprintf_dom_walker::before_dom_children): New function.
8686 (struct call_info): Moved into sprintf_dom_walker class
8687 (compute_formath_length, handle_gimple_call): Likewise.
8688 (sprintf_length::execute): Call the dominator walker rather
8689 than walking the statements.
8691 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
8692 gimple statement locations.
8693 (check_array_bounds): Corresponding changes. Get the statement's
8694 location directly from wi->stmt.
8696 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
8699 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
8701 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
8703 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
8704 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
8706 2017-10-27 Jakub Jelinek <jakub@redhat.com>
8709 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
8710 * config/i386/i386.c (maybe_get_pool_constant): Removed.
8711 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
8712 maybe_get_pool_constant.
8713 * config/i386/predicates.md (zero_extended_scalar_load_operand):
8716 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8718 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
8719 2.26 caveat. Update gas and gld versions.
8720 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
8723 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
8725 * cgraph.h (set_malloc_flag): Declare.
8726 * cgraph.c (set_malloc_flag_1): New function.
8727 (set_malloc_flag): Likewise.
8728 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
8729 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
8731 (read_ipa_call_summary): Add support for reading is_return_callee.
8732 (write_ipa_call_summary): Stream is_return_callee.
8733 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
8734 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
8735 ipa-prop.h, ipa-fnsummary.h.
8736 (pure_const_names): Change to static.
8737 (malloc_state_e): Define.
8738 (malloc_state_names): Define.
8739 (funct_state_d): Add field malloc_state.
8740 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
8741 (check_retval_uses): New function.
8742 (malloc_candidate_p): Likewise.
8743 (analyze_function): Add support for malloc attribute.
8744 (pure_const_write_summary): Stream malloc_state.
8745 (pure_const_read_summary): Add support for reading malloc_state.
8746 (dump_malloc_lattice): New function.
8747 (propagate_malloc): New function.
8748 (warn_function_malloc): New function.
8749 (ipa_pure_const::execute): Call propagate_malloc and
8750 ipa_free_fn_summary.
8751 (pass_local_pure_const::execute): Add support for malloc attribute.
8752 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
8753 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
8755 2017-10-27 Martin Liska <mliska@suse.cz>
8757 PR gcov-profile/82457
8758 * doc/invoke.texi: Document that one needs a non-strict ISO mode
8759 for fork-like functions to be properly instrumented.
8761 2017-10-27 Richard Biener <rguenther@suse.de>
8764 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
8765 info when we redirected EH.
8767 2017-10-26 Michael Collison <michael.collison@arm.com>
8769 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
8771 (<optab>_trunchf<GPI:mode>2: New pattern.
8772 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
8773 * config/aarch64/iterators.md (wv): New mode attribute.
8774 (vf, VF): New mode attributes.
8775 (vgp, VGP): New mode attributes.
8776 (s): Update attribute with SImode and DImode prefixes.
8778 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
8780 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
8781 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
8782 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
8783 (nios2_option_overide): Initialize it. Don't allow R0-relative
8784 addressing with PIC.
8785 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
8786 (nios2_symbolic_constant_p): Likewise.
8787 (nios2_legitimate_address_p): Likewise.
8788 (nios2_r0rel_section_name_p): New.
8789 (nios2_symbol_ref_in_r0rel_data_p): New.
8790 (nios2_emit_move_sequence): Handle r0rel_constant_p.
8791 (r0rel_constant_p): New.
8792 (nios2_print_operand_address): Handle r0rel_constant_p.
8793 (nios2_cdx_narrow_form_p): Likewise.
8794 * config/nios2/nios2.opt (mr0rel-sec=): New option.
8795 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
8796 (Nios II Options): Document -mr0rel-sec.
8798 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
8800 * config/nios2/nios2.c: Include xregex.h.
8801 (nios2_gprel_sec_regex): New.
8802 (nios2_option_overide): Initialize it. Don't allow GP-relative
8803 addressing with PIC.
8804 (nios2_small_section_name_p): Check for regex match.
8805 * config/nios2/nios2.opt (mgprel-sec=): New option.
8806 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
8807 (Nios II Options): Document -mgprel-sec.
8809 2017-10-26 Jim Wilson <wilson@tuliptree.org>
8811 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
8813 2017-10-26 Tom de Vries <tom@codesourcery.com>
8815 PR tree-optimization/82707
8816 * gimple.c (gimple_copy): Fix unsharing of
8817 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
8819 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
8821 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
8822 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
8823 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
8824 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
8825 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
8826 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
8827 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
8828 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
8829 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
8830 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
8831 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
8832 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
8833 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
8834 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
8835 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
8836 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
8837 _mm512_mask_cmpunord_ps_mask): New intrinsics.
8839 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
8841 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
8843 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
8844 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
8845 warning to rs6000.c. Remove the Undocumented flag, since it has
8847 (-mabi=ibmlongdouble): Likewise.
8848 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
8849 already set, set the default format for long double.
8850 (rs6000_debug_reg_global): Print whether long double is IBM or
8852 (rs6000_option_override_internal): Rework setting long double
8853 format. Only warn if the user is changing the long double default
8854 and they did not use -Wno-psabi.
8855 * doc/invoke.texi (PowerPC options): Update the documentation for
8856 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
8858 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
8859 Alan Hayward <alan.hayward@arm.com>
8860 David Sherwood <david.sherwood@arm.com>
8862 * rtl.h (wider_subreg_mode): New function.
8863 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
8864 rather than an unsigned int *.
8865 * ira-color.c (regno_max_ref_width): Replace with...
8866 (regno_max_ref_mode): ...this new variable.
8867 (coalesced_pseudo_reg_slot_compare): Update accordingly.
8868 Use wider_subreg_mode.
8869 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
8870 rather than an unsigned int *.
8871 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
8872 (process_alt_operands): Likewise.
8873 (invariant_p): Likewise.
8874 * lra-spills.c (assign_mem_slot): Likewise.
8875 (add_pseudo_to_slot): Likewise.
8876 * lra.c (collect_non_operand_hard_regs): Likewise.
8877 (add_regs_to_insn_regno_info): Likewise.
8878 * reload1.c (regno_max_ref_width): Replace with...
8879 (regno_max_ref_mode): ...this new variable.
8880 (reload): Update accordingly. Update call to
8881 ira_sort_regnos_for_alter_reg.
8882 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
8883 (init_eliminable_invariants): Update to use regno_max_ref_mode.
8884 (scan_paradoxical_subregs): Likewise.
8886 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
8888 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
8889 (aarch64_frame): Add emit_frame_chain boolean.
8890 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
8891 Move eh_return case to aarch64_layout_frame.
8892 (aarch64_layout_frame): Initialize emit_frame_chain.
8893 (aarch64_expand_prologue): Use emit_frame_chain.
8895 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
8897 * config/aarch64/aarch64.c (aarch64_layout_frame):
8898 Ensure LR is always stored at the bottom of the callee-saves.
8899 Remove rarely used frame layout which saves callee-saves at top of
8900 frame, so the store of LR can be used as a valid probe in all cases.
8902 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
8904 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
8905 Improve unaligned TImode/TFmode base/offset split.
8907 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
8908 Alan Hayward <alan.hayward@arm.com>
8909 David Sherwood <david.sherwood@arm.com>
8911 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
8912 * combine.c (find_single_use_1): Likewise.
8913 (expand_field_assignment): Likewise.
8914 (move_deaths): Likewise.
8915 * lra-constraints.c (simplify_operand_subreg): Likewise.
8916 (curr_insn_transform): Likewise.
8917 * lra.c (collect_non_operand_hard_regs): Likewise.
8918 (add_regs_to_insn_regno_info): Likewise.
8919 * rtlanal.c (reg_referenced_p): Likewise.
8920 (covers_regno_no_parallel_p): Likewise.
8922 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
8924 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
8925 Don't print any bits outside the precision of the value.
8926 * wide-int.cc (test_printing): Add some new tests.
8928 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8930 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
8931 supports -xbrace_comment option.
8932 * configure: Regenerate.
8933 * config.in: Regenerate.
8934 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
8935 (ASM_CPU_SPEC): Use it.
8937 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
8939 * target.def (static_rtx_alignment): New hook.
8940 * targhooks.h (default_static_rtx_alignment): Declare.
8941 * targhooks.c (default_static_rtx_alignment): New function.
8942 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
8943 * doc/tm.texi: Regenerate.
8944 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
8945 instead of targetm.constant_alignment. Remove call to
8947 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
8948 (cris_preferred_mininum_alignment): New function, split out from...
8949 (cris_constant_alignment): ...here.
8950 (cris_static_rtx_alignment): New function.
8951 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
8953 (ix86_constant_alignment): ...here.
8954 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
8955 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
8956 (mmix_static_rtx_alignment): New function.
8957 * config/spu/spu.c (spu_static_rtx_alignment): New function.
8958 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
8960 2017-10-26 Tamar Christina <tamar.christina@arm.com>
8963 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
8964 Add flag_trapping_math and flag_fp_int_builtin_inexact.
8966 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
8968 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
8969 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
8970 mark as a sign-extending load.
8971 (local_pic_load_u): Define.
8973 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
8976 * fold-const.c (operand_equal_for_comparison_p): Also return true
8977 if ARG0 is a simple variant of ARG1 with narrower precision.
8978 (fold_ternary_loc): Always pass unstripped operands to the predicate.
8980 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
8982 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
8984 * i386.h (processor_costs): Add gather_static, gather_per_elt,
8985 scatter_static, scatter_per_elt.
8986 * x86-tune-costs.h: Add new cost entries.
8988 2017-10-25 Richard Biener <rguenther@suse.de>
8990 * tree-ssa-sccvn.h (vn_eliminate): Declare.
8991 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
8992 class pass_fre): Move to ...
8993 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
8994 class pass_fre): ... here and adjust for statistics.
8996 2017-10-25 Jakub Jelinek <jakub@redhat.com>
8999 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
9000 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
9001 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
9002 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
9005 2017-10-25 Richard Biener <rguenther@suse.de>
9007 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
9008 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
9009 eliminate_push_avail, eliminate_insert): Move inside...
9010 (class eliminate_dom_walker): ... this class in preparation
9012 (fini_eliminate): Remove by merging with ...
9013 (eliminate): ... this function. Adjust for class changes.
9014 (pass_pre::execute): Remove fini_eliminate call.
9015 (pass_fre::execute): Likewise.
9017 2017-10-24 Jakub Jelinek <jakub@redhat.com>
9020 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
9021 (VPERMI2, VPERMI2I): New mode iterators.
9022 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
9023 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
9025 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
9026 mode iterator. Remove 3 old define_insn patterns.
9027 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
9028 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
9029 VPERMI2 mode iterator, remove the other two expanders.
9030 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
9031 to use VPERMI2 mode iterator, add another alternative for vpermi2*
9032 instructions, remove the other two patterns.
9033 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
9034 mode iterator, remove the other two patterns.
9035 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
9036 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
9037 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
9038 and adjust argument order accordingly.
9039 (ix86_expand_vec_perm): Adjust caller.
9040 (expand_vec_perm_1): Likewise.
9041 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
9042 (expand_vec_perm_vpermt2_vpshub2): ... this.
9043 (ix86_expand_vec_perm_const_1): Adjust caller.
9044 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
9047 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
9048 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
9049 (vec_shl_<mode>): Remove unused expander.
9050 (avx512bw_<shift_insn><mode>3): New define_insn.
9051 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
9052 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
9054 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
9057 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
9060 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
9062 PR rtl-optimization/82396
9063 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
9064 (autopref_multipass_init): Simplify initialization.
9065 (autopref_rank_data): Simplify sort order.
9066 * gcc/sched-int.h (autopref_multipass_data_): Remove
9067 multi_mem_insn_p, min_offset and max_offset.
9069 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
9072 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
9073 Check special value of flag_omit_frame_pointer.
9074 (aarch64_can_eliminate): Likewise.
9075 (aarch64_override_options_after_change_1): Simplify handling of
9076 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
9078 2017-10-24 Richard Biener <rguenther@suse.de>
9080 PR tree-optimization/82697
9081 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
9082 zero for conditional load and unconditional store.
9084 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
9086 * doc/install.texi: Document bootstrap-cet.
9088 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
9091 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
9092 ENDBR instruction at function entrance if function is only
9095 2017-10-24 Jakub Jelinek <jakub@redhat.com>
9098 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
9099 patterns to better describe from which operation the CF is computed.
9100 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
9101 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
9102 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
9103 is 0, use _0 suffixed expanders instead of emitting a comparison
9106 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
9108 * config/i386/i386.md(*movsf_internal, *movdf_internal):
9109 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
9111 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
9114 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
9115 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
9116 * loop-iv.c (iv_get_reaching_def): Likewise.
9117 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
9118 variable is promoted and the partition contains undefined values.
9120 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9122 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
9124 (nios2_address_cost): Define.
9125 (nios2_legitimize_address): Recognize (exp + constant) directly.
9126 (TARGET_ADDRESS_COST): Define.
9128 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9130 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
9131 (nios2_symbolic_memory_operand_p): Declare.
9132 (nios2_split_large_constant): Declare.
9133 (nios2_split_symbolic_memory_operand): Declare.
9134 * config/nios2/nios2.c: Adjust includes.
9135 (nios2_symbolic_constant_allowed): New.
9136 (nios2_symbolic_constant_p): New.
9137 (nios2_plus_symbolic_constant_p): New.
9138 (nios2_valid_addr_expr_p): Recognize addresses involving
9140 (nios2_legitimate_address_p): Likewise, also LO_SUM.
9141 (nios2_symbolic_memory_operand_p): New.
9142 (nios2_large_constant_p): New.
9143 (nios2_split_large_constant): New.
9144 (nios2_split_plus_large_constant): New.
9145 (nios2_split_symbolic_memory_operand): New.
9146 (nios2_legitimize_address): Code refactoring. Handle addresses
9147 involving symbolic constants.
9148 (nios2_emit_move_sequence): Likewise.
9149 (nios2_print_operand): Improve error output.
9150 (nios2_print_operand_address): Handle LO_SUM.
9151 (nios2_cdx_narrow_form_p): Likewise.
9152 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
9153 operands involving symbolic constants.
9154 (movhi_internal, movsi_internal): Likewise.
9155 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
9156 (extendhisi2, extendqi<mode>2): Likewise.
9158 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9160 * tree-pass.h (PROP_rtl_split_insns): Define.
9161 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
9163 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9165 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
9167 2017-10-23 Jakub Jelinek <jakub@redhat.com>
9170 * target.def (const_not_ok_for_debug_p): Default to
9171 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
9172 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
9173 * targhooks.c (default_const_not_ok_for_debug_p): New function.
9174 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
9175 which targetm.const_not_ok_for_debug_p returned true.
9176 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
9178 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
9180 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
9181 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
9182 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
9184 (ix86_const_not_ok_for_debug_p): New function.
9185 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
9186 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
9188 2017-10-23 David Malcolm <dmalcolm@redhat.com>
9191 * system.h: Conditionally include "unique-ptr.h" if
9192 INCLUDE_UNIQUE_PTR is defined.
9193 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
9194 of defining INCLUDE_UNIQUE_PTR before including "system.h".
9196 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
9198 * config/rl78/rl78.md: New define_expand "subdi3".
9200 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
9203 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
9204 DF_REF_INSN if DF_REF_INSN_INFO is false.
9206 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
9208 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
9209 xmm_move instead of sse_move.
9210 (sse_store_index): New function.
9211 (ix86_register_move_cost): Be more sensible about mismatch stall;
9212 model AVX moves correctly; make difference between sse->integer and
9214 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
9215 moves; make difference between SSE and AVX.
9216 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
9217 and zmm_move. Increase size of sse load and store tables;
9218 add unaligned load and store tables; add ssemmx_to_integer.
9219 * x86-tune-costs.h: Update all entries according to real
9220 move latencies from Agner Fog's manual and chip documentation.
9222 2017-10-23 Jakub Jelinek <jakub@redhat.com>
9225 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
9226 * config/i386/constraints.md (Wf): New constraint.
9227 * config/i386/i386.md (UNSPEC_SBB): New unspec.
9228 (cmp<dwi>_doubleword): Removed.
9229 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
9230 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
9231 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
9232 expand with cmp<dwi>_doubleword. For LTU and GEU use
9233 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
9235 * common.opt (gcolumn-info): Enable by default.
9236 * doc/invoke.texi (gcolumn-info): Document new default.
9238 2017-10-23 Richard Biener <rguenther@suse.de>
9240 PR tree-optimization/82672
9241 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
9242 Fold the stmt if we propagated into it.
9244 2017-10-23 Richard Biener <rguenther@suse.de>
9246 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
9247 (bitmap_remove_expr_from_set): ... this. All callers call this
9248 for non-constant values.
9249 (bitmap_set_subtract): Rename to...
9250 (bitmap_set_subtract_expressions): ... this. Adjust and
9252 (bitmap_set_contains_value): Remove superfluous check.
9253 (bitmap_set_replace_value): Inline into single caller ...
9254 (bitmap_value_replace_in_set): ... here and simplify.
9255 (dependent_clean): Merge into ...
9256 (clean): ... this using an overload. Adjust.
9257 (prune_clobbered_mems): Adjust.
9258 (compute_antic_aux): Likewise.
9259 (compute_partial_antic_aux): Likewise.
9261 2017-10-23 Richard Biener <rguenther@suse.de>
9263 PR tree-optimization/82129
9265 2017-08-01 Richard Biener <rguenther@suse.de>
9267 PR tree-optimization/81181
9268 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
9269 (compute_antic): ... end of iteration here.
9271 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9273 * target.def (starting_frame_offset): New hook.
9274 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
9275 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
9276 * doc/tm.texi.in: Regenerate.
9277 * hooks.h (hook_hwi_void_0): Declare.
9278 * hooks.c (hook_hwi_void_0): New function.
9279 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
9280 STARTING_FRAME_OFFSET.
9281 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
9282 * reload1.c (reload): Likewise.
9283 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
9284 instead of STARTING_FRAME_OFFSET.
9285 * function.c (try_fit_stack_local): Likewise.
9286 (assign_stack_local_1): Likewise
9287 (instantiate_virtual_regs): Likewise.
9288 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
9289 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
9290 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
9291 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
9292 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
9293 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
9294 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
9295 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
9296 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
9297 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
9298 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
9299 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
9300 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
9301 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
9302 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
9303 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
9304 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
9305 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
9306 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
9307 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
9308 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
9309 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
9310 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
9311 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
9312 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
9313 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
9314 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
9315 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
9316 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
9317 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
9318 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
9319 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
9320 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
9321 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
9322 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
9323 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
9324 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
9325 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
9326 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
9327 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
9328 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
9329 * config/avr/avr.c (avr_starting_frame_offset): Make static and
9330 return a HOST_WIDE_INT.
9331 (avr_builtin_setjmp_frame_value): Use it instead of
9332 STARTING_FRAME_OFFSET.
9333 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9334 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
9335 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
9337 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9338 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
9339 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
9340 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9341 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
9342 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
9343 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9344 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
9345 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
9346 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9347 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
9348 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
9350 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9351 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
9352 * config/mips/mips.c (mips_compute_frame_info): Refer to
9353 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
9354 (mips_starting_frame_offset): New function.
9355 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9356 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
9357 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
9358 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
9359 and return a HOST_WIDE_INT.
9360 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9361 (mmix_initial_elimination_offset): Refer to
9362 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
9363 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
9364 * config/pa/pa.c (pa_starting_frame_offset): New function.
9365 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
9366 (pa_expand_prologue): Likewise.
9367 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9368 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
9369 !FRAME_GROWS_DOWNWARD handling to...
9370 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9371 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
9372 !FRAME_GROWS_DOWNWARD handling to...
9373 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9374 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
9375 !FRAME_GROWS_DOWNWARD handling to...
9376 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9377 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
9379 (rs6000_starting_frame_offset): New function.
9380 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
9381 !FRAME_GROWS_DOWNWARD handling to...
9382 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9383 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
9384 !FRAME_GROWS_DOWNWARD handling to...
9385 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9386 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
9387 !FRAME_GROWS_DOWNWARD handling to...
9388 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9389 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
9390 (rs6000_starting_frame_offset): New function.
9391 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
9392 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
9393 * config/vax/vax.c (vax_starting_frame_offset): New function.
9394 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
9395 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9396 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
9397 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
9398 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9399 * system.h (STARTING_FRAME_OFFSET): Poison.
9401 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9403 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
9404 SCALAR_TYPE_MODE instead of TYPE_MODE.
9406 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9407 Alan Hayward <alan.hayward@arm.com>
9408 David Sherwood <david.sherwood@arm.com>
9410 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
9412 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9413 Alan Hayward <alan.hayward@arm.com>
9414 David Sherwood <david.sherwood@arm.com>
9416 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
9418 2017-10-23 Richard Biener <rguenther@suse.de>
9420 PR tree-optimization/82129
9421 * tree-ssa-pre.c (bitmap_set_and): Remove.
9422 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
9423 canonicalizing expressions in the set to those with lowest
9424 ID rather than taking that from the first edge.
9426 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9428 * combine.c (rtx_equal_for_field_assignment_p): Use
9429 byte_lowpart_offset.
9431 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9432 Alan Hayward <alan.hayward@arm.com>
9433 David Sherwood <david.sherwood@arm.com>
9435 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
9436 to a SUBREG_PROMOTED_VAR.
9438 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9439 Alan Hayward <alan.hayward@arm.com>
9440 David Sherwood <david.sherwood@arm.com>
9442 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
9443 (expand_debug_source_expr): Likewise.
9444 * combine.c (combine_simplify_rtx): Likewise.
9445 * cse.c (fold_rtx): Likewise.
9446 * optabs.c (expand_float): Likewise.
9447 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9448 (simplify_binary_operation_1): Likewise.
9450 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9451 Alan Hayward <alan.hayward@arm.com>
9452 David Sherwood <david.sherwood@arm.com>
9454 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
9455 (record_promoted_value): Likewise.
9456 * expr.c (expand_expr_real_2): Likewise.
9457 * ree.c (update_reg_equal_equiv_notes): Likewise.
9458 (combine_set_extension): Likewise.
9459 * rtlanal.c (low_bitmask_len): Likewise.
9460 * simplify-rtx.c (neg_const_int): Likewise.
9461 (simplify_binary_operation_1): Likewise.
9463 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9464 Alan Hayward <alan.hayward@arm.com>
9465 David Sherwood <david.sherwood@arm.com>
9467 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
9468 * regcprop.c (maybe_mode_change): Likewise.
9469 * reload1.c (alter_reg): Likewise.
9471 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9473 * inchash.h (inchash::hash::add_wide_int): New function.
9474 * lto-streamer-out.c (hash_tree): Use it.
9476 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9478 * inchash.h (inchash::hash::add_wide_int): Rename to...
9479 (inchash::hash::add_hwi): ...this.
9480 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
9481 (polymorphic_call_target_hasher::hash): Likewise.
9482 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
9483 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
9484 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
9485 * lto-streamer-out.c (hash_tree): Likewise.
9486 * optc-save-gen.awk: Likewise.
9487 * tree.c (add_expr): Likewise.
9489 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
9492 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
9493 for ordered inequality comparisons even with TARGET_IEEE_FP.
9495 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
9498 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
9499 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
9500 Expand with cmp<dwi>_doubleword.
9502 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
9504 * extend.texi: Add x86 specific to 'nocf_check' attribute.
9505 List CET intrinsics.
9506 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
9507 specific to -fcf-protection option.
9509 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
9511 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
9512 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
9513 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
9514 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
9515 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
9516 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
9517 (extra_objs): Add cet.o for Linux/x86 targets.
9518 (tmake_file): Add i386/t-cet for Linux/x86 targets.
9519 * config/i386/cet.c: New file.
9520 * config/i386/cetintrin.h: Likewise.
9521 * config/i386/t-cet: Likewise.
9522 * config/i386/cpuid.h (bit_SHSTK): New.
9523 (bit_IBT): Likewise.
9524 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
9525 pass IBT and SHSTK bits.
9526 * config/i386/i386-builtin-types.def
9527 (VOID_FTYPE_UNSIGNED_PVOID): New.
9528 (VOID_FTYPE_UINT64_PVOID): Likewise.
9529 * config/i386/i386-builtin.def: Add CET intrinsics.
9530 * config/i386/i386-c.c (ix86_target_macros_internal): Add
9531 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
9532 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
9533 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
9535 * config/i386/i386.c (rest_of_insert_endbranch): New.
9536 (pass_data_insert_endbranch): Likewise.
9537 (pass_insert_endbranch): Likewise.
9538 (make_pass_insert_endbranch): Likewise.
9539 (ix86_notrack_prefixed_insn_p): Likewise.
9540 (ix86_target_string): Add -mibt, -mshstk flags.
9541 (ix86_option_override_internal): Add flag_cf_protection
9543 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
9544 (ix86_print_operand): Add 'notrack' prefix output.
9545 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
9546 (ix86_expand_builtin): Expand CET intrinsics.
9547 (x86_output_mi_thunk): Add 'endbranch' instruction.
9548 * config/i386/i386.h (TARGET_IBT): New.
9549 (TARGET_IBT_P): Likewise.
9550 (TARGET_SHSTK): Likewise.
9551 (TARGET_SHSTK_P): Likewise.
9552 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
9553 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
9554 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
9555 (builtin_setjmp_setup): New pattern.
9556 (builtin_longjmp): Likewise.
9557 (rdssp<mode>): Likewise.
9558 (incssp<mode>): Likewise.
9559 (saveprevssp): Likewise.
9560 (rstorssp): Likewise.
9561 (wrss<mode>): Likewise.
9562 (wruss<mode>): Likewise.
9563 (setssbsy): Likewise.
9564 (clrssbsy): Likewise.
9565 (nop_endbr): Likewise.
9566 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
9568 * config/i386/immintrin.h: Include <cetintrin.h>.
9569 * config/i386/linux-common.h
9570 (file_end_indicate_exec_stack_and_cet): New prototype.
9571 (TARGET_ASM_FILE_END): New.
9573 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
9575 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
9576 latencies instead of having separate table; make difference between
9577 integer and float costs.
9578 * i386.h (processor_costs): Remove scalar_stmt_cost,
9579 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
9580 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
9582 * x86-tune-costs.h: Remove entries which has been removed in
9583 procesor_costs from all tables; make cond_taken_branch_cost
9584 and cond_not_taken_branch_cost COST_N_INSNS based.
9586 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
9588 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
9590 2017-10-20 Jakub Jelinek <jakub@redhat.com>
9592 * config/i386/i386.md (isa): Remove fma_avx512f.
9593 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
9594 <avx512>_fmadd_<mode>_mask3<round_name>,
9595 <avx512>_fmsub_<mode>_mask<round_name>,
9596 <avx512>_fmsub_<mode>_mask3<round_name>,
9597 <avx512>_fnmadd_<mode>_mask<round_name>,
9598 <avx512>_fnmadd_<mode>_mask3<round_name>,
9599 <avx512>_fnmsub_<mode>_mask<round_name>,
9600 <avx512>_fnmsub_<mode>_mask3<round_name>,
9601 <avx512>_fmaddsub_<mode>_mask<round_name>,
9602 <avx512>_fmaddsub_<mode>_mask3<round_name>,
9603 <avx512>_fmsubadd_<mode>_mask<round_name>,
9604 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
9605 (*vec_widen_umult_even_v16si<mask_name>,
9606 *vec_widen_smult_even_v16si<mask_name>): Likewise.
9607 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
9609 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
9611 * extend.texi: Add 'nocf_check' documentation.
9612 * gimple.texi: Add second parameter to
9613 gimple_build_call_from_tree.
9614 * invoke.texi: Add -fcf-protection documentation.
9615 * rtl.texi: Add REG_CALL_NOTRACK documenation.
9617 2017-10-20 Richard Biener <rguenther@suse.de>
9619 PR tree-optimization/82473
9620 * tree-vect-loop.c (vectorizable_reduction): Properly get at
9621 the largest input type.
9623 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
9625 * c-attribs.c (handle_nocf_check_attribute): New function.
9626 (c_common_attribute_table): Add 'nocf_check' handling.
9627 * gimple-parser.c: Add second argument NULL to
9628 gimple_build_call_from_tree.
9629 * attrib.c (comp_type_attributes): Check nocf_check attribute.
9630 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
9632 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
9633 * common.opt: Add fcf-protection flag.
9634 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
9635 * flag-types.h: Add enum cf_protection_level.
9636 * gimple.c (gimple_build_call_from_tree): Add second parameter.
9637 Add 'nocf_check' attribute propagation to gimple call.
9638 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
9639 (gimple_build_call_from_tree): Update prototype.
9640 (gimple_call_nocf_check_p): New function.
9641 (gimple_call_set_nocf_check): Likewise.
9642 * gimplify.c: Add second argument to gimple_build_call_from_tree.
9643 * ipa-icf.c: Add nocf_check attribute in statement hash.
9644 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
9645 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
9646 * toplev.c (process_options): Add flag_cf_protection handling.
9648 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
9650 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
9652 2017-10-20 Richard Biener <rguenther@suse.de>
9654 PR tree-optimization/82603
9655 * tree-if-conv.c (predicate_mem_writes): Make sure to only
9656 remove false predicated stores.
9658 2017-10-20 Richard Biener <rguenther@suse.de>
9660 * graphite-isl-ast-to-gimple.c
9661 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
9662 Remove return value and simplify, dump copied stmt after lhs
9664 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
9665 Reduce dump verbosity.
9666 (gsi_insert_earliest): Likewise.
9667 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
9668 * graphite.c (print_global_statistics): Adjust dumping.
9669 (print_graphite_scop_statistics): Likewise.
9670 (print_graphite_statistics): Do not dump loops here.
9671 (graphite_transform_loops): But here.
9673 2017-10-20 Nicolas Roche <roche@adacore.com>
9675 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
9676 * configure: Regenerate.
9678 2017-10-20 Jakub Jelinek <jakub@redhat.com>
9681 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
9682 functions when optimizing replace GIMPLE_RETURN stmts with
9683 calls to __builtin_unreachable ().
9686 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
9687 for -fsanitize=thread link of executables.
9688 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
9689 link of executables.
9692 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
9694 (<shift_insn><mode>3<mask_name>): Change the last of the 3
9695 define_insns for logical vector shifts to use VI248_AVX512BW
9696 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
9697 condition, useless isa and prefix attributes. Change the first
9698 2 of these define_insns to ...
9699 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
9700 define_insn for avx512vl.
9701 (<shift_insn><mode>3): ... and this, new define_insn without
9702 masking for non-avx512vl.
9705 * config/i386/sse.md (*andnot<mode>3,
9706 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
9707 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
9708 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
9709 not applied use empty suffix even for TARGET_AVX512VL.
9710 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
9711 is applied, supply evex,evex or evex,evex,evex instead of just
9714 2017-10-20 Julia Koval <julia.koval@intel.com>
9716 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
9717 (OPTION_MASK_ISA_GFNI_UNSET): New.
9718 (ix86_handle_option): Handle OPT_mgfni.
9719 * config/i386/cpuid.h (bit_GFNI): New.
9720 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
9721 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
9722 * config/i386/i386.c (ix86_target_string): Add -mgfni.
9723 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
9724 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
9725 * config/i386/i386.opt: Add mgfni.
9727 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
9729 * config/msp430/msp430.c (msp430_option_override): Disable
9730 -fdelete-null-pointer-checks.
9731 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
9733 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
9735 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
9736 of x87 and SSE instructions.
9738 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
9740 * asan.c (create_cond_insert_point): Do not update edge count.
9741 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
9742 (afdo_propagate_circuit): Likewise.
9743 (afdo_calculate_branch_prob): Likewise.
9744 (afdo_annotate_cfg): Likewise.
9745 * basic-block.h (struct edge_def): Remove count.
9746 (edge_def::count): New accessor.
9747 * bb-reorder.c (rotate_loop): Update.
9748 (find_traces_1_round): Update.
9749 (connect_traces): Update.
9750 (sanitize_hot_paths): Update.
9751 * cfg.c (unchecked_make_edge): Update.
9752 (make_single_succ_edge): Update.
9753 (check_bb_profile): Update.
9754 (dump_edge_info): Update.
9755 (update_bb_profile_for_threading): Update.
9756 (scale_bbs_frequencies_int): Update.
9757 (scale_bbs_frequencies_gcov_type): Update.
9758 (scale_bbs_frequencies_profile_count): Update.
9759 (scale_bbs_frequencies): Update.
9760 * cfganal.c (connect_infinite_loops_to_exit): Update.
9761 * cfgbuild.c (compute_outgoing_frequencies): Update.
9762 (find_many_sub_basic_blocks): Update.
9763 * cfgcleanup.c (try_forward_edges): Update.
9764 (try_crossjump_to_edge): Update
9765 * cfgexpand.c (expand_gimple_cond): Update
9766 (expand_gimple_tailcall): Update
9767 (construct_exit_block): Update
9768 * cfghooks.c (verify_flow_info): Update
9769 (redirect_edge_succ_nodup): Update
9770 (split_edge): Update
9771 (make_forwarder_block): Update
9772 (duplicate_block): Update
9773 (account_profile_record): Update
9774 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
9775 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
9776 * cfgloopmanip.c (scale_loop_profile): Update.
9778 (lv_adjust_loop_entry_edge): Update.
9779 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
9780 (force_nonfallthru_and_redirect): Update.
9781 (purge_dead_edges): Update.
9782 (rtl_flow_call_edges_add): Update.
9783 * cgraphunit.c (init_lowered_empty_function): Update.
9784 (cgraph_node::expand_thunk): Update.
9785 * gimple-pretty-print.c (dump_probability): Update.
9786 (dump_edge_probability): Update.
9787 * gimple-ssa-isolate-paths.c (isolate_path): Update.
9788 * haifa-sched.c (sched_create_recovery_edges): Update.
9789 * hsa-gen.c (convert_switch_statements): Update.
9790 * ifcvt.c (dead_or_predicable): Update.
9791 * ipa-inline-transform.c (inline_transform): Update.
9792 * ipa-split.c (split_function): Update.
9793 * ipa-utils.c (ipa_merge_profiles): Update.
9794 * loop-doloop.c (add_test): Update.
9795 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
9796 * lto-streamer-in.c (input_cfg): Update.
9797 (input_function): Update.
9798 * lto-streamer-out.c (output_cfg): Update.
9799 * modulo-sched.c (sms_schedule): Update.
9800 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
9801 * predict.c (maybe_hot_edge_p): Update.
9802 (unlikely_executed_edge_p): Update.
9803 (probably_never_executed_edge_p): Update.
9804 (dump_prediction): Update.
9805 (drop_profile): Update.
9806 (propagate_unlikely_bbs_forward): Update.
9807 (determine_unlikely_bbs): Update.
9808 (force_edge_cold): Update.
9809 * profile.c (compute_branch_probabilities): Update.
9810 * reg-stack.c (better_edge): Update.
9811 * shrink-wrap.c (handle_simple_exit): Update.
9812 * tracer.c (better_p): Update.
9813 * trans-mem.c (expand_transaction): Update.
9814 (split_bb_make_tm_edge): Update.
9815 * tree-call-cdce.c: Update.
9816 * tree-cfg.c (gimple_find_sub_bbs): Update.
9817 (gimple_split_edge): Update.
9818 (gimple_duplicate_sese_region): Update.
9819 (gimple_duplicate_sese_tail): Update.
9820 (gimple_flow_call_edges_add): Update.
9821 (insert_cond_bb): Update.
9822 (execute_fixup_cfg): Update.
9823 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
9824 * tree-complex.c (expand_complex_div_wide): Update.
9825 * tree-eh.c (lower_resx): Update.
9826 (unsplit_eh): Update.
9827 (cleanup_empty_eh_move_lp): Update.
9828 * tree-inline.c (copy_edges_for_bb): Update.
9829 (freqs_to_counts): Update.
9830 (copy_cfg_body): Update.
9831 * tree-ssa-dce.c (remove_dead_stmt): Update.
9832 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
9833 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
9834 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
9835 (unloop_loops): Update.
9836 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
9837 * tree-ssa-loop-split.c (connect_loops): Update.
9838 (split_loop): Update.
9839 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
9840 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
9841 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
9842 * tree-ssa-reassoc.c (branch_fixup): Update.
9843 * tree-ssa-tail-merge.c (replace_block_by): Update.
9844 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
9845 (compute_path_counts): Update.
9846 (update_profile): Update.
9847 (recompute_probabilities): Update.
9848 (update_joiner_offpath_counts): Update.
9849 (estimated_freqs_path): Update.
9850 (freqs_to_counts_path): Update.
9851 (clear_counts_path): Update.
9852 (ssa_fix_duplicate_block_edges): Update.
9853 (duplicate_thread_path): Update.
9854 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
9855 (case_bit_test_cmp): Update.
9856 (collect_switch_conv_info): Update.
9857 (gen_inbound_check): Update.
9858 (do_jump_if_equal): Update.
9859 (emit_cmp_and_jump_insns): Update.
9860 * tree-tailcall.c (decrease_profile): Update.
9861 (eliminate_tail_call): Update.
9862 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
9863 (vect_do_peeling): Update.
9864 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
9865 * ubsan.c (ubsan_expand_null_ifn): Update.
9866 (ubsan_expand_ptr_ifn): Update.
9867 * value-prof.c (gimple_divmod_fixed_value): Update.
9868 (gimple_mod_pow2): Update.
9869 (gimple_mod_subtract): Update.
9870 (gimple_ic): Update.
9871 (gimple_stringop_fixed_value): Update.
9873 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
9876 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
9878 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
9880 PR rtl-optimization/82395
9881 * ira-color.c (allocno_priority_compare_func): Fix comparison step
9882 based on non_spilled_static_chain_regno_p.
9884 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
9886 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
9887 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
9888 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
9890 2017-10-19 Martin Sebor <msebor@redhat.com>
9892 PR tree-optimization/82596
9893 * tree.c (array_at_struct_end_p): Handle STRING_CST.
9895 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
9897 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
9898 (get_mem_refs_of_builtin_call): Likewise.
9899 * builtins.c (expand_builtin_apply): Adjust call to
9900 allocate_dynamic_stack_space.
9901 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
9902 the third argument to allocate_dynamic_stack_space, otherwise -1.
9903 (expand_builtin): Deal with all alloca variants.
9904 (is_inexpensive_builtin): Likewise.
9905 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
9906 * calls.c (special_function_p): Deal with all alloca variants.
9907 (initialize_argument_information): Adjust call to
9908 allocate_dynamic_stack_space.
9909 (expand_call): Likewise.
9910 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
9911 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
9912 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
9913 use it for the stack usage computation.
9914 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
9915 * function.c (gimplify_parameters): Call build_alloca_call_expr.
9916 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
9917 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
9918 (in_loop_p): Remove first argument and useless check.
9919 (pass_walloca::execute): Remove useless test and adjust call to above.
9920 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
9921 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
9922 (gimplify_call_expr): Deal with all alloca variants.
9923 * hsa-gen.c (gen_hsa_alloca): Likewise.
9924 (gen_hsa_insns_for_call): Likewise.
9925 * ipa-pure-const.c (special_builtin_state): Likewise.
9926 * tree-chkp.c (chkp_build_returned_bound): Likewise.
9927 * tree-object-size.c (alloc_object_size): Likewise.
9928 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
9929 (call_may_clobber_ref_p_1): Likewise.
9930 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
9931 (ccp_fold_stmt): Likewise.
9932 (optimize_stack_restore): Likewise.
9933 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
9934 (mark_all_reaching_defs_necessary_1): Likewise.
9935 (propagate_necessity): Likewise.
9936 (eliminate_unnecessary_stmts): Likewise.
9937 * tree.c (build_common_builtin_nodes): Build
9938 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
9939 (build_alloca_call_expr): New function.
9940 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
9941 (CASE_BUILT_IN_ALLOCA): Likewise.
9942 (build_alloca_call_expr): Declare.
9943 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
9945 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
9948 * dwarf2out.c (new_die_raw): New static inline function.
9949 (new_die): Use it to create the DIE.
9950 (add_AT_external_die_ref): Likewise.
9951 (clone_die): Likewise.
9952 (clone_as_declaration): Likewise.
9953 (dwarf2out_vms_debug_main_pointer): Likewise.
9954 (base_type_die): Likewise. Remove early return for corner cases.
9955 Do not call add_pubtype on the DIE here.
9956 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
9957 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
9958 typedefs for base types with DW_AT_endianity. Make sure a DIE with
9959 native order exists for base types, attach the DIE manually and call
9960 add_pubtype on it. Do not equate a reverse order DIE to the type.
9962 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
9964 * config/arm/arm.c (align_ok_ldrd_strd): New function.
9965 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
9967 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
9969 2017-10-19 Jakub Jelinek <jakub@redhat.com>
9971 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
9972 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
9973 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
9974 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
9975 * opts.c (sanitizer_opts): Add builtin.
9976 * ubsan.c (instrument_builtin): New function.
9977 (pass_ubsan::execute): Call it.
9978 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
9979 * doc/invoke.texi: Document -fsanitize=builtin.
9981 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
9982 builtins, store max (log2 (align), 0) into uchar field instead of
9983 align into uptr field.
9984 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
9985 store uchar 0 field instead of uptr 0 field.
9986 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
9987 instead of passing one address of struct with 2 locations pass
9988 two addresses of structs with 1 location each.
9989 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
9990 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
9991 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
9992 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
9993 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
9994 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
9995 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
9996 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
9998 2017-10-19 Martin Liska <mliska@suse.cz>
10001 * file-find.c (remove_prefix): Remove.
10002 * file-find.h (remove_prefix): Likewise.
10003 * gcc-ar.c: Remove smartness of lookup.
10005 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
10007 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
10008 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
10009 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
10011 2017-10-19 Jakub Jelinek <jakub@redhat.com>
10014 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
10015 (setcc + and to xor + setcc): New peephole2.
10017 2017-10-19 Tom de Vries <tom@codesourcery.com>
10019 * doc/sourcebuild.texi (Test Directives, Variants of
10020 dg-require-support): Add dg-require-stack-size.
10022 2017-10-19 Martin Liska <mliska@suse.cz>
10025 * gimplify.c (gimplify_decl_expr): Do not instrument variables
10026 that have a large alignment.
10027 (gimplify_target_expr): Likewise.
10029 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
10031 PR rtl-optimization/82602
10032 * ira.c (rtx_moveable_p): Return false for volatile asm.
10034 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
10037 * config/i386/i386-modes.def (CCGZ): New CC mode.
10038 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
10039 * config/i386/predicates.md (ix86_comparison_operator):
10041 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
10042 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
10043 with double-word subtraction.
10044 (put_condition_code): Handle CCGZmode.
10046 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
10048 * wide-int.cc (debug (const wide_int &)): New.
10049 (debug (const wide_int *)): New.
10050 (debug (const widest_int &)): New.
10051 (debug (const widest_int *)): New.
10053 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
10055 PR middle-end/82556
10056 * lra-constraints.c (curr_insn_transform): Use non-input operand
10057 instead of output one for matched reload.
10059 2017-10-18 Bin Cheng <bin.cheng@arm.com>
10061 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
10062 (tree-ssa-loop-ivopts.h): New header file.
10063 (struct builtin_info): New fields.
10064 (classify_builtin_1): Compute and record base and offset parts for
10065 memset builtin partition by calling strip_offset.
10066 (offset_cmp, fuse_memset_builtins): New functions.
10067 (finalize_partitions): Fuse adjacent memset partitions by calling
10069 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
10070 Expose the interface.
10071 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
10073 2017-10-18 Bin Cheng <bin.cheng@arm.com>
10075 PR tree-optimization/82574
10076 * tree-loop-distribution.c (find_single_drs): New parameter. Check
10077 that data reference must be executed exactly once per iteration
10078 against the outermost loop in nest.
10079 (classify_partition): Update call to above function.
10081 2017-10-18 Richard Biener <rguenther@suse.de>
10083 PR tree-optimization/82591
10084 * graphite.c (graphite_transform_loops): Move code gen message
10086 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
10087 Here. Handle scop_to_isl_ast failing.
10088 (scop_to_isl_ast): Limit the number of ISL operations.
10090 2017-10-18 Richard Biener <rguenther@suse.de>
10092 * graphite-isl-ast-to-gimple.c
10093 (translate_isl_ast_to_gimple::set_rename): Simplify.
10094 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
10095 (graphite_copy_stmts_from_block): ... here.
10096 (copy_bb_and_scalar_dependences): Simplify.
10097 (add_parameters_to_ivs_params): Canonicalize.
10098 (generate_entry_out_of_ssa_copies): Simplify.
10099 * graphite-sese-to-poly.c (extract_affine_name): Simplify
10100 by passing in ISL dimension.
10101 (parameter_index_in_region_1): Rename to ...
10102 (parameter_index_in_region): ... this.
10103 (extract_affine): Adjust assert, pass down parameter index.
10104 (add_param_constraints): Use range-info when available.
10105 (build_scop_context): Adjust.
10106 * sese.c (new_sese_info): Adjust.
10107 (free_sese_info): Likewise.
10108 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
10109 Remove unused typedefs.
10110 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
10112 2017-10-18 Martin Liska <mliska@suse.cz>
10114 * combine.c (simplify_compare_const): Add gcc_fallthrough.
10116 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
10118 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
10119 (s390_sched_init): Do not reset s390_sched_state if we entered the
10120 current basic block via a fallthru edge and all others are unlikely.
10122 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
10124 * config/s390/s390.c (NUM_SIDES): New variable.
10125 (LONGRUNNING_THRESHOLD): New variable.
10126 (LATENCY_FACTOR): New variable.
10127 (s390_sched_score): Decrease score for long-running instructions on
10129 (s390_sched_variable_issue): Perform bookkeeping for long-running
10132 2017-10-18 Richard Biener <rguenther@suse.de>
10134 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
10135 Simplify with removal of the parameter rename map.
10136 (set_rename): Likewise.
10137 (should_copy_to_new_region): Likewise.
10138 (graphite_copy_stmts_from_block): Likewise.
10139 (copy_bb_and_scalar_dependences): Remove initialization of
10140 unused copied_bb_map.
10141 (copy_def): Remove.
10142 (copy_internal_parameters): Likewise.
10143 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
10144 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
10145 Use INTEGRAL_TYPE_P.
10146 (parameter_index_in_region_1): Rename to ...
10147 (assign_parameter_index_in_region): ... this. Assert we have
10148 a parameter we handle.
10149 (scan_tree_for_params): Adjust.
10150 * sese.h (parameter_rename_map_t): Remove.
10151 (struct sese_info_t): Remove unused parameter_rename_map and
10152 copied_bb_map members.
10153 * sese.c (new_sese_info): Adjust.
10154 (free_sese_info): Likewise.
10156 2017-10-18 Martin Liska <mliska@suse.cz>
10159 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
10160 on an abnormal edge.
10162 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
10164 * doc/invoke.texi (ffunction-sections and fdata-sections):
10167 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
10169 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
10170 the use statement can throw internally.
10172 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
10174 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
10175 any RTX present on the RHS of a SET.
10176 * compare-elim.c (try_eliminate_compare): Restore comment.
10178 2017-10-17 Jakub Jelinek <jakub@redhat.com>
10180 * langhooks.h (struct lang_hooks): Document that tree_size langhook
10181 may be also called on tcc_type nodes.
10182 * langhooks.c (lhd_tree_size): Likewise.
10184 2017-10-17 David Malcolm <dmalcolm@redhat.com>
10186 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
10187 format_warning_at_substring.
10188 (maybe_warn): Convert source_range * param to a location_t. Pass
10189 UNKNOWN_LOCATION rather than NULL to fmtwarn.
10190 (format_directive): Remove code to extract source_ranges and
10191 source_range * in favor of just a location_t.
10192 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
10194 * substring-locations.c (format_warning_va): Convert
10195 source_range * param to a location_t.
10196 (format_warning_at_substring): Likewise.
10197 * substring-locations.h (format_warning_va): Likewise.
10198 (format_warning_at_substring): Likewise.
10200 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
10202 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
10204 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
10205 and scatter/gather ops.
10207 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
10208 vec_gather_load and vec_scatter_store.
10209 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
10210 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
10211 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
10212 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
10213 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
10214 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
10216 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
10218 * reg-stack.c (compare_for_stack_reg): Add bool argument.
10219 Detect FTST instruction and handle its register pops. Only pop
10220 second operand if can_pop_second_op is true.
10221 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
10222 set can_pop_second_op to false in the compare_for_stack_reg call.
10224 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
10225 output_fp_compare for stack register operands.
10226 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
10227 instructions here. Do not emit stack register pops here. Assert
10228 that FCOMPP pops next to top stack register. Rewrite function.
10230 2017-10-17 Nathan Sidwell <nathan@acm.org>
10232 PR middle-end/82577
10233 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
10234 use DECL_ASSEMBLER_NAME_RAW.
10236 PR middle-end/82546
10237 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
10240 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
10241 Wilco Dijkstra <wilco.dijkstra@arm.com>
10243 * builtins.c (expand_builtin_update_setjmp_buf): Add a
10244 converstion to Pmode from the buf_addr.
10246 2017-10-17 Richard Biener <rguenther@suse.de>
10248 * graphite-dependences.c (scop_get_reads_and_writes): Change
10249 output parameters to references.
10251 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
10253 PR 71026/tree-optimization
10254 * fold-const.c (distribute_real_division): Removed.
10255 (fold_binary_loc): Remove calls to distribute_real_divison.
10257 2017-10-17 Richard Biener <rguenther@suse.de>
10259 * graphite-scop-detection.c
10260 (scop_detection::stmt_has_simple_data_refs_p): Always use
10261 the full nest as region.
10262 (try_generate_gimple_bb): Likewise.
10263 * sese.c (scalar_evolution_in_region): Simplify now that
10264 SCEV can handle instantiation in regions.
10265 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
10266 in the non-loop part of a function if requested.
10268 2017-10-17 Richard Biener <rguenther@suse.de>
10270 PR tree-optimization/82563
10271 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
10273 (graphite_regenerate_ast_isl): Call it.
10274 * graphite-scop-detection.c (build_scops): Remove entry edge split.
10276 2017-10-17 Jakub Jelinek <jakub@redhat.com>
10278 PR tree-optimization/82549
10279 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
10280 Formatting fixes. Instead of calling make_bit_field_ref with negative
10283 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
10285 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
10286 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
10287 _mm_maskz_reduce_ss): New.
10288 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
10289 __builtin_ia32_reducess_mask): Ditto..
10290 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
10291 * config/i386/sse.md (reduces<mode>): Renamed to ...
10292 (reduces<mode><mask_scalar_name>): ... this.
10293 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
10295 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
10296 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
10298 2017-10-16 David Malcolm <dmalcolm@redhat.com>
10300 * Makefile.in (OBJS): Add unique-ptr-tests.o.
10301 * selftest-run-tests.c (selftest::run_tests): Call
10302 selftest::unique_ptr_tests_cc_tests.
10303 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
10304 * unique-ptr-tests.cc: New file.
10306 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
10309 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
10311 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
10312 (make_hard_regno_born, make_hard_regno_dead): Update
10313 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
10315 2017-10-16 Jeff Law <law@redhat.com>
10317 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
10319 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
10321 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
10323 2017-10-16 Olivier Hainque <hainque@adacore.com>
10325 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
10326 with_cpu if we were configured for an e500v2 target cpu name.
10328 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
10330 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
10331 * doc/invoke.texi: Document +nodsp as a valid extension for
10334 2017-10-16 Martin Liska <mliska@suse.cz>
10336 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
10337 (test_set_range): Likewise.
10338 (test_range_functions): Rename to ...
10339 (test_bit_in_range): ... this.
10340 (sbitmap_c_tests): Add new test.
10342 2017-10-16 Tamar Christina <tamar.christina@arm.com>
10344 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
10346 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
10347 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
10349 2017-10-16 Tamar Christina <tamar.christina@arm.com>
10351 * config/aarch64/aarch64-builtins.c
10352 (aarch64_types_quadopu_lane_qualifiers): New.
10353 (TYPES_QUADOPU_LANE): New.
10354 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
10355 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
10356 (aarch64_<sur>dot_laneq<vsi2qi>): New.
10357 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
10358 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
10359 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
10360 (Vdottype, DOTPROD): New.
10361 (sur): Add SDOT and UDOT.
10363 2017-10-16 Tamar Christina <tamar.christina@arm.com>
10365 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
10366 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
10367 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
10368 Add TARGET_DOTPROD.
10369 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
10370 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
10371 Enable TARGET_DOTPROD.
10372 (cortex-a75.cortex-a55): Likewise.
10373 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
10375 2017-10-16 Tamar Christina <tamar.christina@arm.com>
10377 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
10378 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
10380 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
10382 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
10383 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
10384 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
10385 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
10386 * config/arm/types.md (neon_dot, neon_dot_q): New.
10387 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
10389 2017-10-16 Tamar Christina <tamar.christina@arm.com>
10391 * config/arm/arm.h (TARGET_DOTPROD): New.
10392 * config/arm/arm.c (arm_arch_dotprod): New.
10393 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
10394 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
10395 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
10396 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
10397 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
10398 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
10399 * doc/invoke.texi (armv8.2-a): Document dotprod
10401 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
10403 * i386.c (ix86_vec_cost): New function.
10404 (ix86_rtx_costs): Handle vector operations better.
10405 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
10406 * x86-tune-costs.h: Add new costs to all tables.
10408 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
10410 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
10412 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
10413 divsd, sqrtss and sqrtsd
10414 * x86-tune-costs.h: Add new entries to all costs.
10415 (znver1_cost): Fix to match real instruction latencies.
10417 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10418 Michael Collison <michael.collison@arm.com>
10420 * compare-elim.c: Include emit-rtl.h.
10421 (can_merge_compare_into_arith): New function.
10422 (try_validate_parallel): Likewise.
10423 (try_merge_compare): Likewise.
10424 (try_eliminate_compare): Call the above when no previous clobber
10426 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
10429 2017-10-14 Jakub Jelinek <jakub@redhat.com>
10431 PR middle-end/62263
10432 PR middle-end/82498
10433 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
10434 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
10436 PR middle-end/62263
10437 PR middle-end/82498
10438 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
10439 to be any operand_equal_p operands. For & (B - 1) require
10440 B to be power of 2. Recognize
10441 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
10443 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
10446 * optabs.c (expand_memory_blockage): Fix call of
10447 targetm.have_memory_blockage.
10449 2017-10-14 Jakub Jelinek <jakub@redhat.com>
10452 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
10453 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
10454 objects to extra_objs instead of overwriting it.
10456 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
10458 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
10459 Use any_fp_register_operand as operand[3] predicate. Simplify
10460 equality test for operands[2] and operands[4] memory location.
10461 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
10462 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
10463 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
10464 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
10465 any_fp_register_operand as operand[1] predicate. Simplify
10466 equality test for operands[0] and operands[3] memory location.
10467 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
10468 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
10469 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
10471 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
10473 * target-insns.def: Add memory_blockage.
10474 * optabs.c (expand_memory_blockage): New function.
10475 (expand_asm_memory_barrier): Rename ...
10476 (expand_asm_memory_blockage): ... to this.
10477 (expand_mem_thread_fence): Call expand_memory_blockage
10478 instead of expand_asm_memory_barrier.
10479 (expand_mem_singnal_fence): Ditto.
10480 (expand_atomic_load): Ditto.
10481 (expand_atomic_store): Ditto.
10482 * doc/md.texi (Standard Pattern Names For Generation):
10483 Document memory_blockage instruction pattern.
10485 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
10487 * config/rl78/rl78.c (rl78_emit_libcall): New function.
10488 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
10489 * config/rl78/rl78.md: New define_expand "adddi3".
10491 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
10493 * cfghooks.c (verify_flow_info): Disable check that all probabilities
10496 2017-10-13 Jeff Law <law@redhat.com>
10498 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
10500 2017-10-13 Jakub Jelinek <jakub@redhat.com>
10503 * internal-fn.c (expand_mul_overflow): If both operands have
10504 the same highpart of -1 or 0 and the topmost bit of lowpart
10505 is different, overflow is if res <= 0 rather than res < 0.
10507 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
10509 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
10510 TARGET_P9_VECTOR code for unaligned_load case.
10512 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
10514 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
10516 2017-10-13 Nathan Sidwell <nathan@acm.org>
10518 * tree-core.h (tree_contains_struct): Make bool.
10519 * tree.c (tree_contains_struct): Likewise.
10520 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
10521 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
10522 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
10523 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
10524 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
10526 2017-10-13 Richard Biener <rguenther@suse.de>
10528 * graphite-isl-ast-to-gimple.c
10529 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
10530 parameters and dominance check.
10531 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
10532 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
10533 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
10534 Do not update SSA form here or do intermediate IL verification.
10535 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
10536 (graphite_initialize): Remove check on the number of loops in
10537 the function and inline into graphite_transform_loops.
10538 (graphite_finalize): Inline into graphite_transform_loops.
10539 (graphite_transform_loops): Perform SSA update and IL verification
10541 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
10543 2017-10-13 Richard Biener <rguenther@suse.de>
10545 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
10546 graphite_expression_type_precision): Avoid global constructor
10548 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
10549 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
10550 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
10551 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
10552 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
10553 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
10555 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
10558 * config/i386/i386.h (ix86_red_zone_size): New.
10559 * config/i386/i386.md (push peephole2s): Replace
10560 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
10562 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
10563 Alan Hayward <alan.hayward@arm.com>
10564 David Sherwood <david.sherwood@arm.com>
10566 * combine.c (can_change_dest_mode): Reject changes in
10567 REGMODE_NATURAL_SIZE.
10569 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
10570 Alan Hayward <alan.hayward@arm.com>
10571 David Sherwood <david.sherwood@arm.com>
10573 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
10574 (expand_debug_source_expr): Likewise.
10575 * combine.c (combine_simplify_rtx): Likewise.
10576 * cse.c (fold_rtx): Likewise.
10577 * fwprop.c (canonicalize_address): Likewise.
10578 * targhooks.c (default_shift_truncation_mask): Likewise.
10580 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
10581 Alan Hayward <alan.hayward@arm.com>
10582 David Sherwood <david.sherwood@arm.com>
10584 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
10585 (widened_mode): Likewise.
10586 (expand_unop): Likewise.
10587 * ree.c (transform_ifelse): Likewise.
10588 (merge_def_and_ext): Likewise.
10589 (combine_reaching_defs): Likewise.
10590 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10592 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
10593 Alan Hayward <alan.hayward@arm.com>
10594 David Sherwood <david.sherwood@arm.com>
10596 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
10597 * combine.c (gen_lowpart_for_combine): Likewise.
10598 * dwarf2out.c (rtl_for_decl_location): Likewise.
10599 * final.c (alter_subreg): Likewise.
10600 * rtlhooks.c (gen_lowpart_general): Likewise.
10601 (gen_lowpart_if_possible): Likewise.
10603 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
10604 Alan Hayward <alan.hayward@arm.com>
10605 David Sherwood <david.sherwood@arm.com>
10607 * calls.c (expand_call): Use subreg_lowpart_offset.
10608 * cse.c (cse_insn): Likewise.
10609 * regcprop.c (copy_value): Likewise.
10610 (copyprop_hardreg_forward_1): Likewise.
10612 2017-10-13 Jakub Jelinek <jakub@redhat.com>
10615 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
10616 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
10617 =Q constraints to +Q and into insn condition add check
10618 that operands[0] and operands[1] are equal.
10619 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
10620 =Q constraints to +Q and into insn condition add check
10621 that operands[0] is equal to either operands[1] or operands[2].
10624 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
10625 instead of handling MINUS_EXPR twice (once for each argument),
10626 canonicalize operand order and handle just once, use rtype where
10627 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
10630 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
10631 any values of __C while still being pattern recognizable as a simple
10632 rotate instruction.
10634 2017-10-13 Richard Biener <rguenther@suse.de>
10636 PR tree-optimization/82451
10638 2017-10-02 Richard Biener <rguenther@suse.de>
10640 PR tree-optimization/82355
10641 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
10642 a mapping for the enclosing loop but avoid generating one for
10643 the loop tree root.
10644 (copy_bb_and_scalar_dependences): Remove premature codegen
10645 error on PHIs in blocks duplicated into multiple places.
10646 * graphite-scop-detection.c
10647 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
10648 in the region use it as loop and nest to analyze the DR in.
10649 (try_generate_gimple_bb): Likewise.
10650 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
10651 (add_loop_constraints): For blocks in a loop not in the region
10652 create a dimension with a single iteration.
10653 * sese.h (gbb_loop_at_index): Remove assert.
10655 * cfgloop.c (loop_preheader_edge): For the loop tree root
10656 return the single successor of the entry block.
10657 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
10658 Reset the SCEV hashtable and niters.
10659 * graphite-scop-detection.c
10660 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
10661 assert that we only have POLYNOMIAL_CHREC that vary in loops
10662 contained in the region.
10663 (scop_detection::graphite_can_represent_expr): Adjust.
10664 (scop_detection::stmt_has_simple_data_refs_p): For loops
10665 not in the region set loop to NULL. The nest is now the
10666 entry edge to the region.
10667 (try_generate_gimple_bb): Likewise.
10668 * sese.c (scalar_evolution_in_region): Adjust for
10669 instantiate_scev change.
10670 * tree-data-ref.h (graphite_find_data_references_in_stmt):
10671 Make nest parameter the edge into the region.
10672 (create_data_ref): Likewise.
10673 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
10674 entry edge into a region and adjust instantiate_scev calls.
10675 (create_data_ref): Likewise.
10676 (graphite_find_data_references_in_stmt): Likewise.
10677 (find_data_references_in_stmt): Pass the loop preheader edge
10678 from the nest argument.
10679 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
10680 parameter the edge into the region.
10681 (instantiate_parameters): Use the loop preheader edge as entry.
10682 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
10684 (get_instantiated_value_entry): Make instantiate_below parameter
10685 the edge into the region.
10686 (instantiate_scev_name): Likewise. Adjust dominance checks,
10687 when we cannot use loop-based instantiation instantiate by
10688 walking use-def chains.
10689 (instantiate_scev_poly): Adjust.
10690 (instantiate_scev_binary): Likewise.
10691 (instantiate_scev_convert): Likewise.
10692 (instantiate_scev_not): Likewise.
10693 (instantiate_array_ref): Remove.
10694 (instantiate_scev_3): Likewise.
10695 (instantiate_scev_2): Likewise.
10696 (instantiate_scev_1): Likewise.
10697 (instantiate_scev_r): Do not blindly handle N-operand trees.
10698 Do not instantiate array-refs. Handle all constants and invariants.
10699 (instantiate_scev): Make instantiate_below parameter
10700 the edge into the region.
10701 (resolve_mixers): Use the loop preheader edge for the region
10702 parameter to instantiate_scev_r.
10703 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
10705 2017-10-13 Richard Biener <rguenther@suse.de>
10707 PR tree-optimization/82525
10708 * graphite-isl-ast-to-gimple.c
10709 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
10711 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
10712 Fail code generation when we cannot represent the isl integer.
10713 (binary_op_to_tree): Elide modulo operations that are no-ops
10714 in the type we code generate. Remove now superfluous code
10717 2017-10-13 Richard Biener <rguenther@suse.de>
10719 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
10720 (scop_detection::harmful_loop_in_region): Remove premature
10721 IV type restriction.
10722 (scop_detection::graphite_can_represent_scev): We can handle
10723 pointer IVs just fine.
10725 2017-10-13 Alan Modra <amodra@gmail.com>
10727 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
10728 "Clobbers and Scratch Registers". Add paragraph on
10729 alternative to clobbers for scratch registers and OpenBLAS
10732 2017-10-13 Alan Modra <amodra@gmail.com>
10734 * doc/extend.texi (Clobbers): Correct vax example. Delete old
10735 example of a memory input for a string of known length. Move
10736 commentary out of table. Add a number of new examples
10737 covering array memory inputs.
10739 2017-10-12 Martin Liska <mliska@suse.cz>
10741 PR tree-optimization/82493
10742 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
10743 (test_range_functions): New function.
10744 (sbitmap_c_tests): Likewise.
10745 * selftest-run-tests.c (selftest::run_tests): Run new tests.
10746 * selftest.h (sbitmap_c_tests): New function.
10748 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
10750 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
10752 * config/rs6000/amo.h: Fix spacing issue.
10754 2017-10-12 Jakub Jelinek <jakub@redhat.com>
10757 * config/i386/i386.md (*ashl<mode>3_mask_1,
10758 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
10759 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
10762 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
10764 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
10765 (profile_probability): Set max_probability
10766 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
10768 * profile-count.c (profile_probability::differs_from_p): Do not
10769 rely on max_probaiblity == 10000
10771 2017-10-12 Jeff Law <law@redhat.com>
10773 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
10776 2017-10-12 Martin Sebor <msebor@redhat.com>
10780 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
10781 (handle_alias_pairs): Call it.
10782 * common.opt (-Wattribute-alias): New option.
10783 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
10784 * doc/invoke.texi (-Wattribute-alias): Document.
10786 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
10789 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
10791 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
10793 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
10794 (make_hard_regno_born, make_hard_regno_dead): Update
10795 bb_killed_pseudos and bb_gen_pseudos.
10797 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
10799 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
10801 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
10803 * config/alpha/alpha.c (alpha_split_conditional_move):
10804 Use std::swap instead of manually swapping.
10805 (alpha_stdarg_optimize_hook): Ditto.
10806 (alpha_canonicalize_comparison): Ditto.
10808 2017-10-12 Bin Cheng <bin.cheng@arm.com>
10810 * tree-loop-distribution.c (struct builtin_info): New struct.
10811 (struct partition): Refactor fields into struct builtin_info.
10812 (partition_free): Free struct builtin_info.
10813 (build_size_arg_loc, build_addr_arg_loc): Delete.
10814 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
10815 information from struct builtin_info.
10816 (find_single_drs): New function refactored from classify_partition.
10817 Also moved builtin validity checks to this function.
10818 (compute_access_range, alloc_builtin): New functions.
10819 (classify_builtin_st, classify_builtin_ldst): New functions.
10820 (classify_partition): Refactor code into functions find_single_drs,
10821 classify_builtin_st and classify_builtin_ldst.
10822 (distribute_loop): Don't do runtime alias check when distributing
10824 (find_seed_stmts_for_distribution): New function.
10825 (pass_loop_distribution::execute): Refactor code finding seed
10826 stmts into above function. Support distribution for the innermost
10827 two-level loop nest. Adjust dump information.
10829 2017-10-12 Bin Cheng <bin.cheng@arm.com>
10831 * tree-loop-distribution.c: Adjust the general comment.
10832 (NUM_PARTITION_THRESHOLD): New macro.
10833 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
10834 (classify_partition): Skip builtin pattern of loop nest's inner loop.
10835 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
10836 in call to build_partition_graph.
10837 (finalize_partitions): New parameter. Make loop distribution more
10838 conservative by fusing more partitions.
10839 (distribute_loop): Don't do runtime alias check in case of loop nest
10841 (find_seed_stmts_for_distribution): New function.
10842 (prepare_perfect_loop_nest): New function.
10843 (pass_loop_distribution::execute): Refactor code finding seed stmts
10844 and loop nest into above functions. Support loop nest distribution.
10845 Adjust dump information accordingly.
10847 2017-10-12 Bin Cheng <bin.cheng@arm.com>
10849 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
10850 and set PTYPE_SEQUENTIAL for merged partition.
10852 2017-10-12 Richard Biener <rguenther@suse.de>
10854 PR tree-optimization/69728
10856 2017-09-19 Richard Biener <rguenther@suse.de>
10858 PR tree-optimization/69728
10859 * graphite-sese-to-poly.c (schedule_error): New global.
10860 (add_loop_schedule): Handle empty domain by failing the
10862 (build_original_schedule): Handle schedule_error.
10864 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
10865 domain by returning an unchanged schedule.
10867 2017-10-12 Jakub Jelinek <jakub@redhat.com>
10869 * genrecog.c (validate_pattern): For VEC_SELECT verify that
10870 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
10872 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
10874 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
10877 2017-10-12 Jakub Jelinek <jakub@redhat.com>
10880 * expr.c (store_field): Don't optimize away bitsize == 0 store
10881 from CALL_EXPR with addressable return type.
10883 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
10885 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
10886 * config/rs6000/rs6000.md (sel): Delete mode attribute.
10887 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
10888 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
10889 TARGET_ISEL instead of TARGET_ISEL<sel>.
10891 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
10893 * config/rs6000/rs6000.c
10894 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
10896 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
10898 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
10900 (reg_or_cint_operand): Fix comment.
10901 (reg_or_zero_operand): New predicate.
10902 * config/rs6000/rs6000-protos.h (output_isel): Delete.
10903 * config/rs6000/rs6000.c (output_isel): Delete.
10904 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
10905 instead of reg_or_cint_operand. Output instruction directly (not via
10907 (isel_unsigned_<mode>): Ditto.
10908 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
10909 gpc_reg_operand. Add an instruction alternative for this. Output
10910 instruction directly.
10911 (*isel_reversed_unsigned_<mode>): Ditto.
10913 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
10915 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
10916 (TARGET_CANONICALIZE_COMPARISON): Define.
10918 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
10921 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
10922 Check whether the dest is REG before adding REG_EQUIV note.
10924 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
10927 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
10929 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
10930 (make_hard_regno_born, make_hard_regno_dead): Update
10931 bb_killed_pseudos and bb_gen_pseudos.
10933 2017-10-11 Nathan Sidwell <nathan@acm.org>
10935 * incpath.h (enum incpath_kind): Name enum, prefix values.
10936 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
10937 * incpath.c (heads, tails): Use INC_MAX.
10938 (add_env_var_paths, add_standard_paths): Use incpath_kind.
10939 (merge_include_chains, split_quote_chain,
10940 register_include_chains): Update incpath_kind names.
10941 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
10942 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
10944 (add_framework_path, darwin_register_objc_includes): Likewise.
10945 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
10947 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
10949 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
10950 Do not use float_operator operator predicate.
10951 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
10952 * config/i386/predicates.md (float_operator): Remove predicate.
10954 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
10956 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
10957 (*jccxf_i387): Ditto.
10958 (*jcc<mode>_i387): Ditto.
10959 (*jccu<mode>_i387): Ditto.
10960 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
10961 (*jcc_*_i387 splitters): Remove.
10962 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
10963 * config/i386/i386.c (ix86_split_fp_branch): Remove.
10964 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
10967 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
10969 * profile-count.h (slow_safe_scale_64bit): New function.
10970 (safe_scale_64bit): New inline.
10971 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
10972 * profile-count.c: Include wide-int.h
10973 (slow_safe_scale_64bit): New.
10975 2017-10-11 Nathan Sidwell <nathan@acm.org>
10977 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
10978 HAS_DECL_ASSEMBLER_NAME_P.
10979 * gimple-expr.c (gimple_decl_printable_name: Check
10980 HAS_DECL_ASSEMBLER_NAME_P too.
10981 * ipa-utils.h (type_in_anonymous_namespace_p): Check
10982 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
10983 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
10984 * passes.c (rest_of_decl_compilation): Check
10985 HAS_DECL_ASSEMBLER_NAME_P too.
10986 * recog.c (verify_changes): Likewise.
10987 * tree-pretty-print.c (dump_decl_name): Likewise.
10988 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
10990 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
10991 (SET_DECL_ASSEMBLER_NAME): Use it.
10992 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
10993 (COPY_DECL_ASSEMBLER_NAME): Likewise.
10994 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
10996 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
10998 * config.gcc (i386, x86_64): Add extra objects.
10999 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
11000 (ix86_min_insn_size): Declare.
11001 (ix86_issue_rate): Declare.
11002 (ix86_adjust_cost): Declare.
11003 (ia32_multipass_dfa_lookahead): Declare.
11004 (ix86_macro_fusion_p): Declare.
11005 (ix86_macro_fusion_pair_p): Declare.
11006 (ix86_bd_has_dispatch): Declare.
11007 (ix86_bd_do_dispatch): Declare.
11008 (ix86_core2i7_init_hooks): Declare.
11009 (ix86_atom_sched_reorder): Declare.
11010 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
11011 (COSTS_N_BYTES): Move to x86-tune-costs.h.
11012 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
11013 (rip_relative_addr_p): Rename to ...
11014 (ix86_rip_relative_addr_p): ... this one; export.
11015 (memory_address_length): Update.
11016 (ix86_issue_rate): Move to x86-tune-sched.c.
11017 (ix86_flags_dependent): Move to x86-tune-sched.c.
11018 (ix86_agi_dependent): Move to x86-tune-sched.c.
11019 (exact_dependency_1): Move to x86-tune-sched.c.
11020 (exact_store_load_dependency): Move to x86-tune-sched.c.
11021 (ix86_adjust_cost): Move to x86-tune-sched.c.
11022 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
11023 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
11024 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
11025 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
11026 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
11027 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
11028 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
11029 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
11030 (min_insn_size): Rename to ...
11031 (ix86_min_insn_size): ... this one; export.
11032 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
11033 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
11034 (core2i7_first_cycle_multipass_backtrack): Move to
11035 x86-tune-sched-core.c.
11036 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
11037 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
11038 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
11039 (ix86_avoid_jump_mispredicts): Update.
11040 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
11041 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
11042 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
11043 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
11044 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
11045 (MAX_INSN): Move to ix86-tune-sched-bd.c.
11046 (MAX_IMM): Move to ix86-tune-sched-bd.c.
11047 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
11048 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
11049 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
11050 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
11051 (MAX_STORE): Move to ix86-tune-sched-bd.c.
11052 (BIG): Move to ix86-tune-sched-bd.c.
11053 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
11054 (enum insn_path): Move to ix86-tune-sched-bd.c.
11055 (get_mem_group): Move to ix86-tune-sched-bd.c.
11056 (is_cmp): Move to ix86-tune-sched-bd.c.
11057 (dispatch_violation): Move to ix86-tune-sched-bd.c.
11058 (is_branch): Move to ix86-tune-sched-bd.c.
11059 (is_prefetch): Move to ix86-tune-sched-bd.c.
11060 (init_window): Move to ix86-tune-sched-bd.c.
11061 (allocate_window): Move to ix86-tune-sched-bd.c.
11062 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
11063 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
11064 (process_end_window): Move to ix86-tune-sched-bd.c.
11065 (allocate_next_window): Move to ix86-tune-sched-bd.c.
11066 (find_constant): Move to ix86-tune-sched-bd.c.
11067 (get_num_immediates): Move to ix86-tune-sched-bd.c.
11068 (has_immediate): Move to ix86-tune-sched-bd.c.
11069 (get_insn_path): Move to ix86-tune-sched-bd.c.
11070 (get_insn_group): Move to ix86-tune-sched-bd.c.
11071 (count_num_restricted): Move to ix86-tune-sched-bd.c.
11072 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
11073 (add_insn_window): Move to ix86-tune-sched-bd.c.
11074 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
11075 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
11076 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
11077 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
11078 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
11079 (do_dispatch): Move to ix86-tune-sched-bd.c.
11080 (has_dispatch): Move to ix86-tune-sched-bd.c.
11081 * config/i386/t-i386: Add new object files.
11082 * config/i386/x86-tune-costs.h: New file.
11083 * config/i386/x86-tune-sched-atom.c: New file.
11084 * config/i386/x86-tune-sched-bd.c: New file.
11085 * config/i386/x86-tune-sched-core.c: New file.
11086 * config/i386/x86-tune-sched.c: New file.
11088 2017-10-11 Liu Hao <lh_mouse@126.com>
11090 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
11091 the generic version below instead.
11092 (should_colorize): Recognize Windows consoles as terminals
11094 * pretty-print.c [__MINGW32__] (write_all): New function.
11095 [__MINGW32__] (find_esc_head): Likewise.
11096 [__MINGW32__] (find_esc_terminator): Likewise.
11097 [__MINGW32__] (eat_esc_sequence): Likewise.
11098 [__MINGW32__] (mingw_ansi_fputs): New function that handles
11100 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
11103 2017-10-11 Richard Biener <rguenther@suse.de>
11105 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
11106 Properly call analyze_scalar_evolution with the loop of the stmt.
11108 2017-10-11 Richard Biener <rguenther@suse.de>
11110 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
11111 * tree-core.h (tree_base): Add chrec_var union member.
11112 * tree.h (CHREC_VAR): Remove.
11113 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
11114 * tree-chrec.h (build_polynomial_chrec): Adjust.
11115 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
11116 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
11118 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
11120 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
11121 * match.pd: ... here.
11122 ((T) X == (T) Y): Relax condition.
11124 2017-10-11 Bin Cheng <bin.cheng@arm.com>
11126 PR tree-optimization/82472
11127 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
11129 (break_alias_scc_partitions): Update postorder number.
11131 2017-10-11 Martin Liska <mliska@suse.cz>
11134 * opts.c (parse_no_sanitize_attribute): Do not use error_value
11136 * opts.h (parse_no_sanitize_attribute): Remove last argument.
11138 2017-10-11 Martin Liska <mliska@suse.cz>
11140 * print-rtl.c (print_insn): Move declaration of idbuf
11141 to same scope as name.
11143 2017-10-11 Martin Liska <mliska@suse.cz>
11148 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
11149 volatile arguments.
11151 2017-10-11 Martin Liska <mliska@suse.cz>
11154 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
11155 volatile arguments.
11157 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
11159 * config.gcc (default_gnu_indirect_function): Default to yes for
11160 arm*-*-linux* with glibc.
11162 2017-10-11 Richard Biener <rguenther@suse.de>
11164 * tree-scalar-evolution.c (get_scalar_evolution): Handle
11165 default-defs and types we do not want to analyze.
11166 (interpret_loop_phi): Replace unreachable code with an assert.
11167 (compute_scalar_evolution_in_loop): Remove and inline ...
11168 (analyze_scalar_evolution_1): ... here, replacing condition with
11169 what makes the intent clearer. Remove handling of cases
11170 get_scalar_evolution now handles.
11172 2017-10-10 Jim Wilson <wilson@tuliptree.org>
11174 PR rtl-optimization/81434
11175 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
11176 comment for main loop. In sched_group_found if, also add checks for
11177 pass and min_cost_group.
11179 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
11181 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
11182 (rs6000_insn_cost): New function.
11183 * config/rs6000/rs6000.md (cost): New attribute.
11185 2017-10-10 Jakub Jelinek <jakub@redhat.com>
11186 H.J. Lu <hongjiu.lu@intel.com>
11190 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
11191 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
11192 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
11193 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
11194 to OPTION_MASK_ISA_AVX512VL - builtins that have both
11195 OPTION_MASK_ISA_MMX and some other bit set require both
11196 mmx and the ISAs without the mmx bit.
11197 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
11198 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
11199 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
11200 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
11201 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
11202 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
11203 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
11204 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
11205 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
11206 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
11207 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
11208 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
11209 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
11210 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
11211 Add OPTION_MASK_ISA_MMX.
11213 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
11215 * config.gcc (armv7*-*-freebsd*): New target.
11216 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
11218 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
11220 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
11221 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
11224 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
11226 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
11227 a template parameter.
11228 (WIDE_INT_REF_FOR): Update accordingly.
11229 * tree.h (wi::int_traits <const_tree>): Delete.
11230 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
11231 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
11232 (wi::tree_to_wide_ref): New typedef.
11233 (wi::to_wide): New function.
11234 * calls.c (get_size_range): Use wi::to_wide when operating on
11235 trees as wide_ints.
11236 * cgraph.c (cgraph_node::create_thunk): Likewise.
11237 * config/i386/i386.c (ix86_data_alignment): Likewise.
11238 (ix86_local_alignment): Likewise.
11239 * dbxout.c (stabstr_O): Likewise.
11240 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
11241 * expr.c (const_vector_from_tree): Likewise.
11242 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
11243 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
11244 (fold_negate_expr_1, int_const_binop_1, const_binop)
11245 (fold_convert_const_int_from_real, optimize_bit_field_compare)
11246 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
11247 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
11248 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
11249 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
11250 (fold_not_const, round_up_loc): Likewise.
11251 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
11252 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
11253 (alloca_call_type): Likewise.
11254 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
11255 * godump.c (go_output_typedef): Likewise.
11256 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
11257 * internal-fn.c (get_min_precision): Likewise.
11258 * ipa-cp.c (ipcp_store_vr_results): Likewise.
11259 * ipa-polymorphic-call.c
11260 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
11261 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
11262 (ipa_modify_call_arguments): Likewise.
11263 * match.pd: Likewise.
11264 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
11265 * print-tree.c (print_node_brief, print_node): Likewise.
11266 * stmt.c (expand_case): Likewise.
11267 * stor-layout.c (layout_type): Likewise.
11268 * tree-affine.c (tree_to_aff_combination): Likewise.
11269 * tree-cfg.c (group_case_labels_stmt): Likewise.
11270 * tree-data-ref.c (dr_analyze_indices): Likewise.
11271 (prune_runtime_alias_test_list): Likewise.
11272 * tree-dump.c (dequeue_and_dump): Likewise.
11273 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
11274 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
11275 * tree-pretty-print.c (dump_generic_node): Likewise.
11276 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
11277 (simple_iv_with_niters): Likewise.
11278 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
11279 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
11280 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
11281 * tree-ssa-loop-niter.c (split_to_var_and_offset)
11282 (refine_value_range_using_guard, number_of_iterations_ne_max)
11283 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
11284 (get_cst_init_from_scev, record_nonwrapping_iv)
11285 (scev_var_range_cant_overflow): Likewise.
11286 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
11287 * tree-ssa-pre.c (compute_avail): Likewise.
11288 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
11289 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
11290 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
11291 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
11292 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
11293 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
11294 (dump_case_nodes, try_switch_expansion): Likewise.
11295 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
11296 (vect_do_peeling): Likewise.
11297 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
11298 * tree-vect-stmts.c (vectorizable_load): Likewise.
11299 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
11300 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
11301 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
11302 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
11303 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
11304 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
11305 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
11306 (evrp_dom_walker::before_dom_children): Likewise.
11307 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
11308 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
11309 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
11310 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
11311 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
11312 (get_range_pos_neg): Likewise.
11313 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
11314 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
11315 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
11316 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
11317 * config/avr/avr.c (avr_fold_builtin): Likewise.
11318 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
11319 * config/msp430/msp430.c (msp430_attr): Likewise.
11320 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
11321 * config/powerpcspe/powerpcspe-c.c
11322 (altivec_resolve_overloaded_builtin): Likewise.
11323 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
11324 (rs6000_expand_ternop_builtin): Likewise.
11325 * config/rs6000/rs6000-c.c
11326 (altivec_resolve_overloaded_builtin): Likewise.
11327 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
11328 (rs6000_expand_ternop_builtin): Likewise.
11329 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
11331 2017-10-10 Bin Cheng <bin.cheng@arm.com>
11333 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
11334 when copying loop nest with only one inner loop.
11336 2017-10-10 Richard Biener <rguenther@suse.de>
11338 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
11339 blocks if SCEV is active.
11340 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
11342 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
11343 (scev_initialize): Assert we are not yet initialized.
11345 2017-10-10 Bin Cheng <bin.cheng@arm.com>
11347 * tree-loop-distribution.c (generate_loops_for_partition): Remove
11348 inner loop's exit stmt by making it always exit the loop, otherwise
11349 we would generate an infinite empty loop.
11351 2017-10-10 Bin Cheng <bin.cheng@arm.com>
11353 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
11354 renaming variables in new preheader if it's deleted.
11356 2017-10-10 Bin Cheng <bin.cheng@arm.com>
11358 * tree-loop-distribution.c (struct partition): Remove unused field
11359 loops of the structure.
11360 (partition_alloc, partition_free): Ditto.
11361 (build_rdg_partition_for_vertex): Ditto.
11363 2017-10-09 Jeff Law <law@redhat.com>
11365 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
11366 return type to match prototype and documentation.
11368 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
11370 * config/rs6000/rs6000.c (processor_costs): Move to ...
11371 * config/rs6000/rs6000.h: ... here.
11372 (rs6000_cost): Declare.
11374 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
11376 * except.c (setjmp_fn): New global variable.
11377 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
11378 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
11379 if DONT_USE_BUILTIN_SETJMP is defined.
11381 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
11383 * target.def (insn_cost): New hook.
11384 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
11385 * doc/tm.texi: Regenerate.
11386 * rtlanal.c (insn_cost): Use the new hook.
11388 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
11390 * combine.c (combine_validate_cost): Compute the new insn_cost,
11391 not just pattern_cost.
11392 (try_combine): Adjust comment.
11394 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
11396 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
11398 * combine.c (uid_insn_cost): Adjust comment.
11399 (combine_validate_cost): Adjust comment. Use pattern_cost instead
11401 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
11402 * dse.c (find_shift_sequence): Ditto.
11403 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
11404 (bb_valid_for_noce_process_p): Use pattern_cost.
11405 * rtl.h (insn_rtx_cost): Delete.
11406 (pattern_cost): New prototype.
11407 (insn_cost): New prototype.
11408 * rtlanal.c (insn_rtx_cost): Rename to...
11409 (pattern_cost): ... this.
11412 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
11414 * config/i386/i386.md (*jcc_2): Remove insn pattern.
11415 (*jcc<mode>_0_r_i387): Ditto.
11416 (*jccxf_r_i387): Ditto.
11417 (*jcc<mode>_r_i387): Ditto.
11418 (*jccu<mode>_r_i387): Ditto.
11419 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
11420 (*jcc): Rename from *jcc_1.
11422 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11424 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
11425 deferred rescans after the lvx/stvx recombination pre-pass.
11427 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
11429 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
11430 memory operation instruction support.
11431 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
11432 (rs6000-ibm-aix[789]*): Likewise.
11433 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
11434 Document new functions.
11436 2017-10-09 Richard Biener <rguenther@suse.de>
11438 PR tree-optimization/82397
11439 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
11440 equality only for semantically equal trees.
11442 2017-10-09 Richard Biener <rguenther@suse.de>
11444 PR tree-optimization/82449
11445 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
11446 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
11447 allow constant addresses.
11448 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
11451 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11453 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
11456 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11459 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
11462 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11465 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
11467 2017-10-09 Jakub Jelinek <jakub@redhat.com>
11470 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
11471 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
11473 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
11475 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
11476 (WI_BINARY_PREDICATE_RESULT): Likewise.
11477 (wi::binary_traits::operator_result): New type.
11478 (wi::binary_traits::predicate_result): Likewise.
11479 (generic_wide_int::operator~, unary generic_wide_int::operator-)
11480 (generic_wide_int::operator==, generic_wide_int::operator!=)
11481 (generic_wide_int::operator&, generic_wide_int::and_not)
11482 (generic_wide_int::operator|, generic_wide_int::or_not)
11483 (generic_wide_int::operator^, generic_wide_int::operator+
11484 (binary generic_wide_int::operator-, generic_wide_int::operator*):
11486 (operator~, unary operator-, operator==, operator!=, operator&)
11487 (operator|, operator^, operator+, binary operator-, operator*): New
11489 * expr.c (get_inner_reference): Use wi::bit_and_not.
11490 * fold-const.c (fold_binary_loc): Likewise.
11491 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
11492 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
11493 (bit_value_binop): Likewise.
11494 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
11495 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
11496 (extract_range_from_binary_expr_1): Likewise.
11497 (masked_increment): Likewise.
11498 (simplify_bit_ops_using_ranges): Likewise.
11500 2017-10-09 Martin Jambor <mjambor@suse.cz>
11503 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
11504 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
11505 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
11507 (hsa_fixup_mov_insn_type): New function.
11508 (hsa_op_with_type::get_in_type): Use it.
11509 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
11510 immediates in an assert.
11511 (hsa_op_with_type::extend_int_to_32bit): New method.
11512 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
11513 types. Convert to dest type if necessary.
11514 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
11515 (reg_for_gimple_ssa): Pass false as min32int to
11516 hsa_type_for_scalar_tree_type.
11517 (gen_hsa_addr): Fixup type when creating addresable temporary.
11518 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
11519 (gen_hsa_unary_operation): Extend operands and convert to dest type if
11520 necessary. Call hsa_fixup_mov_insn_type.
11521 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
11522 extend operands and convert to dest type if necessary.
11523 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
11524 to dest type if necessary.
11525 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
11526 if conversion nt necessary and size matches.
11527 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
11528 to dest type if necessary.
11529 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
11530 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
11532 (gen_hsa_clrsb): Likewise.
11533 (gen_hsa_ffs): Likewise.
11534 (gen_hsa_divmod): Extend operands and convert to dest type if
11536 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
11538 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
11540 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
11541 Remove empty default arguments. Use a brace block as output
11543 (conditional return): Ditto.
11545 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
11546 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
11548 (group_ending_nop): Ditto.
11549 (doloop_end): Ditto.
11550 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
11551 (splitters for those): Ditto.
11553 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
11555 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
11556 a conditional jump (and the compare for it) so that pc_rtx is the
11558 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
11559 for the deleted and renamed ctr<mode>_internal[234] patterns.
11560 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
11561 Delete second conditional return pattern.
11562 (ctr<mode>_internal2): Delete this second bdnz pattern.
11563 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
11564 (ctr<mode>_internal4): Delete this second bdz pattern.
11566 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
11568 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
11569 (always_initialized_rtx_for_ssa_name_p): New predicate.
11570 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
11571 (finish_out_of_ssa): Free new field of SA.
11572 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
11573 * tree-ssa-coalesce.c: Include tree-ssa.h.
11574 (get_parm_default_def_partitions): Remove extern keyword.
11575 (get_undefined_value_partitions): New function.
11576 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
11577 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
11578 uninitialized bits.
11579 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
11581 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
11583 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
11585 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
11587 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
11588 for targets that preffer 128bit.
11590 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
11592 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
11594 2017-10-08 Olivier Hainque <hainque@adacore.com>
11596 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
11597 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
11599 (thumb_set_return_address): Likewise.
11601 2017-10-08 Olivier Hainque <hainque@adacore.com>
11603 * common/config/arm/arm-common.c (arm_except_unwind_info):
11604 Handle DWARF2_UNWIND_INFO.
11606 2017-10-07 Michael Collison <michael.collison@arm.com>
11608 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
11611 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
11613 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
11614 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
11615 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
11616 defined, force the creation of a new block for a dispatch label.
11618 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
11620 * invoke.texi (Wsuggest-attribute=cold): Document.
11621 * common.opt (Wsuggest-attribute=cold): New
11622 * ipa-pure-const.c (warn_function_cold): New function.
11623 * predict.c (compute_function_frequency): Use it.
11624 * predict.h (warn_function_cold): Declare.
11626 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
11628 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
11631 2017-10-06 Martin Liska <mliska@suse.cz>
11633 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
11634 keyword for member functions.
11635 (struct sanopt_tree_couple): New struct.
11636 (struct sanopt_tree_couple_hash): New function.
11637 (struct sanopt_ctx): Add new hash_map.
11638 (has_dominating_ubsan_ptr_check): New function.
11639 (record_ubsan_ptr_check_stmt): Likewise.
11640 (maybe_optimize_ubsan_ptr_ifn): Likewise.
11641 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
11642 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
11644 2017-10-06 Sudakshina Das <sudi.das@arm.com>
11647 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
11648 aarch64_simd_valid_immediate on CONST_VECTORs.
11649 (aarch64_reg_or_bic_imm): Likewise.
11651 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
11653 PR rtl-optimization/82396
11654 * haifa-sched.c (ready_sort_real): Disable qsort checking.
11656 2017-10-06 Sebastian Pop <sebpop@gmail.com>
11658 * graphite-dependences.c (scop_get_reads): Move code to...
11659 (scop_get_must_writes): Move code to...
11660 (scop_get_may_writes): Move code to...
11661 (scop_get_reads_and_writes): ... here.
11662 (scop_get_dependences): Call scop_get_reads_and_writes.
11664 2017-10-06 Jakub Jelinek <jakub@redhat.com>
11666 PR tree-optimization/82434
11667 * fold-const.h (can_native_encode_type_p,
11668 can_native_encode_string_p): Remove.
11669 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
11670 don't encode anything, just return what would be otherwise returned.
11671 (native_encode_fixed, native_encode_complex, native_encode_vector):
11673 (native_encode_string): Likewise. Inline by hand
11674 can_native_encode_string_p.
11675 (can_native_encode_type_p): Remove.
11676 (can_native_encode_string_p): Remove.
11677 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
11678 STRING_CSTs using can_native_encode_string_p, test all
11679 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
11680 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
11681 argument from native_encode_expr.
11682 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
11683 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
11686 2017-10-06 Richard Biener <rguenther@suse.de>
11688 PR tree-optimization/82397
11689 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
11690 operand_equal_p but rely on data_ref_compare_tree for detecting
11692 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
11693 to match up with dr_group_sort_cmp.
11695 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11698 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
11700 * config/s390/s390-builtin-types.def: Regenerate.
11702 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11705 * config/s390/s390-builtin-types.def: Regenerate.
11706 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
11707 Change flag from B_VXE to B_VX.
11708 (s390_vec_min_dbl): Remove B_VXE flag.
11710 2017-10-06 Richard Biener <rguenther@suse.de>
11712 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
11713 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
11714 translate_isl_ast_to_gimple::is_valid_rename,
11715 translate_isl_ast_to_gimple::get_rename,
11716 translate_isl_ast_to_gimple::get_def_bb_for_const,
11717 translate_isl_ast_to_gimple::get_new_name,
11718 translate_isl_ast_to_gimple::collect_all_ssa_names,
11719 translate_isl_ast_to_gimple::copy_loop_phi_args,
11720 translate_isl_ast_to_gimple::collect_all_ssa_names,
11721 translate_isl_ast_to_gimple::copy_loop_phi_args,
11722 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
11723 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
11724 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
11725 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
11726 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
11727 translate_isl_ast_to_gimple::copy_cond_phi_args,
11728 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
11729 translate_isl_ast_to_gimple::edge_for_new_close_phis,
11730 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
11731 translate_isl_ast_to_gimple::rename_uses,
11732 translate_isl_ast_to_gimple::rename_all_uses): Remove.
11733 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
11734 (set_rename_for_each_def): Likewise.
11735 (graphite_copy_stmts_from_block): Handle debug stmt resetting
11736 here. Handle rewriting SCEV analyzable uses here.
11737 (copy_bb_and_scalar_dependences): Generate code for PHI
11739 (graphite_regenerate_ast_isl): Adjust.
11740 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
11741 (add_write, add_read): New functions.
11742 (build_cross_bb_scalars_def): Use it and simplify.
11743 (build_cross_bb_scalars_use): Likewise.
11744 (graphite_find_cross_bb_scalar_vars): Inline into...
11745 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
11746 simulating out-of-SSA. Compute liveout and add dependencies.
11747 (build_scops): Force an empty entry block.
11748 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
11750 (sese_build_liveouts): Declare.
11751 (sese_trivially_empty_bb_p): Likewise.
11752 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
11753 compute liveout and debug_liveout.
11754 (sese_bad_liveouts_use): Remove.
11755 (sese_reset_debug_liveouts_bb): Likewise.
11756 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
11757 (sese_build_liveouts): Build liveout and debug_liveout and store
11759 (new_sese_info): Adjust.
11760 (free_sese_info): Likewise.
11761 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
11762 do not build liveout here.
11763 (move_sese_in_condition): Adjust region entry.
11764 (scev_analyzable_p): Match up with chrec_apply requirements.
11765 (sese_trivially_empty_bb_p): New.
11766 * tree-into-ssa.c (get_reaching_def): Properly support generating
11767 default-defs for incremental rewrite of anonymous names.
11769 2017-10-06 Richard Biener <rguenther@suse.de>
11771 * graphite-sese-to-poly.c (extract_affine): For casts increasing
11772 precision do not perform modulo reduction.
11774 2017-10-06 Richard Biener <rguenther@suse.de>
11776 PR tree-optimization/82436
11777 * tree-vect-slp.c (vect_supported_load_permutation_p): More
11778 conservatively choose the vectorization factor when checking
11779 whether we can perform the required load permutation.
11780 (vect_transform_slp_perm_load): Assert when we may not fail.
11782 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
11784 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
11785 message for incompatible -msdata=* and -mcall-* options.
11787 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
11789 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
11790 rate for post-reload scheduling.
11792 2017-10-05 Tamar Christina <tamar.christina@arm.com>
11794 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
11796 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
11798 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
11799 to improve monte carlo in scimark.
11801 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
11803 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
11804 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
11805 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
11806 pentium4_cost, nocona_cost): Set reassociation width to 1.
11807 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
11808 width to 2 for fp operations and 1 otherwise.
11809 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
11811 (atom_cost): Set reassociation width to 2.
11812 (slm_cost, generic_cost): Set fp reassociation width
11813 to 2 and 1 otherwise.
11814 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
11815 (core_cost): Set fp reassociation width to 4 and vector to 2.
11816 (ix86_reassociation_width): Rewrite using cost table; special case
11817 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
11818 and TARGET_AVX128_OPTIMAL.
11819 * config/i386/i386.h (processor_costs): Add
11820 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
11821 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
11822 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
11823 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
11824 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
11825 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
11827 2017-10-05 Nathan Sidwell <nathan@acm.org>
11829 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
11831 2017-10-05 Tamar Christina <tamar.christina@arm.com>
11833 * config/arm/arm.c (arm_test_fpu_data): New.
11834 (arm_run_selftests): Call arm_test_fpu_data.
11836 2017-10-04 Nathan Sidwell <nathan@acm.org>
11838 * toplev.c (toplev::main): Remove excess parens on pretty_printer
11840 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
11842 2017-10-04 Sudakshina Das <sudi.das@arm.com>
11844 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
11845 check type for aarch64_simd_valid_immediate.
11846 (aarch64_output_simd_mov_immediate): Update prototype.
11847 (aarch64_simd_valid_immediate): Update prototype.
11848 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
11849 support for ORR-immediate.
11850 (and<mode>3): modified pattern to add support for BIC-immediate.
11851 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
11852 now checks for valid immediate for BIC and ORR based on new enum
11854 (aarch64_output_simd_mov_immediate): Function now used to output
11855 BIC/ORR imm as well based on new enum argument.
11856 * config/aarch64/constraints.md (Do): New vector immediate constraint.
11858 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
11859 (aarch64_reg_or_bic_imm): Likewise.
11861 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11863 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
11864 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
11866 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
11870 PR rtl-optimization/82396
11871 * haifa-sched.c (autopref_multipass_init): Simplify
11873 (autopref_rank_data): Simplify sort order.
11874 * sched-int.h (autopref_multipass_data_): Remove
11875 multi_mem_insn_p, min_offset and max_offset.
11877 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11879 * doc/sourcebuild.texi: Document vect_peeling_profitable.
11881 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11883 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
11884 vect_doubleint_cvt.
11886 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11888 * doc/sourcebuild.texi: Document vect_long_mult.
11890 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
11892 PR tree-optimization/82413
11893 * fold-const.c (build_range_check): Use widest_int when comparing
11894 the maximum ETYPE value with HIGH.
11896 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
11898 PR rtl-optimization/82396
11899 * haifa-sched.c (autopref_multipass_init): Simplify
11901 (autopref_rank_data): Simplify sort order.
11902 * sched-int.h (autopref_multipass_data_): Remove
11903 multi_mem_insn_p, min_offset and max_offset.
11905 2017-10-04 Jakub Jelinek <jakub@redhat.com>
11907 PR tree-optimization/82381
11908 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
11909 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
11912 PR tree-optimization/82374
11913 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
11914 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
11915 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
11916 current_function_decl to the new decl.
11918 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
11920 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
11921 helper macro for IEEE float128 hardware built-in functions.
11922 (SQRTF128_ODD): Add built-in functions with the round-to-odd
11924 (TRUNCF128_ODD): Likewise.
11925 (ADDF128_ODD): Likewise.
11926 (SUBF128_ODD): Likewise.
11927 (MULF128_ODD): Likewise.
11928 (DIVF128_ODD): Likewise.
11929 (FMAF128_ODD): Likewise.
11930 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
11931 UNSPEC_TRUNC_ROUND_TO_ODD.
11932 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
11933 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
11934 floating point round to odd instructions.
11935 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
11936 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
11937 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
11938 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
11939 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
11940 (trunc<mode>sf2_hw): Change the truncate with round to odd
11941 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
11942 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
11943 to odd hardware instructions.
11944 (sub<mode>3_odd): Likewise.
11945 (mul<mode>3_odd): Likewise.
11946 (div<mode>3_odd): Likewise.
11947 (sqrt<mode>2_odd): Likewise.
11948 (fma<mode>4_odd): Likewise.
11949 (fms<mode>4_odd): Likewise.
11950 (nfma<mode>4_odd): Likewise.
11951 (nfms<mode>4_odd): Likewise.
11952 (trunc<mode>df2_odd): Change the truncate with round to odd
11953 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
11955 * doc/extend.texi (PowerPC built-in functions): Update documentation
11956 for existing IEEE float128-bit built-in functions. Add built-in
11957 functions that generate the IEEE 128-bit floating point round to
11960 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
11962 PR rtl-optimization/77729
11963 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
11964 to (X&(C1&~C2))|C2 transformations.
11966 2017-10-03 Martin Jambor <mjambor@suse.cz>
11968 PR tree-optimization/82363
11969 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
11970 mismatch, mark lacc written regardless of racc.
11972 2017-10-03 Jakub Jelinek <jakub@redhat.com>
11974 PR tree-optimization/82381
11975 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
11976 stmt_to_insert nor wheather SSA_NAMEs are default defs.
11977 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
11978 fallthrough into reassoc_stmt_dominates_stmt_p.
11981 * combine.c (combine_instructions): Don't combine in unreachable
11984 2017-10-02 Peter Bergner <bergner@vnet.ibm.com>
11987 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
11988 function to not use the have_cpu variable. Do not set cpu_index,
11989 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
11990 or the default cpu.
11991 (rs6000_valid_attribute_p): Remove duplicate initializations of
11992 old_optimize and func_optimize.
11993 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
11994 (rs6000_activate_target_options): Make global.
11995 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
11998 2017-10-02 Jakub Jelinek <jakub@redhat.com>
12000 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
12001 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
12002 Set *poffset to 0 and *psize and *pmax_size to -1 if
12003 *poffset + *psize overflows in HOST_WIDE_INT.
12005 PR tree-optimization/82387
12006 PR tree-optimization/82388
12007 PR tree-optimization/82389
12008 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
12009 instead of live_bytes non-NULL.
12011 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
12014 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
12017 2017-10-02 Richard Biener <rguenther@suse.de>
12019 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
12020 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
12021 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
12023 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
12025 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
12026 requested precision matches the type's.
12027 * calls.c (alloc_max_size): Calculate the new candidate size as
12028 a widest_int and use wi::to_widest when comparing it with the
12029 current candidate size.
12030 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
12031 zero rather than integer_zero_node.
12032 * match.pd: Check for a no-op conversion before using wi::add
12033 rather than after. Use tree_to_uhwi when summing small shift
12034 counts into an unsigned int.
12036 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
12037 Alan Hayward <alan.hayward@arm.com>
12038 David Sherwood <david.sherwood@arm.com>
12041 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
12042 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
12043 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
12044 POINTER_AND_FP_REGS.
12046 2017-10-02 Richard Biener <rguenther@suse.de>
12048 PR tree-optimization/82355
12049 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
12050 a mapping for the enclosing loop but avoid generating one for
12051 the loop tree root.
12052 (copy_bb_and_scalar_dependences): Remove premature codegen
12053 error on PHIs in blocks duplicated into multiple places.
12054 * graphite-scop-detection.c
12055 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
12056 in the region use it as loop and nest to analyze the DR in.
12057 (try_generate_gimple_bb): Likewise.
12058 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
12059 (add_loop_constraints): For blocks in a loop not in the region
12060 create a dimension with a single iteration.
12061 * sese.h (gbb_loop_at_index): Remove assert.
12063 2017-10-01 Kevin Buettner <kevinb@redhat.com>
12065 * omp-expand.c (adjust_context_scope): New function.
12066 (expand_parallel_call): Call adjust_context_scope.
12068 2017-10-01 Jeff Law <law@redhat.com>
12070 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
12071 dom_opt_dom_walker class with direct access to private members.
12072 Add comments. Call test_for_singularity.
12073 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
12074 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
12075 m_dummy_cond anymore.
12076 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
12078 (pass_dominator:execute): Build the dummy_cond here and pass it
12079 to the dom_opt_dom_walker ctor.
12080 (test_for_singularity): New function.
12082 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
12083 Maya Rashish <coypu@sdf.org>
12085 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
12086 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
12087 (alpha*-*-netbsd*) Use nbsd_tm_file.
12088 (arm*-*-netbsdelf*) Likewise.
12089 (i[34567]86-*-netbsdelf*) Likewise.
12090 (x86_64-*-netbsd*) Likewise.
12091 (mips*-*-netbsd*) Likewise.
12092 (powerpc-*-netbsd*) Likewise.
12093 (sh*-*-netbsd*) Likewise.
12094 (sparc-*-netbsdelf*) Likewise.
12095 (sparc64-*-netbsd*) Likewise.
12096 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
12098 (vax-*-netbsdelf*) Likewise.
12099 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
12100 (UINT_FAST8_TYPE) Likewise.
12101 (INT_FAST16_TYPE) Check CHAR_FAST16.
12102 (UINT_FAST16_TYPE) Likewise.
12104 2017-09-30 Jakub Jelinek <jakub@redhat.com>
12107 * config/i386/i386.md
12108 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
12109 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
12110 *divmodsi4_zext_2): New define_insn_and_split.
12111 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
12112 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
12113 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
12114 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
12115 New define_insn_and_split.
12116 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
12117 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
12118 operands[1] having DImode when mode is SImode.
12120 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
12121 always SImode for DIV and MOD in REG_EQUAL notes.
12123 2017-09-29 Yury Gribov <tetra2005@gmail.com>
12125 PR middle-end/82319
12126 * match.pd: Fix handling of NaNs in pattern.
12128 2017-09-29 Jeff Law <law@redhat.com>
12130 * sbitmap.c (bitmap_bit_in_range_p): New function.
12131 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
12132 * tree-ssa-dse.c (live_bytes_read): New function.
12133 (dse_classify_store): Ignore reads of dead bytes.
12135 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
12136 typos and whitespace errors.
12137 * config/i386/predicates.md (address_no_seg_operand): Likewise.
12138 * config/s390/s390.c (s390_emit_prologue): Likewise.
12140 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
12143 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
12144 with a symbol for LRA.
12146 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
12148 PR rtl-optimization/82338
12149 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
12151 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
12153 * genmodes.c (calc_wider_mode): Suppress qsort macro.
12154 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
12155 (qsort_chk): Declare.
12156 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
12157 (qsort_chk): New function.
12159 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12161 PR tree-optimization/82337
12162 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
12163 phi definition if the PHI result appears in an abnormal PHI.
12164 (find_basis_for_base_expr): Don't record a basis if the LHS of the
12165 basis appears in an abnormal PHI.
12167 2017-09-29 Richard Biener <rguenther@suse.de>
12169 * graphite-isl-ast-to-gimple.c
12170 (translate_isl_ast_to_gimple::set_codegen_error): New function.
12171 (binary_op_to_tree): Use it.
12172 (get_rename_from_scev): Likewise.
12173 (copy_loop_phi_nodes): Likewise.
12174 (copy_bb_and_scalar_dependences): Likewise.
12175 (translate_pending_phi_nodes): Likewise.
12177 2017-09-29 Jakub Jelinek <jakub@redhat.com>
12180 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
12181 for movabsq $(i32 << shift), r64.
12183 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
12185 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
12186 index when encoding %esp as %rsp to avoid 0x67 prefix.
12188 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
12190 * config/i386/i386.md (*movsf_internal, *movdf_internal):
12191 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
12193 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
12195 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
12196 Extensions with more than 16 double VFP registers.
12197 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
12198 to_clear_mask and all code related to it. Replace the remaining
12199 entry by a sbitmap and adapt code accordingly.
12201 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
12203 * brig-builtins.def: Change pure attributes to const.
12205 2017-09-28 Joseph Myers <joseph@codesourcery.com>
12207 * config.gcc (default_gnu_indirect_function): Default to yes for
12208 sparc*-*-linux* with glibc.
12210 2017-09-28 Joseph Myers <joseph@codesourcery.com>
12212 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
12213 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
12214 when creating .init_array and .fini_array sections with priority
12217 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
12220 * config/aarch64/aarch64.c
12221 (aarch64_builtin_support_vector_misalignment): Always return false
12222 when misalignment is unknown.
12224 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
12226 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
12227 this function to return false if the definition used by the swap
12228 instruction is artificial, or if the memory address from which the
12229 constant value is loaded is not represented by a base address held
12230 in a register or if the base address register is a frame or stack
12231 pointer. Additionally, return false if the base address of the
12232 loaded constant is a SYMBOL_REF but is not considered to be a
12234 (replace_swapped_load_constant): New function.
12235 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
12236 loaded constant vector with a load of a swapped constant vector.
12238 2017-09-27 Carl Love <cel@us.ibm.com>
12240 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
12241 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
12242 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
12245 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
12247 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
12248 first, always call autopref_rank_data otherwise.
12250 2017-09-27 Richard Biener <rguenther@suse.de>
12252 * graphite-scop-detection.c (find_scop_parameters): Move
12253 loop bound handling ...
12254 (gather_bbs::before_dom_children): ... here, avoiding the need
12255 to build scop_info->loop_nest.
12256 (record_loop_in_sese): Remove.
12257 * sese.h (sese_info_t::loop_nest): Remove.
12258 * sese.c (new_sese_info): Do not allocate loop_nest.
12259 (free_sese_info): Do not free loop_nest.
12261 2017-09-27 Jakub Jelinek <jakub@redhat.com>
12264 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
12265 lhs from calls if the lhs has addressable type.
12267 2017-09-27 Richard Biener <rguenther@suse.de>
12269 * graphite.h (scop::max_alias_set): New member.
12270 * graphite-scop-detection.c: Remove references to non-existing
12271 --param in comments.
12272 (build_alias_sets): Record the maximum alias set used for drs.
12273 (build_scops): Support zero as unlimited for
12274 --param graphite-max-arrays-per-scop.
12275 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
12276 and inline into ...
12277 (build_poly_sr_1): ... here. Compute alias set based on the
12278 maximum alias set used for drs rather than
12279 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
12281 2017-09-27 Richard Biener <rguenther@suse.de>
12283 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
12284 --param loop-block-tile-size=0 to disable tiling.
12286 2017-09-27 Richard Biener <rguenther@suse.de>
12288 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
12289 (graphite-max-nb-scop-params): Document special value zero.
12290 * domwalk.h (dom_walker::STOP): New symbolical constant.
12291 (dom_walker::dom_walker): Add optional parameter for bb to
12293 (dom_walker::~dom_walker): Declare.
12294 (dom_walker::before_dom_children): Document STOP return value.
12295 (dom_walker::m_user_bb_to_rpo): New member.
12296 (dom_walker::m_bb_to_rpo): Likewise.
12297 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
12298 mapping here if not provided by the user.
12299 (dom_walker::~dom_walker): Free bb to RPO mapping if not
12300 provided by the user.
12301 (dom_walker::STOP): Define.
12302 (dom_walker::walk): Do not compute bb to RPO mapping here.
12303 Support STOP return value from before_dom_children to stop
12305 * graphite-optimize-isl.c (optimize_isl): If the schedule
12306 is the same still generate code if -fgraphite-identity
12307 or -floop-parallelize-all are given.
12308 * graphite-scop-detection.c: Include cfganal.h.
12309 (gather_bbs::gather_bbs): Get and pass through bb to RPO
12311 (gather_bbs::before_dom_children): Return STOP for BBs
12313 (build_scops): Compute bb to RPO mapping and pass it to
12314 the domwalk. Treat --param graphite-max-nb-scop-params=0
12315 as not limiting the number of params.
12316 * graphite.c (graphite_initialize): Remove limit on the
12317 number of basic-blocks in a function.
12318 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
12319 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
12320 default value of 10.
12322 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
12324 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
12325 Adjust code to eliminate needing to do the shift right 32-bits
12326 operation after XSCVDPSPN.
12328 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
12330 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
12331 ((X / Y) != 0 -> X >= Y): Likewise.
12333 2017-09-26 Carl Love <cel@us.ibm.com>
12335 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
12336 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
12337 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
12338 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
12339 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
12340 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
12341 definitions and overloading.
12342 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
12343 statement for P9V_BUILTIN_XST_LEN_R.
12344 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
12345 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
12346 define_expand and define_insn for the instructions and builtins.
12347 * doc/extend.texi: Update the built-in documentation file for the new
12348 built-in functions.
12349 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
12350 define_insn for the instructions
12352 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
12355 * gcc/config/netbsd-protos.h: New file.
12356 * gcc/config/netbsd.c: New file.
12357 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
12358 * gcc/config/t-netbsd: New file.
12359 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
12360 (tmake_file) Add t-netbsd.
12361 (extra_objs) Add netbsd.o.
12363 2017-09-26 Janus Weil <janus@gcc.gnu.org>
12367 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
12369 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
12371 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
12372 sign extension from a vector register to a GPR by doing a 32-bit
12373 direct move and then an EXTSW.
12374 (extendsi<mode>2 splitter): Likewise.
12375 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
12376 right or vector extract after doing XSCVDPSPN. Use
12377 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
12379 (movdi_from_sf_zero_ext): Likewise.
12380 (reload_gpr_from_vsxsf): Likewise.
12381 (p8_mfvsrd_4_disf): Delete, no longer used.
12382 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
12383 then needing to move the SFmode to a GPR to use the XSCVDPSP
12384 instruction instead of FRSP and XSCVDPSPN.
12385 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
12386 it is adjacent to the other XSCVSPDP insns.
12387 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
12388 SFmode to be in traditional Altivec registers.
12389 (vsx_xscvdpspn): Eliminate useless alternative constraint.
12390 (vsx_xscvspdpn): Likewise.
12391 (vsx_xscvspdpn_scalar): Likewise.
12393 2017-09-26 Martin Jambor <mjambor@suse.cz>
12395 * tree-sra.c (compare_access_positions): Put integral types first,
12396 stabilize sorting of integral types, remove conditions putting
12397 non-full-precision integers last.
12398 (sort_and_splice_var_accesses): Disable scalarization if a
12399 non-integert would be represented by a non-full-precision integer.
12401 2017-09-26 Joseph Myers <joseph@codesourcery.com>
12403 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
12404 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
12405 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
12406 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
12407 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
12408 conditionals inside the function instead of around it. Call
12409 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
12410 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
12412 2017-09-26 Richard Biener <rguenther@suse.de>
12414 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
12416 (scop_detection::build_scop_breadth): ... this. Removed.
12417 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
12418 (scop_detection::harmful_stmt_in_bb): Likewise.
12419 (scop_detection::graphite_can_represent_stmt): Likewise.
12420 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
12421 (scop_detection::can_represent_loop): Remove recursion, fold in ...
12422 (scop_detection::can_represent_loop_1): ... this. Removed.
12423 (scop_detection::harmful_loop_in_region): Simplify after inlining
12424 the above and remove more quadraticness.
12425 (build_scops): Adjust.
12426 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
12429 2017-09-26 Jakub Jelinek <jakub@redhat.com>
12432 * config/i386/i386.c (ix86_print_operand_address_as): Only test
12433 REGNO (base) == SP_REG if base is a REG.
12435 PR middle-end/35691
12436 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
12437 if it is different SSA_NAME.
12438 (optimize_range_tests_cmp_bitwise): New function.
12439 (optimize_range_tests): Call it.
12441 2017-09-26 Richard Biener <rguenther@suse.de>
12443 PR tree-optimization/82321
12444 * graphite.c (canonicalize_loop_closed_ssa): Properly check
12445 for the def being inside the loop.
12447 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12449 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
12451 * config/s390/s390-builtins.def: Fix constraint on op4.
12453 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12455 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
12456 independent expanders.
12457 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
12458 ("vec_ordered", "vec_unordered"): New expanders.
12460 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12462 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
12465 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12467 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
12468 vec_unpacks_lo_v16qi.
12469 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
12471 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12473 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
12474 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
12475 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
12477 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12479 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
12481 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
12483 ("*vec_slb<mode>"): New insn pattern.
12484 ("vec_shr_<mode>"): New expander.
12485 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
12486 and force the shift count operand to V16QImode.
12487 ("vec_srb<mode>"): Set shift count mode to V16QI.
12489 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12491 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
12492 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
12493 ("vec_widen_smult_hi_<mode>"): New expander definitions.
12495 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
12498 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
12500 2017-09-26 Richard Biener <rguenther@suse.de>
12502 PR tree-optimization/82320
12503 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
12506 2017-09-25 Jeff Law <law@redhat.com>
12508 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
12509 prototype for new argument.
12510 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
12511 mostly extracted from rs6000_emit_allocate_stack.
12512 (rs6000_emit_probe_stack_range_stack_clash): New function.
12513 (rs6000_emit_allocate_stack): Call
12514 rs6000_emit_probe_stack_range_stack_clash as needed.
12515 (rs6000_emit_probe_stack_range): Add additional argument
12516 to call to gen_probe_stack_range{si,di}.
12517 (output_probe_stack_range): New.
12518 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
12519 (output_probe_stack_range_stack_clash): New.
12520 (rs6000_emit_prologue): Emit notes into dump file as requested.
12521 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
12522 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
12523 Add additional operand and pass it to output_probe_stack_range.
12525 2017-09-25 Bin Cheng <bin.cheng@arm.com>
12527 PR tree-optimization/82163
12528 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
12529 (checking_verify_loop_closed_ssa): New parameter.
12530 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
12531 (check_loop_closed_ssa_stmt): Delete.
12532 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
12533 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
12534 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
12537 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
12539 * brig-builtins.def: Treat HSAIL barrier builtins as
12540 setjmp/longjump style functions.
12542 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
12544 * target.def (constant_alignment): New hook.
12545 * defaults.h (CONSTANT_ALIGNMENT): Delete.
12546 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
12547 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
12548 * doc/tm.texi: Regenerate.
12549 * targhooks.h (default_constant_alignment): Declare.
12550 (constant_alignment_word_strings): Likewise.
12551 * targhooks.c (default_constant_alignment): New function.
12552 (constant_alignment_word_strings): Likewise.
12553 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
12554 instead of CONSTANT_ALIGNMENT.
12555 * varasm.c (align_variable, get_variable_align, build_constant_desc)
12556 (force_const_mem): Likewise.
12557 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
12558 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
12559 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
12560 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12561 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
12563 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
12564 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12565 constant_alignment_word_strings.
12566 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
12567 (CONSTANT_ALIGNMENT): Likewise.
12568 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12569 (arm_constant_alignment): New function.
12570 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
12571 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12572 constant_alignment_word_strings.
12573 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
12574 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12575 constant_alignment_word_strings.
12576 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
12577 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12578 (cris_constant_alignment): New function.
12579 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
12580 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12581 (epiphany_constant_alignment): New function.
12582 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
12583 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12584 constant_alignment_word_strings.
12585 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
12586 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12587 constant_alignment_word_strings.
12588 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
12589 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12590 constant_alignment_word_strings.
12591 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
12592 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
12593 * config/i386/i386.c (ix86_constant_alignment): Make static.
12594 Use the same interface as the target hook.
12595 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12596 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
12597 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12598 constant_alignment_word_strings.
12599 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
12600 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
12601 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12602 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
12603 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12604 constant_alignment_word_strings.
12605 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
12606 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12607 constant_alignment_word_strings.
12608 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
12609 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12610 constant_alignment_word_strings.
12611 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
12612 * config/microblaze/microblaze.c (microblaze_constant_alignment):
12614 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12615 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
12616 * config/mips/mips.c (mips_constant_alignment): New function.
12617 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12618 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
12619 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
12620 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12621 (mmix_constant_alignment): Make static. Use the same interface
12622 as the target hook.
12623 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
12624 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12625 constant_alignment_word_strings.
12626 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
12627 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12628 constant_alignment_word_strings.
12629 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
12630 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12631 constant_alignment_word_strings.
12632 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
12633 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12634 (rs6000_constant_alignment): New function.
12635 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
12636 * config/riscv/riscv.c (riscv_constant_alignment): New function.
12637 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12638 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
12639 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12640 (rs6000_constant_alignment): New function.
12641 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
12642 * config/s390/s390.c (s390_constant_alignment): New function.
12643 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12644 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
12645 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12646 constant_alignment_word_strings.
12647 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
12648 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12649 (sparc_constant_alignment): New function.
12650 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
12651 * config/spu/spu.c (spu_constant_alignment): New function.
12652 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12653 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
12654 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12655 constant_alignment_word_strings.
12656 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
12657 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12658 constant_alignment_word_strings.
12659 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
12660 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12661 constant_alignment_word_strings.
12662 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
12663 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12664 (visium_constant_alignment): New function.
12665 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
12666 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12667 (xtensa_constant_alignment): New function.
12668 * system.h (CONSTANT_ALIGNMENT): Poison.
12670 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
12672 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
12673 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
12674 (rs6000_builtin_valid_without_lhs): New helper function.
12675 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12676 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
12678 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
12680 * target.h (vec_perm_indices): Use unsigned short rather than
12682 (auto_vec_perm_indices): Likewise.
12683 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
12684 Use unsigned int rather than unsigned char.
12685 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
12687 2017-09-25 Richard Biener <rguenther@suse.de>
12689 * cfgloop.h (sort_sibling_loops): Declare.
12690 * cfgloop.c (sort_sibling_loops_cmp): New helper.
12691 (sort_sibling_loops): New function sorting the sibling loop list
12693 * graphite.c (graphite_transform_loops): Sort sibling loops.
12695 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
12697 * target.def (vec_perm_const_ok): Change sel parameter to
12699 * optabs-query.c (can_vec_perm_p): Update accordingly.
12700 * doc/tm.texi: Regenerate.
12701 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
12702 auto_vec_perm_indices and remove separate nelt field.
12703 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
12704 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
12705 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
12706 (aarch64_expand_vec_perm_const): Update accordingly.
12707 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
12708 to vec_perm_indices.
12709 * config/arm/arm.c (expand_vec_perm_d): Change perm to
12710 auto_vec_perm_indices and remove separate nelt field.
12711 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
12712 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
12713 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
12715 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
12716 to vec_perm_indices.
12717 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
12718 sel to vec_perm_indices.
12719 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
12720 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
12721 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
12723 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
12726 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
12729 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
12730 on the FUNCTION_DECL function context if it has a DIE that is a
12733 2017-09-25 Richard Biener <rguenther@suse.de>
12735 PR tree-optimization/82285
12736 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
12739 2017-09-25 Tom de Vries <tom@codesourcery.com>
12743 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
12746 2017-09-25 Richard Biener <rguenther@suse.de>
12748 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
12749 ISL errors other than isl_error_quota happen. Dump if the
12750 schedule is the same.
12751 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
12752 errors instead of aborting inside ISL.
12754 2017-09-25 Iain Sandoe <iain@codesourcery.com>
12757 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
12758 of libgcc_eh for m64.
12759 * config/i386/darwin64.h: Likewise.
12761 2017-09-25 Richard Biener <rguenther@suse.de>
12763 PR middle-end/82144
12764 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
12765 attribute for incomplete types nor twice for complete ones.
12767 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
12770 * config/i386/i386.c (ix86_print_operand_address_as): Encode
12771 %esp as %rsp to avoid 0x67 prefix if there is no index or base
12774 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
12777 * config/i386/i386.opt (mprefer-avx256): Use
12778 ix86_target_flags variable.
12779 * config/i386/i386.c (ix86_target_string): Move
12780 -mprefer-avx256 to flag2_opts.
12782 2017-09-22 Jakub Jelinek <jakub@redhat.com>
12784 PR middle-end/35691
12785 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
12786 and x != -1 | y != -1 into (x & y) != -1.
12788 2017-09-22 Steve Ellcey <sellcey@cavium.com>
12790 * config.gcc: Add new case statement to set
12791 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
12792 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
12793 s390x-*-linux* case statements. Added aarch64 to the list of
12794 supported architectures.
12796 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
12798 PR tree-optimization/82289
12799 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
12800 STMT_VINFO_RELEVANT_P.
12802 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
12803 Alan Hayward <alan.hayward@arm.com>
12804 David Sherwood <david.sherwood@arm.com>
12806 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
12807 for VR_RANGE only; don't allow VR_ANTI_RANGE.
12808 (extract_range_from_binary_expr_1): Don't call
12809 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
12811 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
12812 Alan Hayward <alan.hayward@arm.com>
12813 David Sherwood <david.sherwood@arm.com>
12815 * target.def (preferred_vector_alignment): New hook.
12816 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
12818 * doc/tm.texi: Regenerate.
12819 * targhooks.h (default_preferred_vector_alignment): Declare.
12820 * targhooks.c (default_preferred_vector_alignment): New function.
12821 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
12823 (DR_TARGET_ALIGNMENT): New macro.
12824 (aligned_access_p): Update commentary.
12825 (vect_known_alignment_in_bytes): New function.
12826 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
12828 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
12829 Calculate the misalignment based on the target alignment rather than
12831 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
12832 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
12833 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
12834 the target alignment, rather than masking the element misalignment
12835 with the number of elements in a vector. Also use the target
12836 alignment when calculating the maximum number of peels.
12837 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
12838 instead of TYPE_ALIGN_UNIT.
12839 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
12840 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
12841 (vect_create_addr_base_for_vector_ref): Update call accordingly.
12842 (vect_create_data_ref_ptr): Likewise.
12843 (vect_setup_realignment): Realign by ANDing with
12844 -DR_TARGET_MISALIGNMENT.
12845 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
12846 the number of peels based on DR_TARGET_ALIGNMENT.
12847 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
12848 with the guaranteed alignment boundary when deciding whether
12850 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
12851 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
12852 (ensure_base_align): Remove stmt_info parameter. Get the
12853 target base alignment from DR_TARGET_ALIGNMENT.
12854 (vectorizable_store): Update call accordingly. Interpret
12855 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
12857 (vectorizable_load): Likewise.
12859 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
12860 Alan Hayward <alan.hayward@arm.com>
12861 David Sherwood <david.sherwood@arm.com>
12863 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
12864 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
12865 (vect_enhance_data_refs_alignment): Likewise.
12867 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
12869 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
12870 error. Only quit immediately if parsing is complete.
12871 (BEGIN): Initialize fatal_err and parse_done.
12872 (begin fpu, end fpu): Check number of arguments.
12873 (begin arch, end arch): Likewise.
12874 (begin cpu, end cpu): Likewise.
12875 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
12876 (optalias): Likewise.
12878 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
12880 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
12881 * config/arm/arm-isa.h: Delete. Move definitions to ...
12882 * arm-cpus.in: ... here. Use new feature and fgroup values.
12883 * config/arm/arm.c (arm_option_override): Use lower case for feature
12885 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
12886 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
12887 * config/arm/parsecpu.awk (END): Add new command 'isa'.
12889 (print_isa_bits_for): New function.
12890 (gen_isa): New function.
12891 (gen_comm_data): Use print_isa_bits_for.
12892 (define feature): New keyword.
12893 (define fgroup): New keyword.
12894 * config/arm/t-arm (TM_H): Remove.
12895 (GTM_H): Add arm-isa.h.
12896 (arm-isa.h): Add rule to generate file.
12897 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
12898 case for feature bit names.
12900 2017-09-22 Richard Biener <rguenther@suse.de>
12902 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
12904 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
12905 print of no dependency loops ...
12906 * graphite.c (graphite_transform_loops): ... here.
12907 (canonicalize_loop_closed_ssa_form): Work from inner to outer
12909 (same_close_phi_node, remove_duplicate_close_phi,
12910 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
12911 (canonicalize_loop_closed_ssa): ... here and simplify.
12912 * graphite-optimize-isl.c: Include tree-vectorizer.h.
12913 (optimize_isl): Use dump_printf_loc to tell when we stopped
12914 optimizing because of an ISL timeout.
12916 2017-09-22 Richard Biener <rguenther@suse.de>
12918 PR tree-optimization/82291
12919 * tree-if-conv.c (predicate_mem_writes): Make sure to
12920 remove writes in blocks predicated with false.
12922 2017-09-22 Richard Biener <rguenther@suse.de>
12924 * sese.c: Include cfganal.h.
12925 (if_region_set_false_region): Remove.
12926 (create_if_region_on_edge): Likewise.
12927 (move_sese_in_condition): Re-implement without destroying
12930 2017-09-22 Richard Biener <rguenther@suse.de>
12932 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
12933 Verify both BBs contain loop PHI nodes before dispatching to
12934 copy_loop_phi_args.
12935 (graphite_regenerate_ast_isl): Do not recompute dominators,
12936 do not verify three times. Restructure for clarity.
12937 * graphite-scop-detection.c (same_close_phi_node,
12938 remove_duplicate_close_phi, make_close_phi_nodes_unique,
12939 defined_in_loop_p, canonicalize_loop_closed_ssa,
12940 canonicalize_loop_closed_ssa_form): Simplify, remove excess
12941 checking and SSA rewrite, move to ...
12942 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
12943 (graphite_initialize): Do not pass in ctx, do not reset the
12944 SCEV cache, compute only dominators.
12945 (graphite_transform_loops): Allocate ISL ctx after
12946 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
12947 Maintain post-dominators only around build_scops.
12948 * sese.c (if_region_set_false_region): Make static. Free
12949 and recompute dominators.
12950 (move_sese_in_condition): Assert we don't get called with
12951 post-dominators computed.
12952 * sese.h (if_region_set_false_region): Remove.
12954 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
12956 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
12957 mode attribute for TARGET_AVX512VL.
12959 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
12961 * config/i386/i386.opt (mprefer-avx256): New option.
12962 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
12964 (ix86_preferred_simd_mode): Return 256-bit AVX modes
12965 for TARGET_PREFER_AVX256.
12966 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
12968 2017-09-21 Jeff Law <law@redhat.com>
12970 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
12971 Fix dump output if the only stack space is for pushed registers.
12973 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
12975 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
12978 2017-09-21 Martin Sebor <msebor@redhat.com>
12981 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
12982 code (in C++) or code that triggers warnings.
12984 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
12986 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
12988 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
12990 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
12991 * sched-rgn.c: Ditto.
12992 * sel-sched-ir.c: Ditto.
12994 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
12996 * toplev.h (set_random_seed): Adjust return type.
12997 * toplev.c (init_local_tick): Move eager initialization of random_seed
12998 to get_random_seed. Adjust comment.
12999 (init_random_seed): Inline to get_random_seed, delete.
13000 (get_random_seed): Initialize random_seed lazily.
13001 (set_random_seed): Do not return previous value.
13002 (print_switch_value): Do not call get_random_seed.
13004 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
13006 * cgraph.c (delete_function_version): New, broken out from...
13007 (cgraph_node::delete_function_version): ...here. Rename to
13008 cgraph_node::delete_function_version_by_decl. Update all uses.
13009 (cgraph_node::remove): Call delete_function_version.
13011 2017-09-21 Jakub Jelinek <jakub@redhat.com>
13014 * tree-inline.c (expand_call_inline): Emit clobber stmts for
13015 VAR_DECLs to which addressable non-volatile parameters are mapped
13016 and for id->retvar after the return value assignment. Clear
13017 id->retval and id->retbnd after inlining.
13019 2017-09-21 Richard Biener <rguenther@suse.de>
13021 PR tree-optimization/82276
13022 PR tree-optimization/82244
13023 * tree-vrp.c (build_assert_expr_for): Set
13024 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
13026 (remove_range_assertions): Revert earlier change.
13028 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
13031 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
13033 2017-09-21 Richard Biener <rguenther@suse.de>
13035 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
13036 Restore valid IL after code generation errors.
13037 * graphite.c (graphite_transform_loops): Diagnose code
13038 generation issues as MSG_MISSED_OPTIMIZATION and continue
13039 with processing SCOPs.
13041 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13042 Alan Hayward <alan.hayward@arm.com>
13043 David Sherwood <david.sherwood@arm.com>
13045 * calls.c (compute_argument_addresses): Use simplify_gen_binary
13046 rather than choosing between plus_constant and gen_rtx_<CODE>.
13047 * expr.c (emit_push_insn): Likewise.
13048 (expand_expr_real_2): Likewise.
13050 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13051 Alan Hayward <alan.hayward@arm.com>
13052 David Sherwood <david.sherwood@arm.com>
13054 * loop-unroll.c (split_iv): Call copy_rtx on the step.
13056 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13057 Alan Hayward <alan.hayward@arm.com>
13058 David Sherwood <david.sherwood@arm.com>
13060 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
13061 calling tree_to_uhwi.
13063 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13064 Alan Hayward <alan.hayward@arm.com>
13065 David Sherwood <david.sherwood@arm.com>
13067 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
13068 INTEGER_CST rather than a negative test for ADDR_EXPR.
13070 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13071 Alan Hayward <alan.hayward@arm.com>
13072 David Sherwood <david.sherwood@arm.com>
13074 * tree-vrp.c (extract_range_from_binary_expr_1): Check
13075 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
13077 2017-09-21 Richard Biener <rguenther@suse.de>
13079 PR tree-optimization/71351
13080 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
13081 graphite_create_new_loop_guard): Remove, fold remaining parts
13083 (translate_isl_ast_node_for): ... here and simplify.
13085 2017-09-21 Jakub Jelinek <jakub@redhat.com>
13088 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
13089 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
13090 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
13091 alternative always use QI mode, for -Os imov (=R,R) alternative
13092 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
13095 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13096 Jeff Law <law@redhat.com>
13098 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
13099 (allocate_stack_space): New function, partially extracted from
13100 s390_emit_prologue.
13101 (s390_emit_prologue): Track offset to most recent stack probe.
13102 Code to allocate space moved into allocate_stack_space.
13103 Dump actions when no stack is allocated.
13104 (s390_prologue_plus_offset): New function.
13105 (s390_emit_stack_probe): Likewise.
13107 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
13109 * common.opt (Wa, Wl, Wp, g, gz=): Add
13111 (gno-column-info): Remove.
13112 (gcolumn-info): Drop RejectNegative.
13113 (gno-): New prefix.
13114 (gno-record-gcc-switches): Remove.
13115 (grecord-gcc-switches): Drop RejectNegative.
13116 (gno-split-dwarf): Remove.
13117 (gsplit-dwarf): Drop RejectNegative.
13118 (gno-strict-dwarf): Remove.
13119 (gstrict-dwarf): Drop RejectNegative.
13120 * config/darwin.opt (gfull, gused): Add RejectNegative.
13121 * dwarf2out.c (gen_producer_string): Drop
13122 gno-record-gcc-switches handler.
13123 * optc-gen.awk: Add g to prefixes with negative forms.
13124 * opts-common.c (remapping_prefix_p): New.
13125 (find_opt): Check it.
13126 (generate_canonical_option): Test g prefix.
13127 (option_map): Add -gno- mapping.
13128 (add_misspelling_candidates): Check remapping_prefix_p.
13130 2017-09-20 Jeff Law <law@redhat.com>
13132 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
13133 thinko in stack clash protection support.
13135 * explow.c (compute_stack_clash_protection_loop_data): Use
13136 CONST_INT_P instead of explicit test. Verify object is a
13137 CONST_INT_P before looking at INTVAL.
13138 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
13139 instead of explicit test.
13141 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
13144 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
13145 address instead of to r1 and r11.
13147 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
13149 * config.gcc: Support "knm".
13150 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
13151 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
13153 * config/i386/i386.c (m_KNM): Define.
13154 (processor_target_table): Add "knm".
13156 (ix86_option_override_internal): Add "knm".
13157 (ix86_issue_rate): Add PROCESSOR_KNM.
13158 (ix86_adjust_cost): Ditto.
13159 (ia32_multipass_dfa_lookahead): Ditto.
13160 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
13161 (fold_builtin_cpu): Add M_INTEL_KNM.
13162 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
13163 (processor_type): Add PROCESSOR_KNM.
13164 * config/i386/x86-tune.def: Add m_KNM.
13165 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
13167 2017-09-20 Richard Biener <rguenther@suse.de>
13169 PR tree-optimization/80213
13170 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
13171 are allowed in empty BBs as well.
13172 (canonicalize_loop_closed_ssa): Also look for other complex
13174 (scop_detection::get_sese): Include the loop-closed PHI block
13176 (scop_detection::merge_sese): Remove code adding extra blocks.
13177 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
13178 (build_scops): Assert the final returned scop is invalid.
13180 2017-09-20 Richard Biener <rguenther@suse.de>
13182 PR tree-optimization/82264
13183 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
13185 (vn_phi_lookup): Likewise.
13186 (vn_phi_insert): Likewise.
13188 2017-09-20 Jakub Jelinek <jakub@redhat.com>
13190 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
13191 that fits into uhwi or shwi, add DW_AT_const_value regardless
13192 of early_dwarf without going through RTL, using add_AT_unsigned
13195 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
13196 (DEBUG_LTO_ABBREV_SECTION): Likewise.
13197 (DEBUG_LTO_MACINFO_SECTION): Likewise.
13198 (DEBUG_MACRO_SECTION): Likewise.
13199 (DEBUG_LTO_MACRO_SECTION): Likewise.
13200 (DEBUG_STR_DWO_SECTION): Likewise.
13201 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
13202 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
13203 (DEBUG_LTO_DWO_LINE_SECTION): Define.
13204 (DEBUG_LTO_LINE_STR_SECTION): Define.
13205 (init_sections_and_labels): Initialize debug_line_str_section
13206 variable. Initialize debug_loc_section for -gdwarf-5 to
13207 DEBUG_LOCLISTS_SECTION. Formatting fixes.
13209 2017-09-20 Richard Biener <rguenther@suse.de>
13211 * graphite-sese-to-poly.c (extract_affine): Properly handle
13212 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
13214 2017-09-20 Richard Biener <rguenther@suse.de>
13216 PR tree-optimization/81373
13217 * graphite-scop-detection.c (build_cross_bb_scalars_def):
13218 Force SESE live-out defs to be handled even if they are
13221 2017-09-19 Jeff Law <law@redhat.com>
13223 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
13224 nothing for stack adjustments with REG_STACK_CHECK.
13225 * sched-deps.c (parse_add_or_inc): Reject insns with
13226 REG_STACK_CHECK from dependency breaking.
13227 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
13228 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
13229 * reg-notes.def (STACK_CHECK): New note.
13231 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
13232 (ix86_expand_prologue): Dump stack clash info as needed.
13233 Call ix86_adjust_stack_and_probe_stack_clash as needed.
13235 * function.c (dump_stack_clash_frame_info): New function.
13236 * function.h (dump_stack_clash_frame_info): Prototype.
13237 (enum stack_clash_probes): New enum.
13239 * config/alpha/alpha.c (alpha_expand_prologue): Also check
13240 flag_stack_clash_protection.
13241 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
13242 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
13243 (arm_frame_pointer_required): Likewise.
13244 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
13245 (ia64_expand_prologue): Likewise.
13246 * config/mips/mips.c (mips_expand_prologue): Likewise.
13247 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
13248 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
13249 (sparc_flat_expand_prologue): Likewise.
13250 * config/spu/spu.c (spu_expand_prologue): Likewise.
13252 * explow.c: Include "params.h".
13253 (anti_adjust_stack_and_probe_stack_clash): New function.
13254 (get_stack_check_protect): Likewise.
13255 (compute_stack_clash_protection_loop_data): Likewise.
13256 (emit_stack_clash_protection_loop_start): Likewise.
13257 (emit_stack_clash_protection_loop_end): Likewise.
13258 (allocate_dynamic_stack_space): Use get_stack_check_protect.
13259 Use anti_adjust_stack_and_probe_stack_clash.
13260 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
13261 (emit_stack_clash_protection_loop_start): Likewise.
13262 (emit_stack_clash_protection_loop_end): Likewise.
13263 * rtl.h (get_stack_check_protect): Prototype.
13264 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
13265 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
13266 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
13268 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
13270 * doc/tm.texi: Rebuilt.
13271 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
13272 get_stack_check_protect.
13273 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
13274 * config/arm/arm.c (arm_expand_prologue): Likewise.
13275 (arm_frame_pointer_required): Likewise.
13276 * config/i386/i386.c (ix86_expand_prologue): Likewise.
13277 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
13278 * config/mips/mips.c (mips_expand_prologue): Likewise.
13279 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
13280 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
13281 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
13282 (sparc_flat_expand_prologue): Likewise.
13284 * common.opt (-fstack-clash-protection): New option.
13285 * flag-types.h (enum stack_check_type): Note difference between
13286 -fstack-check= and -fstack-clash-protection.
13287 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
13288 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
13289 * toplev.c (process_options): Issue warnings/errors for cases
13290 not handled with -fstack-clash-protection.
13291 * doc/invoke.texi (-fstack-clash-protection): Document new option.
13292 (-fstack-check): Note additional problem with -fstack-check=generic.
13293 Note that -fstack-check is primarily for Ada and refer users
13294 to -fstack-clash-protection for stack-clash-protection.
13295 Document new params for stack clash protection.
13297 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
13299 * config/i386/i386.c (ix86_split_long_move): Do not handle
13300 address used for LEA in a special way.
13302 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
13304 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
13306 2017-09-19 Martin Sebor <msebor@redhat.com>
13309 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
13310 of incompatible types.
13312 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
13314 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13315 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
13316 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13317 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
13319 2017-09-19 Richard Biener <rguenther@suse.de>
13321 PR tree-optimization/82244
13322 * tree-vrp.c (remove_range_assertions): Do not propagate
13323 a constant to abnormals but replace the assert with a copy.
13325 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
13327 PR rtl-optimization/57878
13328 PR rtl-optimization/68988
13329 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
13330 avoidance test involving non_reload_pseudos. Move frequency test
13331 below the general fragmentation avoidance test.
13333 2017-09-19 Richard Biener <rguenther@suse.de>
13335 PR tree-optimization/69728
13336 * graphite-sese-to-poly.c (schedule_error): New global.
13337 (add_loop_schedule): Handle empty domain by failing the
13339 (build_original_schedule): Handle schedule_error.
13341 2017-09-19 Richard Biener <rguenther@suse.de>
13343 * graphite-scop-detection.c (scop_detection::can_represent_loop):
13344 Do not iterate to sibling loops but only to siblings of inner
13347 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
13350 * config/m68k/m68k.md (moveq feeding equality comparison): Check
13351 that the registers are different.
13353 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
13355 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
13356 to processor_model and "amdfam17h" to arch_names_table.
13357 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
13359 2017-09-18 Jakub Jelinek <jakub@redhat.com>
13362 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
13364 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13365 Alan Hayward <alan.hayward@arm.com>
13366 David Sherwood <david.sherwood@arm.com>
13368 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
13370 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
13372 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
13373 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
13374 vect_schedule_slp_instance.
13375 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
13376 Update call to vect_slp_analyze_node_operations. Simplify return
13378 (vect_slp_analyze_bb_1): Update call accordingly.
13379 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
13380 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
13381 (vect_schedule_slp): Update call accordingly.
13383 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13384 Alan Hayward <alan.hayward@arm.com>
13385 David Sherwood <david.sherwood@arm.com>
13387 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
13388 with types that aren't in fact scalar.
13390 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13392 * tree-vect-slp.c (vect_record_max_nunits): New function,
13394 (vect_build_slp_tree_1): ...here.
13395 (vect_build_slp_tree_2): Call it for phis too.
13397 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13399 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
13400 to vect_get_vec_def_for_operand when getting the mask operand.
13402 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13403 Alan Hayward <alan.hayward@arm.com>
13404 David Sherwood <david.sherwood@arm.com>
13406 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
13409 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13410 Alan Hayward <alan.hayward@arm.com>
13411 David Sherwood <david.sherwood@arm.com>
13413 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
13414 calculation for vector booleans.
13416 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13417 Alan Hayward <alan.hayward@arm.com>
13418 David Sherwood <david.sherwood@arm.com>
13420 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
13422 (vect_transform_stmt): ...here.
13423 (vect_analyze_stmt): Use it instead of calling
13424 vectorizable_live_operation directly.
13426 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
13428 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
13429 non-SIMT targets in acc vector loops.
13431 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
13433 * configure.ac: Add arc and check if assembler supports gdwarf2.
13434 * configure: Regenerate.
13436 2017-09-18 Richard Biener <rguenther@suse.de>
13438 PR tree-optimization/82220
13439 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
13440 epilogue niters from the min_profitable_iters compute.
13442 2017-09-18 Jakub Jelinek <jakub@redhat.com>
13445 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
13446 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
13447 (ix86_init_pic_reg): Revert 2017-09-01 changes.
13449 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
13452 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
13453 switching to a new text section.
13455 2017-09-18 Richard Biener <rguenther@suse.de>
13457 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
13459 (build_alias_set): Reject aliases with no access function.
13461 2017-09-18 Richard Biener <rguenther@suse.de>
13463 PR tree-optimization/79622
13464 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
13466 (build_cross_bb_scalars_use): Likewise.
13468 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
13470 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
13472 2017-09-18 Alan Modra <amodra@gmail.com>
13475 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
13476 stack_pointer_rtx for count 0. Update comments. Break up
13477 large rtl expression.
13479 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
13481 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
13482 Increase to 20 bytes.
13483 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
13484 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
13485 or avx version of the stub.
13487 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
13490 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
13491 compute the minimum stack alignment. Also update preferred stack
13492 boundary for leaf functions.
13494 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
13496 PR tree-optimization/82228
13497 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
13500 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
13502 * common/config/nds32/nds32-common.c
13503 (nds32_option_optimization_table): Refine formatting.
13504 (nds32_option_optimization_table): Use -fsched-pressure and
13505 -fomit-frame-pointer for specific optimization level.
13507 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
13509 * config/nds32/nds32.c: Refine formatting and comments.
13510 * config/nds32/nds32.h: Likewise.
13511 * config/nds32/nds32.md: Likewise.
13512 * config/nds32/nds32-cost.c: Likewise.
13513 * config/nds32/nds32-isr.c: Likewise.
13514 * config/nds32/nds32-md-auxiliary.c: Likewise.
13515 * config/nds32/nds32-multiple.md: Likewise.
13516 * config/nds32/nds32-predicates.c: Likewise.
13518 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
13519 Jakub Jelinek <jakub@redhat.com>
13521 Add support for -std=c++2a.
13522 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
13524 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
13526 2017-09-15 Steve Ellcey <sellcey@cavium.com>
13529 * doc/extend.texi (Common Function Attributes): Add
13530 references to ARM, AArch64, and S/390 specific attributes.
13531 (Function Specific Option Pragmas): Add AArch64 and S/390
13532 to list of back ends that support the target pragma.
13534 2017-09-15 Nathan Sidwell <nathan@acm.org>
13536 * doc/standards.texi: Fix C++17 description. Update URLs for
13539 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
13541 * common.opt (Wcast-align=strict): New warning option.
13542 * doc/invoke.texi: Document -Wcast-align=strict.
13544 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
13546 * cgraph.h (cgraph_thunk_info): Add comments.
13547 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
13548 assert for VIRTUAL_* arguments stricter.
13550 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
13552 PR tree-optimization/71026
13553 * match.pd: Move RDIV patterns from fold-const.c
13554 * fold-const.c (distribute_real_division): Removed.
13555 (fold_binary_loc): Remove calls to distribute_real_divison.
13557 2017-09-15 Jakub Jelinek <jakub@redhat.com>
13559 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
13560 c++1z and gnu++1z as deprecated. Change other references to
13561 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
13562 Change -Wc++1z-compat to -Wc++17-compat.
13563 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
13564 * dwarf2out.c (highest_c_language): Handle C++17.
13565 (gen_compile_unit_die): Likewise.
13567 2017-09-15 Jakub Jelinek <jakub@redhat.com>
13569 PR rtl-optimization/82192
13570 * combine.c (make_extraction): Don't look through non-paradoxical
13571 SUBREGs or TRUNCATE if pos + len is or might be bigger than
13574 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
13575 Alan Hayward <alan.hayward@arm.com>
13576 David Sherwood <david.sherwood@arm.com>
13578 * target.def (function_arg_offset): New hook.
13579 * targhooks.h (default_function_arg_offset): Declare.
13580 * targhooks.c (default_function_arg_offset): New function.
13581 * function.c (locate_and_pad_parm): Use
13582 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
13583 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
13584 (TARGET_FUNCTION_ARG_OFFSET): ...this.
13585 * doc/tm.texi: Regenerate.
13586 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
13587 * config/spu/spu.c (spu_function_arg_offset): New function.
13588 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
13589 * system.h (FUNCTION_ARG_OFFSET): Poison.
13591 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
13592 Alan Hayard <alan.hayward@arm.com>
13593 David Sherwood <david.sherwood@arm.com>
13595 * target.def (truly_noop_truncation): New hook.
13596 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
13597 than TRULY_NOOP_TRUNCATION.
13598 * hooks.h (hook_bool_uint_uint_true): Declare.
13599 * hooks.c (hook_bool_uint_uint_true): New function.
13600 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
13601 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
13602 * doc/tm.texi: Regenerate.
13603 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
13604 rather than TRULY_NOOP_TRUNCATION in comments.
13605 (simplify_comparison): Likewise.
13606 (record_truncated_value): Likewise.
13607 * expmed.c (extract_bit_field_1): Likewise.
13608 (extract_split_bit_field): Likewise.
13609 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
13610 instead of TRULY_NOOP_TRUNCATION.
13611 * function.c (assign_parm_setup_block): Likewise.
13612 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
13613 * rtlhooks.c: Include target.h.
13614 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
13615 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
13616 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
13617 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
13618 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
13619 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
13620 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
13621 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
13622 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
13623 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
13624 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
13625 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
13626 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
13627 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
13628 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
13629 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
13630 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
13631 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
13632 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
13633 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
13634 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
13635 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
13636 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
13637 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
13638 * config/mips/mips.c (mips_truly_noop_truncation): New function.
13639 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
13640 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
13641 rather than TRULY_NOOP_TRUNCATION in comments.
13642 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
13643 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
13644 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
13645 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
13646 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
13647 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
13648 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
13649 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
13650 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
13651 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
13652 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
13653 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
13654 rather than TRULY_NOOP_TRUNCATION in comments.
13655 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
13656 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
13657 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
13658 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
13659 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
13660 TRULY_NOOP_TRUNCATION condition.
13661 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
13662 (TRULY_NOOP_TRUNCATION): Delete.
13663 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
13664 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
13665 * config/spu/spu.c (spu_truly_noop_truncation): New function.
13666 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
13667 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
13668 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
13669 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
13670 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
13671 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
13672 rather than TRULY_NOOP_TRUNCATION in comments.
13673 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
13674 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
13675 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
13676 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
13677 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
13678 * system.h (TRULY_NOOP_TRUNCATION): Poison.
13680 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
13683 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
13684 (*cmp_ior): Likewise.
13685 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
13686 (*ior_scc_scc_cmp): Likewise.
13687 (*and_scc_scc): Likewise.
13688 (*and_scc_scc_cmp): Likewise.
13690 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
13691 Alan Hayard <alan.hayward@arm.com>
13692 David Sherwood <david.sherwood@arm.com>
13694 * target.def (can_change_mode_class): New hook.
13695 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
13696 (hard_regno_nregs): Likewise.
13697 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
13698 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
13699 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
13700 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
13701 (LOAD_EXTEND_OP): Update accordingly.
13702 * doc/tm.texi: Regenerate.
13703 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
13704 CANNOT_CHANGE_MODE_CLASS.
13705 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
13706 (REG_CAN_CHANGE_MODE_P): ...this new macro.
13707 * combine.c (simplify_set): Update accordingly.
13708 * emit-rtl.c (validate_subreg): Likewise.
13709 * recog.c (general_operand): Likewise.
13710 * regcprop.c (mode_change_ok): Likewise.
13711 * reload1.c (choose_reload_regs): Likewise.
13712 (inherit_piecemeal_p): Likewise.
13713 * rtlanal.c (simplify_subreg_regno): Likewise.
13714 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
13715 instead of CANNOT_CHANGE_MODE_CLASS.
13716 (reload_cse_simplify_operands): Likewise.
13717 * reload.c (push_reload): Use targetm.can_change_mode_class
13718 instead of CANNOT_CHANGE_MODE_CLASS.
13719 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
13720 REG_CANNOT_CHANGE_MODE_P.
13721 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13722 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
13723 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13724 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13725 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13726 (arm_can_change_mode_class): New function.
13727 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
13728 than CANNOT_CHANGE_MODE_CLASS in comments.
13729 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13730 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
13731 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
13732 (ix86_can_change_mode_class): ...this new function, inverting the
13733 sense of the return value.
13734 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13735 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13736 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13737 (ia64_can_change_mode_class): New function.
13738 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13739 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
13740 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
13741 (m32c_can_change_mode_class): ...this new function, inverting the
13742 sense of the return value.
13743 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13744 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13745 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
13746 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
13747 (mips_can_change_mode_class): ...this new function, inverting the
13748 sense of the return value.
13749 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13750 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13751 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13752 (msp430_can_change_mode_class): New function.
13753 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13754 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
13755 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13756 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13757 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13758 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
13759 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13760 (pa_cannot_change_mode_class): Replace with...
13761 (pa_can_change_mode_class): ...this new function, inverting the
13762 sense of the return value.
13763 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
13764 than CANNOT_CHANGE_MODE_CLASS in comments.
13765 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13766 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
13767 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13768 (pdp11_cannot_change_mode_class): Replace with...
13769 (pdp11_can_change_mode_class): ...this new function, inverting the
13770 sense of the return value.
13771 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13772 * config/powerpcspe/powerpcspe-protos.h
13773 (rs6000_cannot_change_mode_class_ptr): Delete.
13774 * config/powerpcspe/powerpcspe.c
13775 (rs6000_cannot_change_mode_class_ptr): Delete.
13776 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13777 (rs6000_option_override_internal): Assign to
13778 targetm.can_change_mode_class instead of
13779 rs6000_cannot_change_mode_class_ptr.
13780 (rs6000_cannot_change_mode_class): Replace with...
13781 (rs6000_can_change_mode_class): ...this new function, inverting the
13782 sense of the return value.
13783 (rs6000_debug_cannot_change_mode_class): Replace with...
13784 (rs6000_debug_can_change_mode_class): ...this new function.
13785 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13786 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
13787 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13788 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13789 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
13791 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
13792 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13793 (rs6000_option_override_internal): Assign to
13794 targetm.can_change_mode_class instead of
13795 rs6000_cannot_change_mode_class_ptr.
13796 (rs6000_cannot_change_mode_class): Replace with...
13797 (rs6000_can_change_mode_class): ...this new function, inverting the
13798 sense of the return value.
13799 (rs6000_debug_cannot_change_mode_class): Replace with...
13800 (rs6000_debug_can_change_mode_class): ...this new function.
13801 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13802 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
13803 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
13804 (s390_can_change_mode_class): ...this new function, inverting the
13805 sense of the return value.
13806 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13807 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13808 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
13809 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13810 (sh_cannot_change_mode_class): Replace with...
13811 (sh_can_change_mode_class): ...this new function, inverting the
13812 sense of the return value.
13813 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13814 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13815 (sparc_can_change_mode_class): New function.
13816 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13817 * config/spu/spu.c (spu_can_change_mode_class): New function.
13818 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13819 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13820 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13821 (visium_can_change_mode_class): New function.
13822 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
13824 2017-09-15 Richard Biener <rguenther@suse.de>
13826 PR tree-optimization/82217
13827 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
13828 but not undefined case.
13830 2017-09-15 Jakub Jelinek <jakub@redhat.com>
13833 * postreload.c (reload_cse_simplify_operands): Skip
13834 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
13836 2017-09-15 Richard Biener <rguenther@suse.de>
13838 PR tree-optimization/68823
13839 * graphite-scop-detection.c (build_alias_set): If we have a
13840 possible dependence check whether we can handle them by just
13841 looking at the DRs DR_ACCESS_FNs.
13842 (build_scops): If build_alias_set fails, fail the SCOP.
13844 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
13846 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
13847 to support float128 built-in functions that require the ISA 3.0
13849 (BU_FLOAT128_3_HW): Likewise.
13850 (SQRTF128): Add support for the IEEE 128-bit square root and fma
13851 built-in functions.
13852 (FMAF128): Likewise.
13854 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
13855 support for built-in functions that need the ISA 3.0 IEEE 128-bit
13856 floating point instructions.
13857 (rs6000_invalid_builtin): Likewise.
13858 (rs6000_builtin_mask_names): Likewise.
13859 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
13860 (RS6000_BTM_FLOAT128_HW): Likewise.
13861 (RS6000_BTM_COMMON): Likewise.
13862 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
13864 * doc/extend.texi (RS/6000 built-in functions): Document the
13865 IEEE 128-bit floating point square root and fused multiply-add
13866 built-in functions.
13868 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
13870 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
13871 reg (r2) isn't in the set of registers defined in the prologue.
13873 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13874 Alan Hayward <alan.hayward@arm.com>
13875 David Sherwood <david.sherwood@arm.com>
13877 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
13878 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
13879 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
13880 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
13881 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
13883 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
13884 max_vectorization_factor.
13885 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
13887 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13888 Alan Hayward <alan.hayward@arm.com>
13889 David Sherwood <david.sherwood@arm.com>
13891 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
13892 (vect_worthwhile_without_simd_p): Declare.
13893 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
13894 (vectorizable_reduction): Use it.
13895 * tree-vect-stmts.c (vectorizable_shift): Likewise.
13896 (vectorizable_operation): Likewise.
13898 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13899 Alan Hayward <alan.hayward@arm.com>
13900 David Sherwood <david.sherwood@arm.com>
13902 * tree-vectorizer.h (vect_get_num_copies): New function.
13903 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
13904 * tree-vect-loop.c (vectorizable_reduction): Likewise.
13905 (vectorizable_induction): Likewise.
13906 (vectorizable_live_operation): Likewise.
13907 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
13908 (vectorizable_bswap): Likewise.
13909 (vectorizable_call): Likewise.
13910 (vectorizable_conversion): Likewise.
13911 (vectorizable_assignment): Likewise.
13912 (vectorizable_shift): Likewise.
13913 (vectorizable_operation): Likewise.
13914 (vectorizable_store): Likewise.
13915 (vectorizable_load): Likewise.
13916 (vectorizable_condition): Likewise.
13917 (vectorizable_comparison): Likewise.
13918 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
13920 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13921 Alan Hayward <alan.hayward@arm.com>
13922 David Sherwood <david.sherwood@arm.com>
13924 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
13925 of vect_init_vector.
13927 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13928 Alan Hayward <alan.hayward@arm.com>
13929 David Sherwood <david.sherwood@arm.com>
13931 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
13932 an inline wrapper that provides a location.
13933 (gimple_build_vector): Likewise.
13934 * gimple-fold.c (gimple_build_vector_from_val): New function.
13935 (gimple_build_vector): Likewise.
13936 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
13937 functions to build the initial value. Always return a gimple value.
13938 (get_initial_defs_for_reduction): Likewise. Only compute
13940 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
13941 vect_init_vector on the results from get_initial_def(s)_for_reduction.
13942 (vectorizable_induction): Use gimple_build_vector rather than
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 * target.h (vec_perm_indices): New typedef.
13950 (auto_vec_perm_indices): Likewise.
13951 * optabs-query.h: Include target.h
13952 (can_vec_perm_p): Take a vec_perm_indices *.
13953 * optabs-query.c (can_vec_perm_p): Likewise.
13954 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
13955 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
13956 * tree-vect-generic.c (lower_vec_perm): Likewise.
13957 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
13958 (vect_grouped_load_supported): Likewise.
13959 (vect_shift_permute_load_chain): Likewise.
13960 (vect_permute_store_chain): Use auto_vec_perm_indices.
13961 (vect_permute_load_chain): Likewise.
13962 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
13963 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
13964 Update uses of can_vec_perm_p.
13965 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
13966 mode with a number of elements. Take a vec_perm_indices *.
13967 (vect_create_epilog_for_reduction): Update accordingly.
13968 Use auto_vec_perm_indices.
13969 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
13970 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
13971 (vect_transform_slp_perm_load): Likewise.
13972 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
13973 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
13974 (vect_gen_perm_mask_checked): Likewise.
13975 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
13976 (vect_gen_perm_mask_checked): Likewise.
13977 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
13978 (vectorizable_store): Likewise.
13979 (vectorizable_load): Likewise.
13980 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
13981 (vectorizable_bswap): Likewise.
13983 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13984 Alan Hayward <alan.hayward@arm.com>
13985 David Sherwood <david.sherwood@arm.com>
13987 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
13988 * tree.c (build_vector): Likewise.
13989 (build_vector_from_ctor): Update accordingly.
13990 (build_vector_from_val): Likewise.
13991 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
13992 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
13993 * tree-vect-generic.c (add_rshift): Likewise.
13994 (expand_vector_divmod): Likewise.
13995 (optimize_vector_constructor): Likewise.
13996 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
13997 (vect_transform_slp_perm_load): Likewise.
13998 (vect_schedule_slp_instance): Likewise.
13999 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
14000 (vectorizable_call): Likewise.
14001 (vect_gen_perm_mask_any): Likewise. Add elements in order.
14002 * expmed.c (make_tree): Likewise.
14003 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
14004 a vector passed to build_vector.
14005 (fold_convert_const): Likewise.
14006 (exact_inverse): Likewise.
14007 (fold_ternary_loc): Likewise.
14008 (fold_relational_const): Likewise.
14009 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
14010 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
14011 (const_unop): Likewise. Store the reduction accumulator in a
14012 variable rather than an array.
14013 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
14014 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
14015 the new vector, rather than constructing it after the input arrays.
14016 (native_interpret_vector): Use auto_vec<tree> when building
14017 a vector passed to build_vector. Add elements in order.
14018 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
14019 auto_vec<tree> when building a vector passed to build_vector.
14020 (vect_create_epilog_for_reduction): Likewise.
14021 (vectorizable_induction): Likewise.
14022 (get_initial_def_for_reduction): Likewise. Fix indentation of
14024 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
14026 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
14027 passed to build_vector.
14029 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14030 Alan Hayward <alan.hayward@arm.com>
14031 David Sherwood <david.sherwood@arm.com>
14033 * tree-core.h (tree_base::u): Add an "nelts" field.
14034 (tree_vector): Use VECTOR_CST_NELTS as the length.
14035 * tree.c (tree_size): Likewise.
14036 (make_vector): Initialize VECTOR_CST_NELTS.
14037 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
14038 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
14039 TYPE_VECTOR_SUBPARTS.
14040 * expr.c (const_vector_mask_from_tree): Consistently use "units"
14041 as the number of units, setting it from VECTOR_CST_NELTS.
14042 (const_vector_from_tree): Likewise.
14043 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
14044 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
14045 (fold_negate_expr_1): Likewise.
14046 (fold_convert_const): Likewise.
14047 (const_binop): Likewise. Differentiate the number of output and
14049 (const_unop): Likewise.
14050 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
14051 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
14052 in cases that did the opposite.
14054 2017-09-14 Richard Biener <rguenther@suse.de>
14056 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
14059 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
14061 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
14063 2017-09-14 Jakub Jelinek <jakub@redhat.com>
14066 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
14067 if and where to split a bb, except for splitting before debug insn
14068 sequences followed by non-label real insn. Delete debug insns
14069 in between basic blocks.
14071 * combine.c (make_compound_operation_int): Formatting fixes.
14073 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
14074 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14075 * config/netbsd.h (LINK_EH_SPEC): Likewise.
14076 * config/sol2.h (LINK_EH_SPEC): Likewise.
14077 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14078 * config/s390/linux.h (LINK_SPEC): Likewise.
14079 * config/freebsd.h (LINK_EH_SPEC): Likewise.
14080 * config/openbsd.h (LINK_EH_SPEC): Likewise.
14081 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14082 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
14083 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
14084 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14085 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
14086 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
14088 2017-09-13 Jakub Jelinek <jakub@redhat.com>
14090 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
14092 (ENDFILE_LINUX_SPEC): Likewise.
14093 (LINK_EH_SPEC): Likewise.
14094 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
14095 (LINK_OS_LINUX_SPEC32): Likewise.
14096 (LINK_OS_LINUX_SPEC64): Likewise.
14097 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
14098 (LINK_OS_LINUX_SPEC): Likewise.
14100 2017-09-13 Martin Liska <mliska@suse.cz>
14102 PR middle-end/82154
14103 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
14104 CASE_HIGH is NULL_TREE.
14106 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
14107 Alan Hayward <alan.hayward@arm.com>
14108 David Sherwood <david.sherwood@arm.com>
14110 * target.def (secondary_memory_needed): New hook.
14111 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
14112 instead of SECONDARY_MEMORY_NEEDED.
14113 (secondary_memory_needed_mode): Likewise.
14114 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
14115 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
14116 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
14117 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
14118 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
14119 * doc/tm.texi: Regenerate.
14120 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
14121 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
14122 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14123 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
14124 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
14125 * config/i386/i386.c (inline_secondary_memory_needed): Put the
14126 mode argument first and change the reg_class arguments to reg_class_t.
14127 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
14128 Make static. Update the call to inline_secondary_memory_needed.
14129 (ix86_register_move_cost): Update the call to
14130 inline_secondary_memory_needed.
14131 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14132 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
14134 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
14135 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
14137 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
14138 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
14139 * config/mips/mips.c (mips_secondary_memory_needed): Make static
14140 and match hook interface. Add comment from mips.h.
14141 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14142 * config/mmix/mmix.md (truncdfsf2): Refer to
14143 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
14145 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
14146 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
14147 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14148 (pa_secondary_memory_needed): New function.
14149 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
14150 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
14151 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14152 (pdp11_secondary_memory_needed): Make static and match hook interface.
14153 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
14154 * config/powerpcspe/powerpcspe-protos.h
14155 (rs6000_secondary_memory_needed_ptr): Delete.
14156 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
14158 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14159 (rs6000_option_override_internal): Assign to
14160 targetm.secondary_memory_needed rather than
14161 rs6000_secondary_memory_needed_ptr.
14162 (rs6000_secondary_memory_needed): Match hook interface.
14163 (rs6000_debug_secondary_memory_needed): Likewise.
14164 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
14165 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
14166 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
14167 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14168 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
14169 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
14171 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
14172 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14173 (rs6000_option_override_internal): Assign to
14174 targetm.secondary_memory_needed rather than
14175 rs6000_secondary_memory_needed_ptr.
14176 (rs6000_secondary_memory_needed): Match hook interface.
14177 (rs6000_debug_secondary_memory_needed): Likewise.
14178 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
14179 * config/s390/s390.c (s390_secondary_memory_needed): New function.
14180 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14181 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
14182 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14183 (sparc_secondary_memory_needed): New function.
14184 * lra-constraints.c (check_and_process_move): Refer to
14185 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
14187 (curr_insn_transform): Likewise.
14188 (process_alt_operands): Use targetm.secondary_memory_needed
14189 instead of TARGET_SECONDARY_MEMORY_NEEDED.
14190 (check_secondary_memory_needed_p): Likewise.
14191 (choose_split_class): Likewise.
14192 * reload.c: Unconditionally include code that was previously
14193 conditional on SECONDARY_MEMORY_NEEDED.
14194 (push_secondary_reload): Use targetm.secondary_memory_needed
14195 instead of TARGET_SECONDARY_MEMORY_NEEDED.
14196 (push_reload): Likewise.
14197 * reload1.c: Unconditionally include code that was previously
14198 conditional on SECONDARY_MEMORY_NEEDED.
14199 (choose_reload_regs): Use targetm.secondary_memory_needed
14200 instead of TARGET_SECONDARY_MEMORY_NEEDED.
14201 (gen_reload): Likewise.
14202 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
14204 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
14205 Alan Hayward <alan.hayward@arm.com>
14206 David Sherwood <david.sherwood@arm.com>
14208 * target.def (secondary_memory_needed_mode): New hook:
14209 * targhooks.c (default_secondary_memory_needed_mode): Declare.
14210 * targhooks.h (default_secondary_memory_needed_mode): New function.
14211 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
14212 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
14213 * doc/tm.texi: Regenerate.
14214 * lra-constraints.c (check_and_process_move): Use
14215 targetm.secondary_memory_needed_mode instead of
14216 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
14217 (curr_insn_transform): Likewise.
14218 * reload.c (get_secondary_mem): Likewise.
14219 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14220 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
14222 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14223 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14224 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
14225 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14226 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
14228 * config/powerpcspe/powerpcspe-protos.h
14229 (rs6000_secondary_memory_needed_mode): Delete.
14230 * config/powerpcspe/powerpcspe.c
14231 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14232 (rs6000_secondary_memory_needed_mode): Make static.
14233 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14234 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
14236 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
14238 (rs6000_secondary_memory_needed_mode): Make static.
14239 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14240 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
14241 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14242 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14243 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
14245 (sparc_secondary_memory_needed_mode): New function.
14246 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
14248 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
14250 * config/aarch64/constraints.md (Umq): New constraint.
14251 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
14253 (mov<mode>): Update condition.
14255 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14257 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
14258 when bitposition is the same.
14260 2017-09-13 Richard Biener <rguenther@suse.de>
14262 * dwarf2out.c (output_die_symbol): Remove.
14263 (output_die): Do not output a DIEs symbol.
14265 2017-09-13 Richard Biener <rguenther@suse.de>
14267 PR middle-end/82128
14268 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
14269 default-def to avoid breaking iterator update with the weird
14270 interaction with cgraph_update_edges_for_call_stmt_node.
14272 2017-09-13 Richard Biener <rguenther@suse.de>
14274 * tree-cfg.c (verify_gimple_assign_binary): Add verification
14275 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
14276 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
14277 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
14279 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
14281 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
14282 Disable pc relative literal load irrespective of
14283 TARGET_FIX_ERR_A53_84341 for default.
14285 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
14287 * config/sparc/sparc.c (output_return): Output the source location of
14288 the insn in the delay slot, if any.
14289 (output_sibcall): Likewise.
14291 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
14294 * common.opt (-static-pie): New alias.
14295 (shared): Negate static-pie.
14296 (-no-pie): Update help text.
14298 (static-pie): New option.
14299 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
14300 -static-pie support.
14301 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
14302 (LINK_EH_SPEC): Likewise.
14303 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14304 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
14305 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
14306 * gcc.c (LINK_COMMAND_SPEC): Likewise.
14307 (init_gcc_specs): Likewise.
14308 (init_spec): Likewise.
14309 (display_help): Update help message for -pie.
14310 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
14313 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
14315 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
14316 (movdi_aarch64): Likewise.
14317 (movti_aarch64): Likewise.
14319 2017-09-12 Simon Wright <simon@pushface.org>
14322 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
14323 calculation of the minor version, always output as 0.
14325 2017-09-12 Jakub Jelinek <jakub@redhat.com>
14328 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
14329 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
14330 on it early, rather than manual conversion late. For
14331 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
14332 instead of performing manual conversion.
14334 2017-09-12 Carl Love <cel@us.ibm.com>
14336 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
14337 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
14339 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
14340 VMULOSW): Add definitions.
14341 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14342 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
14343 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
14344 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
14345 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
14347 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
14349 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
14351 (movti_aarch64): Likewise.
14352 (movdf_aarch64): Likewise.
14353 (movtf_aarch64): Likewise.
14354 (load_pairdi): Likewise.
14355 (store_pairdi): Likewise.
14356 (load_pairdf): Likewise.
14357 (store_pairdf): Likewise.
14358 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
14359 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
14360 (ldr_got_small_<mode>): Likewise.
14361 (ldr_got_small_28k_<mode>): Likewise.
14362 (ldr_got_tiny): Likewise.
14363 * config/aarch64/iterators.md (ldst_sz): New.
14364 (ldpstp_sz): Likewise.
14365 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
14367 (thunderx_load): Split load_8 to load_16.
14368 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
14370 (thunderx2t99_storepair_basic): Split store_8 to store_16.
14371 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
14372 (xgene1_store_pair): Split store_8 to store_16.
14373 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
14374 (falkor_st_0_st_sd): Split store_8 to store_16.
14376 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
14378 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
14379 and store1/2/3/4 to store_4/8/12/16.
14380 * config/aarch64/aarch64.md: Update for rename.
14381 * config/arm/arm.md: Likewise.: Likewise.
14382 * config/arm/arm.c: Likewise.
14383 * config/arm/thumb1.md: Likewise.
14384 * config/arm/thumb2.md: Likewise.
14385 * config/arm/vfp.md: Likewise.
14386 * config/arm/arm-generic.md: Likewise.
14387 * config/arm/arm1020e.md: Likewise.
14388 * config/arm/arm1026ejs.md: Likewise.
14389 * config/arm/arm1136jfs.md: Likewise.
14390 * config/arm/arm926ejs.md: Likewise.
14391 * config/arm/cortex-a15.md: Likewise.
14392 * config/arm/cortex-a17.md: Likewise.
14393 * config/arm/cortex-a5.md: Likewise.
14394 * config/arm/cortex-a53.md: Likewise.
14395 * config/arm/cortex-a57.md: Likewise.
14396 * config/arm/cortex-a7.md: Likewise.
14397 * config/arm/cortex-a8.md: Likewise.
14398 * config/arm/cortex-a9.md: Likewise.
14399 * config/arm/cortex-m4.md: Likewise.
14400 * config/arm/cortex-m7.md: Likewise.
14401 * config/arm/cortex-r4.md: Likewise.
14402 * config/arm/exynos-m1.md: Likewise.
14403 * config/arm/fa526.md: Likewise.
14404 * config/arm/fa606te.md: Likewise.
14405 * config/arm/fa626te.md: Likewise.
14406 * config/arm/fa726te.md: Likewise.
14407 * config/arm/fmp626.md: Likewise.
14408 * config/arm/iwmmxt.md: Likewise.
14409 * config/arm/ldmstm.md: Likewise.
14410 * config/arm/marvell-pj4.md: Likewise.
14411 * config/arm/xgene1.md: Likewise.
14412 * config/aarch64/thunderx.md: Likewise.
14413 * config/aarch64/thunderx2t99.md: Likewise.
14414 * config/aarch64/falkor.md: Likewise.
14416 2017-09-12 Martin Liska <mliska@suse.cz>
14418 * attribs.c (private_lookup_attribute): New function.
14419 * attribs.h (private_lookup_attribute): Declared here.
14420 (lookup_attribute): Called from this place.
14422 2017-09-12 Richard Biener <rguenther@suse.de>
14424 PR tree-optimization/82157
14425 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
14426 stmts with side-effects.
14428 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14429 Alan Hayward <alan.hayward@arm.com>
14430 David Sherwood <david.sherwood@arm.com>
14432 * target.def (hard_regno_nregs): New hook.
14433 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
14434 * targhooks.h (default_hard_regno_nregs): Declare.
14435 * targhooks.c (default_hard_regno_nregs): New function.
14436 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
14437 (TARGET_HARD_REGNO_NREGS): ...this hook.
14438 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
14439 (CLASS_MAX_NREGS): Likewise.
14440 * doc/tm.texi: Regenerate.
14441 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
14442 instead of HARD_REGNO_NREGS.
14443 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
14444 HARD_REGNO_NREGS in the comment.
14445 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
14446 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
14447 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
14448 Return an unsigned int.
14449 (TARGET_HARD_REGNO_NREGS): Redefine.
14450 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
14451 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
14452 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
14453 (arc_hard_regno_nregs): New function.
14454 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
14455 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
14456 (arm_hard_regno_nregs): New function.
14457 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
14458 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
14459 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
14460 (TARGET_HARD_REGNO_NREGS): Redefine.
14461 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
14462 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
14463 (HARD_REGNO_NREGS): Delete.
14464 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
14465 (cr16_hard_regno_nregs): New function.
14466 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
14467 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
14468 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
14469 (cris_hard_regno_nregs): New function.
14470 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
14471 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
14472 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
14473 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
14474 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
14475 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
14476 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
14477 (frv_hard_regno_nregs): Make static. Take and return an
14479 (frv_class_max_nregs): Remove outdated copy of documentation.
14480 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
14481 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
14482 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
14483 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
14484 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
14485 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
14486 (TARGET_HARD_REGNO_NREGS): Redefine.
14487 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
14488 (CLASS_MAX_NREGS): Update comment.
14489 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
14490 (ia64_hard_regno_nregs): New function.
14491 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
14492 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
14493 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
14494 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
14495 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
14497 (m32c_hard_regno_nregs): Likewise. Make static.
14498 (TARGET_HARD_REGNO_NREGS): Redefine.
14499 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
14500 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
14501 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
14502 (m68k_hard_regno_nregs): New function.
14503 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
14504 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
14505 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
14506 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
14507 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
14508 Take and return an unsigned int.
14509 (TARGET_HARD_REGNO_NREGS): Redefine.
14510 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
14511 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
14512 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
14513 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
14514 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
14515 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
14516 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
14517 (msp430_hard_regno_nregs): Make static. Take and return an
14519 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
14520 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
14521 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
14522 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
14523 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
14524 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
14525 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
14526 (TARGET_HARD_REGNO_NREGS): Redefine.
14527 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
14528 (PA_HARD_REGNO_NREGS): ...this.
14529 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
14530 (PA_HARD_REGNO_NREGS): ...this.
14531 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
14532 (pa_hard_regno_nregs): New function.
14533 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
14534 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
14535 (pdp11_hard_regno_nregs): New function.
14536 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
14537 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
14538 (rs6000_hard_regno_nregs_hook): New function.
14539 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
14540 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
14541 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
14542 Take and return an unsigned int. Move earlier in file.
14543 (TARGET_HARD_REGNO_NREGS): Redefine.
14544 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
14545 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
14546 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
14547 (rl78_hard_regno_nregs): Make static. Take and return an
14549 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
14550 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
14551 (rs6000_hard_regno_nregs_hook): New function.
14552 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
14553 * config/rx/rx.c (rx_hard_regno_nregs): New function.
14554 (TARGET_HARD_REGNO_NREGS): Redefine.
14555 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
14556 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
14557 instead of HARD_REGNO_NREGS.
14558 (s390_hard_regno_nregs): New function.
14559 (s390_hard_regno_mode_ok): Add comment from s390.h.
14560 (TARGET_HARD_REGNO_NREGS): Redefine.
14561 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
14562 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
14563 (sh_hard_regno_nregs): New function.
14564 (sh_pass_in_reg_p): Use it.
14565 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
14566 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
14567 (sparc_hard_regno_nregs): New function.
14568 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
14569 * config/spu/spu.c (spu_hard_regno_nregs): New function.
14570 (spu_function_arg_advance): Use it, supplying a valid register number.
14571 (TARGET_HARD_REGNO_NREGS): Redefine.
14572 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
14573 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
14574 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
14575 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
14576 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
14577 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
14578 (CLASS_MAX_NREGS): Remove copy of old documentation.
14579 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
14580 (visium_hard_regno_nregs): New function.
14581 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
14582 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
14583 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
14584 xtensa_hard_regno_nregs): New function.
14585 * system.h (HARD_REGNO_NREGS): Poison.
14587 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14589 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
14590 hard_regno_nregs instead of HARD_REGNO_NREGS.
14591 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
14592 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
14593 (c6x_expand_epilogue): Likewise.
14594 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
14595 (frv_read_iacc_argument): Likewise.
14596 * config/sh/sh.c: Include regs.h.
14597 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
14598 (regs_used): Likewise.
14599 (output_stack_adjust): Likewise.
14600 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
14601 * expmed.c: Include regs.h.
14602 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
14603 * ree.c: Include regs.h.
14604 (combine_reaching_defs): Use hard_regno_nregs instead of
14606 (add_removable_extension): Likewise.
14608 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14610 * regs.h (hard_regno_nregs): Turn into a function.
14611 (end_hard_regno): Update accordingly.
14612 * caller-save.c (setup_save_areas): Likewise.
14613 (save_call_clobbered_regs): Likewise.
14614 (replace_reg_with_saved_mem): Likewise.
14615 (insert_restore): Likewise.
14616 (insert_save): Likewise.
14617 * combine.c (can_change_dest_mode): Likewise.
14618 (move_deaths): Likewise.
14619 (distribute_notes): Likewise.
14620 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
14621 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
14622 (rs6000_split_multireg_move): Likewise.
14623 (rs6000_register_move_cost): Likewise.
14624 (rs6000_memory_move_cost): Likewise.
14625 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
14626 (rs6000_split_multireg_move): Likewise.
14627 (rs6000_register_move_cost): Likewise.
14628 (rs6000_memory_move_cost): Likewise.
14629 * cselib.c (cselib_reset_table): Likewise.
14630 (cselib_lookup_1): Likewise.
14631 * emit-rtl.c (set_mode_and_regno): Likewise.
14632 * function.c (aggregate_value_p): Likewise.
14633 * ira-color.c (setup_profitable_hard_regs): Likewise.
14634 (check_hard_reg_p): Likewise.
14635 (calculate_saved_nregs): Likewise.
14636 (assign_hard_reg): Likewise.
14637 (improve_allocation): Likewise.
14638 (calculate_spill_cost): Likewise.
14639 * ira-emit.c (modify_move_list): Likewise.
14640 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
14641 (ira_hard_reg_in_set_p): Likewise.
14642 * ira.c (setup_reg_mode_hard_regset): Likewise.
14643 (clarify_prohibited_class_mode_regs): Likewise.
14644 (check_allocation): Likewise.
14645 * lra-assigns.c (find_hard_regno_for_1): Likewise.
14646 (lra_setup_reg_renumber): Likewise.
14647 (setup_try_hard_regno_pseudos): Likewise.
14648 (spill_for): Likewise.
14649 (assign_hard_regno): Likewise.
14650 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
14651 * lra-constraints.c (in_class_p): Likewise.
14652 (lra_constraint_offset): Likewise.
14653 (simplify_operand_subreg): Likewise.
14654 (lra_constraints): Likewise.
14655 (split_reg): Likewise.
14656 (split_if_necessary): Likewise.
14657 (invariant_p): Likewise.
14658 (inherit_in_ebb): Likewise.
14659 * lra-lives.c (process_bb_lives): Likewise.
14660 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
14661 (get_hard_regs): Likewise.
14662 (do_remat): Likewise.
14663 * lra-spills.c (assign_spill_hard_regs): Likewise.
14664 * mode-switching.c (create_pre_exit): Likewise.
14665 * postreload.c (reload_combine_recognize_pattern): Likewise.
14666 * recog.c (peep2_find_free_register): Likewise.
14667 * regcprop.c (kill_value_regno): Likewise.
14668 (set_value_regno): Likewise.
14669 (copy_value): Likewise.
14670 (maybe_mode_change): Likewise.
14671 (find_oldest_value_reg): Likewise.
14672 (copyprop_hardreg_forward_1): Likewise.
14673 * regrename.c (check_new_reg_p): Likewise.
14674 (regrename_do_replace): Likewise.
14675 * reload.c (push_reload): Likewise.
14676 (combine_reloads): Likewise.
14677 (find_dummy_reload): Likewise.
14678 (operands_match_p): Likewise.
14679 (find_reloads): Likewise.
14680 (find_equiv_reg): Likewise.
14681 (reload_adjust_reg_for_mode): Likewise.
14682 * reload1.c (count_pseudo): Likewise.
14683 (count_spilled_pseudo): Likewise.
14684 (find_reg): Likewise.
14685 (clear_reload_reg_in_use): Likewise.
14686 (free_for_value_p): Likewise.
14687 (allocate_reload_reg): Likewise.
14688 (choose_reload_regs): Likewise.
14689 (reload_adjust_reg_for_temp): Likewise.
14690 (emit_reload_insns): Likewise.
14691 (delete_output_reload): Likewise.
14692 * rtlanal.c (subreg_get_info): Likewise.
14693 * sched-deps.c (sched_analyze_reg): Likewise.
14694 * sel-sched.c (init_regs_for_mode): Likewise.
14695 (mark_unavailable_hard_regs): Likewise.
14696 (choose_best_reg_1): Likewise.
14697 (verify_target_availability): Likewise.
14698 * valtrack.c (dead_debug_insert_temp): Likewise.
14699 * var-tracking.c (track_loc_p): Likewise.
14700 (emit_note_insn_var_location): Likewise.
14701 * varasm.c (make_decl_rtl): Likewise.
14702 * reginfo.c (choose_hard_reg_mode): Likewise.
14703 (init_reg_modes_target): Refer directly to
14704 this_target_regs->x_hard_regno_nregs.
14706 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14708 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
14709 instead of hard_regno_nregs.
14711 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14713 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
14714 end_hard_regno instead of hard_regno_nregs.
14715 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
14716 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
14717 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
14718 * ira-color.c (improve_allocation): Likewise.
14719 * lra-assigns.c (find_hard_regno_for_1): Likewise.
14720 * lra-lives.c (mark_regno_live): Likewise.
14721 (mark_regno_dead): Likewise.
14722 * lra-remat.c (operand_to_remat): Likewise.
14723 * lra.c (collect_non_operand_hard_regs): Likewise.
14724 * postreload.c (reload_combine_note_store): Likewise.
14725 (move2add_valid_value_p): Likewise.
14726 * reload.c (regno_clobbered_p): Likewise.
14728 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14730 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
14732 * config/v850/v850.c (v850_reorg): Likewise.
14733 * reload.c (refers_to_regno_for_reload_p): Likewise.
14734 (find_equiv_reg): Likewise.
14735 * reload1.c (reload_reg_reaches_end_p): Likewise.
14737 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14739 * caller-save.c (add_used_regs): Use REG_NREGS instead of
14741 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
14742 * config/arm/arm.c (output_move_neon): Likewise.
14743 (arm_attr_length_move_neon): Likewise.
14744 (neon_split_vcombine): Likewise.
14745 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
14746 (c6x_mark_reg_written): Likewise.
14747 (c6x_dwarf_register_span): Likewise.
14748 * config/i386/i386.c (ix86_save_reg): Likewise.
14749 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
14750 (rws_access_reg): Likewise.
14751 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
14752 * mode-switching.c (create_pre_exit): Likewise.
14753 * ree.c (combine_reaching_defs): Likewise.
14754 (add_removable_extension): Likewise.
14755 * regcprop.c (find_oldest_value_reg): Likewise.
14756 (copyprop_hardreg_forward_1): Likewise.
14757 * reload.c (reload_inner_reg_of_subreg): Likewise.
14758 (push_reload): Likewise.
14759 (combine_reloads): Likewise.
14760 (find_dummy_reload): Likewise.
14761 (reload_adjust_reg_for_mode): Likewise.
14762 * reload1.c (find_reload_regs): Likewise.
14763 (forget_old_reloads_1): Likewise.
14764 (reload_reg_free_for_value_p): Likewise.
14765 (reload_adjust_reg_for_temp): Likewise.
14766 (emit_reload_insns): Likewise.
14767 (delete_output_reload): Likewise.
14768 * sel-sched.c (choose_best_reg_1): Likewise.
14769 (choose_best_pseudo_reg): Likewise.
14771 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14772 Alan Hayward <alan.hayward@arm.com>
14773 David Sherwood <david.sherwood@arm.com>
14775 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
14776 * target.def (slow_unaligned_access): New hook.
14777 * targhooks.h (default_slow_unaligned_access): Declare.
14778 * targhooks.c (default_slow_unaligned_access): New function.
14779 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
14780 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
14781 * doc/tm.texi: Regenerate.
14782 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
14783 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
14784 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
14786 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
14787 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
14789 (rs6000_slow_unaligned_access): New function.
14790 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
14791 (expand_block_compare): Likewise.
14792 (expand_strn_compare): Likewise.
14793 (rs6000_rtx_costs): Likewise.
14794 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
14795 (riscv_slow_unaligned_access): Likewise.
14796 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
14797 (riscv_slow_unaligned_access_p): ...this and make static.
14798 (riscv_option_override): Update accordingly.
14799 (riscv_slow_unaligned_access): New function.
14800 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
14801 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
14802 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
14803 (rs6000_slow_unaligned_access): New function.
14804 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
14805 (rs6000_rtx_costs): Likewise.
14806 * config/rs6000/rs6000-string.c (expand_block_compare)
14807 (expand_strn_compare): Use targetm.slow_unaligned_access instead
14808 of SLOW_UNALIGNED_ACCESS.
14809 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
14810 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
14811 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
14812 of SLOW_UNALIGNED_ACCESS.
14813 * expmed.c (simple_mem_bitfield_p): Likewise.
14814 * expr.c (alignment_for_piecewise_move): Likewise.
14815 (emit_group_load_1): Likewise.
14816 (emit_group_store): Likewise.
14817 (copy_blkmode_from_reg): Likewise.
14818 (emit_push_insn): Likewise.
14819 (expand_assignment): Likewise.
14820 (store_field): Likewise.
14821 (expand_expr_real_1): Likewise.
14822 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
14823 * lra-constraints.c (simplify_operand_subreg): Likewise.
14824 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
14825 * gimple-ssa-store-merging.c: Likewise in block comment at start
14827 * tree-ssa-strlen.c: Include target.h.
14828 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
14829 of SLOW_UNALIGNED_ACCESS.
14830 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
14832 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14834 PR rtl-optimization/82185
14835 * expmed.c (emit_store_flag_int): Only test tem if it has been
14838 2017-09-12 Richard Biener <rguenther@suse.de>
14840 PR middle-end/82149
14841 * match.pd ((FTYPE) N CMP CST): Fix typo.
14843 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
14845 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
14847 (mips_near_type_p): Add 'short_call' attribute as a synonym
14849 * doc/extend.texi (short_call): Document new function attribute.
14851 2017-09-12 Jakub Jelinek <jakub@redhat.com>
14854 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
14855 assertion check that in the condition.
14856 (get_atomic_generic_size): Likewise. Before testing if parameter
14857 has pointer type, if it has array type, call for C++
14858 default_conversion to perform array-to-pointer conversion.
14860 2017-09-12 Richard Biener <rguenther@suse.de>
14862 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
14863 for operations we cannot scalarize.
14865 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
14867 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
14868 vectors heap vectors. Clean up comments.
14869 Make visited_bbs a reference.
14870 (profitable_jump_thread_path): Make GC
14871 vectors heap vectors. Clean up comments.
14873 (convert_and_register_jump_thread_path): Make GC vectors heap
14875 (check_subpath_and_update_thread_path): Same. Clean up comments.
14876 Make visited_bbs a reference.
14877 (handle_phi): Abstract common code to to
14878 register_jump_thread_path_if_profitable.
14879 Rename VAR_BB to DEF_BB.
14881 Make GC vectors heap vectors.
14882 Make visited_bbs a reference.
14883 (handle_assignment): Same.
14884 (register_jump_thread_path_if_profitable): New.
14885 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
14887 Make GC vectors heap vectors. Clean up comments.
14888 Make visited_bbs a reference.
14889 (find_jump_threads_backwards): Make visited_bbs live in the stack.
14890 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
14893 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
14896 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
14897 words of E_DImode object are reachable by xtensa_uimm8x4 access.
14899 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
14901 Revert r251800 and r251799.
14903 2017-09-11 Martin Jambor <mjambor@suse.cz>
14906 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
14907 arguments in advance.
14908 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
14909 use it to find predecessor edges.
14910 (naive_outof_ssa): Collect vector of predecessors.
14912 2017-09-08 Jason Merrill <jason@redhat.com>
14914 PR c++/70029 - ICE with ref-qualifier and -flto
14915 * langhooks.h (struct lang_hooks_for_types): Add
14916 copy_lang_qualifiers.
14917 * attribs.c (build_type_attribute_qual_variant): Use it.
14918 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
14920 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
14921 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
14923 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
14926 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
14927 (*mulsi3_sp64): New instruction.
14928 (mulsi3): New expander.
14930 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
14932 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
14934 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
14936 * sancov.c: Include memmodel.h.
14938 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
14941 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
14944 2017-09-07 Carl Love <cel@us.ibm.com>
14946 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
14947 the sldi instruction.
14949 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
14951 * sancov.c: Include tm_p.h.
14953 2017-09-07 Jakub Jelinek <jakub@redhat.com>
14956 * output.h (switch_to_other_text_partition): New declaration.
14957 * varasm.c (switch_to_other_text_partition): New function.
14958 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
14959 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
14960 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
14961 to the other text partition before emitting LCL label and switch back
14962 after emitting the word after it.
14964 2017-09-07 Richard Biener <rguenther@suse.de>
14966 * passes.def (pass_split_crit_edges): Remove instance before PRE.
14967 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
14968 critical edges here, after loop init.
14969 (pass_data_pre): Remove PROP_no_crit_edges flags.
14970 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
14971 for valueization of call args to avoid leaking VN_TOP.
14972 (visit_use): Assert we do not visit default defs.
14973 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
14974 Use error_mark_node to more easily detect leaking VN_TOP.
14975 All default-defs are varying, not VN_TOP. Mark them visited.
14976 (run_scc_vn): Make code match comment.
14978 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
14980 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
14981 OPTION_MASK_FLOAT128_KEYWORD.
14982 (POWERPC_MASKS): Likewise.
14983 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
14984 support for the -mfloat128-type option, and make -mfloat128
14985 default on PowerPC Linux systems. Define or undefine
14986 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
14987 Define __float128 to be __ieee128 if IEEE 128-bit support is
14988 enabled, or undefine it.
14989 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
14990 Delete defining __FLOAT128_TYPE__.
14991 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
14992 -mfloat128-type option and make -mfloat128 default on PowerPC
14994 (TARGET_FLOAT128_TYPE): Likewise.
14995 (-mfloat128-type): Likewise.
14996 * config/rs6000/rs6000.c (rs6000_option_override_internal):
14997 Delete the -mfloat128-type option and make -mfloat128 default on
14998 PowerPC Linux systems. Always use __ieee128 to be the keyword for
14999 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
15000 128-bit floating point is enabled. Change tests from using
15001 -mfloat128-type to -mfloat128.
15002 (rs6000_mangle_type): Use the correct mangling for the __float128
15003 type even if normal long double is restricted to 64-bits.
15004 (floatn_mode): Enable the _Float128 type by default on VSX Linux
15006 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
15007 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
15008 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
15010 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
15011 documentation for -mfloat128.
15013 2017-09-06 Olivier Hainque <hainque@adacore.com>
15015 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
15017 2017-09-06 Wish Wu <wishwu007@gmail.com>
15018 Jakub Jelinek <jakub@redhat.com>
15020 * asan.c (initialize_sanitizer_builtins): Add
15021 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
15022 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
15023 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
15024 BT_FN_VOID_UINT64_PTR variables.
15025 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
15026 (BT_FN_VOID_UINT16_UINT16): Likewise.
15027 (BT_FN_VOID_UINT32_UINT32): Likewise.
15028 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
15029 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
15030 (BT_FN_VOID_UINT64_PTR): Likewise.
15031 * common.opt (flag_sanitize_coverage): New variable.
15032 (fsanitize-coverage=trace-pc): Remove.
15033 (fsanitize-coverage=): Add.
15034 * flag-types.h (enum sanitize_coverage_code): New enum.
15035 * fold-const.c (fold_range_test): Disable non-short-circuit
15036 optimization if flag_sanitize_coverage.
15037 (fold_truth_andor): Likewise.
15038 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
15039 * opts.c (COVERAGE_SANITIZER_OPT): Define.
15040 (coverage_sanitizer_opts): New array.
15041 (get_closest_sanitizer_option): Add OPTS argument, handle also
15042 OPT_fsanitize_coverage_.
15043 (parse_sanitizer_options): Adjusted to also handle
15044 OPT_fsanitize_coverage_.
15045 (common_handle_option): Add OPT_fsanitize_coverage_.
15046 * sancov.c (instrument_comparison, instrument_switch): New function.
15047 (sancov_pass): Add trace-cmp support.
15048 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
15049 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
15050 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
15051 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
15052 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
15053 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
15054 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
15055 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
15056 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
15057 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
15059 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
15061 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
15062 error. Only quit immediately if parsing is complete.
15063 (BEGIN): Initialize fatal_err and parse_done.
15064 (begin fpu, end fpu): Check number of arguments.
15065 (begin arch, end arch): Likewise.
15066 (begin cpu, end cpu): Likewise.
15067 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
15068 (optalias): Likewise.
15070 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
15072 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
15073 * config/arm/arm-isa.h: Delete. Move definitions to ...
15074 * arm-cpus.in: ... here. Use new feature and fgroup values.
15075 * config/arm/arm.c (arm_option_override): Use lower case for feature
15077 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
15078 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
15079 * config/arm/parsecpu.awk (END): Add new command 'isa'.
15081 (print_isa_bits_for): New function.
15082 (gen_isa): New function.
15083 (gen_comm_data): Use print_isa_bits_for.
15084 (define feature): New keyword.
15085 (define fgroup): New keyword.
15086 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
15087 (arm-isa.h): Add rule to generate file.
15088 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
15089 case for feature bit names.
15091 2017-09-06 Richard Biener <rguenther@suse.de>
15093 * tree-ssa-pre.c (NECESSARY): Remove.
15094 (create_expression_by_pieces): Do not touch pass-local flags.
15095 (insert_into_preds_of_block): Likewise.
15096 (do_pre_regular_insertion): Likewise.
15097 (eliminate_insert): Likewise.
15098 (eliminate_dom_walker::before_dom_children): Likewise.
15099 (fini_eliminate): Do not look at inserted_exprs.
15100 (mark_operand_necessary): Remove.
15101 (remove_dead_inserted_code): Replace with simple work-list
15102 algorithm based on inserted_exprs and SSA uses.
15103 (pass_pre::execute): Re-order fini_eliminate and
15104 remove_dead_inserted_code.
15106 2017-09-06 Olivier Hainque <hainque@adacore.com>
15108 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
15109 for VxWorks 7. Adjust surrounding comments.
15111 2017-09-06 Richard Biener <rguenther@suse.de>
15113 * gimple-ssa-strength-reduction.c
15114 (find_candidates_dom_walker::before_dom_children): Also allow
15117 2017-09-06 Richard Biener <rguenther@suse.de>
15119 PR tree-optimization/82108
15120 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
15121 for gap in the non-permutation SLP case.
15123 2017-09-06 Martin Jambor <mjambor@suse.cz>
15125 PR tree-optimization/82078
15126 * tree-sra.c (sort_and_splice_var_accesses): Move call to
15127 add_access_to_work_queue...
15128 (build_accesses_from_assign): ...here.
15129 (propagate_all_subaccesses): Make sure racc is the group
15130 representative, if there is one.
15132 2017-09-06 Jakub Jelinek <jakub@redhat.com>
15134 PR middle-end/82095
15135 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
15138 2017-09-06 Richard Biener <rguenther@suse.de>
15140 * gimple-ssa-strength-reduction.c
15141 (find_candidates_dom_walker::before_doom_children): Use a
15142 type and not a mode check.
15144 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
15147 * config/arm/predicates.md (arm_general_adddi_operand): Create new
15149 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
15151 2017-09-05 Jeff Law <law@redhat.com>
15153 PR tree-optimization/64910
15154 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
15155 cases where we have 3 or more operands.
15157 2017-09-05 Jakub Jelinek <jakub@redhat.com>
15159 PR middle-end/81768
15160 * omp-low.c (lower_omp_for): Recompute tree invariant if
15161 gimple_omp_for_initial/final is ADDR_EXPR.
15163 PR middle-end/81768
15164 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
15165 into gimple val before gimplification fo the COND_EXPR.
15167 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
15169 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
15170 REGION_COPY argument.
15171 (thread_through_all_blocks): Remove unused argument to
15172 duplicate_thread_path.
15174 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15175 Alan Hayward <alan.hayward@arm.com>
15176 David Sherwood <david.sherwood@arm.com>
15178 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
15179 Take a scalar_mode rather than a machine_mode.
15180 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
15181 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
15182 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
15183 (aarch64_gen_adjusted_ldpstp): Likewise.
15184 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
15186 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15187 Alan Hayward <alan.hayward@arm.com>
15188 David Sherwood <david.sherwood@arm.com>
15190 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
15191 Take a scalar_int_mode instead of a machine_mode.
15192 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
15193 (aarch64_output_scalar_simd_mov_immediate): Likewise.
15194 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
15195 (aarch64_simd_attr_length_rglist): Delete.
15196 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
15197 a scalar_int_mode instead of a machine_mode.
15198 (aarch64_add_offset): Likewise.
15199 (aarch64_internal_mov_immediate): Likewise
15200 (aarch64_add_constant_internal): Likewise.
15201 (aarch64_add_constant): Likewise.
15202 (aarch64_movw_imm): Likewise.
15203 (aarch64_rtx_arith_op_extract_p): Likewise.
15204 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
15205 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
15206 Remove assert that the mode isn't a vector.
15207 (aarch64_output_scalar_simd_mov_immediate): Likewise.
15208 (aarch64_expand_mov_immediate): Update calls after above changes.
15209 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
15210 (aarch64_and_bitmask_imm): Check for scalar integer modes.
15211 (aarch64_move_imm): Likewise.
15212 (aarch64_can_const_movi_rtx_p): Likewise.
15213 (aarch64_strip_extend): Likewise.
15214 (aarch64_extr_rtx_p): Likewise.
15215 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
15216 a CONST_INT when the mode parameter is VOIDmode.
15217 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
15219 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15221 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
15222 * stor-layout.c (bitwise_mode_for_mode): Likewise.
15223 (bitwise_type_for_mode): Update accordingly.
15225 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15227 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
15228 * stor-layout.c (mode_for_size_tree): Likewise.
15229 (mode_for_array): Update accordingly.
15230 (layout_decl): Likewise.
15231 (compute_record_mode): Likewise. Only set the mode once.
15233 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15235 * target.def (get_mask_mode): Change return type to opt_mode.
15237 * doc/tm.texi: Regenerate.
15238 * targhooks.h (default_get_mask_mode): Return an opt_mode.
15239 * targhooks.c (default_get_mask_mode): Likewise.
15240 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
15241 * optabs-query.c (can_vec_mask_load_store_p): Update use of
15242 targetm.get_mask_mode.
15243 * tree.c (build_truth_vector_type): Likewise.
15245 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15247 * machmode.h (mode_for_vector): Return an opt_mode.
15248 * stor-layout.c (mode_for_vector): Likewise.
15249 (mode_for_int_vector): Update accordingly.
15250 (layout_type): Likewise.
15251 * config/i386/i386.c (emit_memmov): Likewise.
15252 (ix86_expand_set_or_movmem): Likewise.
15253 (ix86_expand_vector_init): Likewise.
15254 (ix86_get_mask_mode): Likewise.
15255 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
15257 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
15258 * expmed.c (extract_bit_field_1): Likewise.
15259 * expr.c (expand_expr_real_2): Likewise.
15260 * optabs-query.c (can_vec_perm_p): Likewise.
15261 (can_vec_mask_load_store_p): Likewise.
15262 * optabs.c (expand_vec_perm): Likewise.
15263 * targhooks.c (default_get_mask_mode): Likewise.
15264 * tree-vect-stmts.c (vectorizable_store): Likewise.
15265 (vectorizable_load): Likewise.
15266 (get_vectype_for_scalar_type_and_size): Likewise.
15268 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15270 * machmode.h (mode_for_int_vector): New function.
15271 * stor-layout.c (mode_for_int_vector): Likewise.
15272 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
15273 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
15274 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
15275 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
15276 (s390_expand_vcond): Likewise.
15278 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15280 * machmode.h (opt_machine_mode): New type.
15281 (opt_mode<T>): Allow construction from anything that can be
15283 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
15284 (mode_for_size): Return an opt_machine_mode.
15285 * stor-layout.c (mode_for_size): Likewise.
15286 (mode_for_size_tree): Update call accordingly.
15287 (bitwise_mode_for_mode): Likewise.
15288 (make_fract_type): Likewise.
15289 (make_accum_type): Likewise.
15290 * caller-save.c (replace_reg_with_saved_mem): Update call
15292 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15293 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15294 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15295 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15296 * expmed.c (extract_bit_field_1): Likewise.
15297 * reload.c (get_secondary_mem): Likewise.
15298 * varasm.c (assemble_integer): Likewise.
15299 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
15302 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15304 * machmode.h (decimal_float_mode_for_size): New function.
15305 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
15306 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
15307 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
15308 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
15309 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
15310 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
15312 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15314 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
15315 (get_builtin_sync_mode): Likewise.
15316 (expand_ifn_atomic_compare_exchange): Likewise.
15317 (expand_builtin_atomic_clear): Likewise.
15318 (expand_builtin_atomic_test_and_set): Likewise.
15319 (fold_builtin_atomic_always_lock_free): Likewise.
15320 * calls.c (compute_argument_addresses): Likewise.
15321 (emit_library_call_value_1): Likewise.
15322 (store_one_arg): Likewise.
15323 * combine.c (combine_instructions): Likewise.
15324 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
15325 * config/arm/arm.c (arm_function_value): Likewise.
15326 (aapcs_allocate_return_reg): Likewise.
15327 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
15328 * config/i386/i386.c (construct_container): Likewise.
15329 (ix86_gimplify_va_arg): Likewise.
15330 (ix86_expand_sse_cmp): Likewise.
15331 (emit_memmov): Likewise.
15332 (emit_memset): Likewise.
15333 (expand_small_movmem_or_setmem): Likewise.
15334 (ix86_expand_pextr): Likewise.
15335 (ix86_expand_pinsr): Likewise.
15336 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
15337 * config/microblaze/microblaze.c (microblaze_block_move_straight):
15339 * config/mips/mips.c (mips_function_value_1) Likewise.
15340 (mips_block_move_straight): Likewise.
15341 (mips_expand_ins_as_unaligned_store): Likewise.
15342 * config/powerpcspe/powerpcspe.c
15343 (rs6000_darwin64_record_arg_advance_flush): Likewise.
15344 (rs6000_darwin64_record_arg_flush): Likewise.
15345 * config/rs6000/rs6000.c
15346 (rs6000_darwin64_record_arg_advance_flush): Likewise.
15347 (rs6000_darwin64_record_arg_flush): Likewise.
15348 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
15349 (sparc_function_value_1): Likewise.
15350 * config/spu/spu.c (adjust_operand): Likewise.
15351 (spu_emit_branch_or_set): Likewise.
15352 (arith_immediate_p): Likewise.
15353 * emit-rtl.c (gen_lowpart_common): Likewise.
15354 * expr.c (expand_expr_real_1): Likewise.
15355 * function.c (assign_parm_setup_block): Likewise.
15356 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
15357 * reload1.c (alter_reg): Likewise.
15358 * stor-layout.c (mode_for_vector): Likewise.
15359 (layout_type): Likewise.
15361 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15363 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
15364 (spu_convert_move): Likewise.
15365 * lower-subreg.c (resolve_simple_move): Likewise.
15367 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15370 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
15371 define_insn to a define_expand.
15372 (altivec_vsum2sws_direct): New define_insn.
15373 (altivec_vsumsws): Convert from a define_insn to a define_expand.
15375 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
15377 * config/arm/arm.c (arm_option_params_internal): Improve setting of
15380 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
15385 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
15386 realign stack if stack alignment needed is less than incoming
15389 2017-09-05 Marek Polacek <polacek@redhat.com>
15392 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
15395 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
15397 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
15399 2017-09-05 Richard Biener <rguenther@suse.de>
15401 PR tree-optimization/82084
15402 * fold-const.c (can_native_encode_string_p): Handle wide characters.
15404 2017-09-05 Richard Biener <rguenther@suse.de>
15406 PR tree-optimization/82102
15407 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
15409 2017-09-05 Martin Liska <mliska@suse.cz>
15411 PR tree-optimization/82032
15412 * tree-cfg.c (generate_range_test): New function.
15413 * tree-cfg.h (generate_range_test): Declared here.
15414 * tree-cfgcleanup.c (convert_single_case_switch): New function.
15415 (cleanup_control_expr_graph): Use it.
15416 * tree-switch-conversion.c (try_switch_expansion): Remove
15418 (emit_case_nodes): Use generate_range_test.
15420 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
15423 * config/i386/i386.md (*<btsc><mode>_mask): Add
15424 TARGET_USE_BT to insn constraint.
15425 (*btr<mode>_mask): Ditto.
15427 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
15429 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
15430 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
15432 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
15435 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
15436 TARGET_NEON and TARGET_IWMMXT.
15437 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
15438 TARGET_NEON and TARGET_IWMMXT.
15439 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
15441 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
15443 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
15444 (ix86_rewrite_tls_address): Ditto.
15445 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
15446 (ix86_rewrite_tls_address_1): Ditto.
15447 (ix86_rewrite_tls_address): Ditto.
15448 * config/i386/predicates.md (tls_address_pattern): New predicate.
15449 * config/i386/i386.md (TLS address splitter): New splitter.
15451 2017-09-04 Richard Biener <rguenther@suse.de>
15453 PR tree-optimization/82084
15454 * fold-const.h (can_native_encode_string_p): Declare.
15455 * fold-const.c (can_native_encode_string_p): Factor out from ...
15456 (native_encode_string): ... here.
15457 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
15458 vectorizing stores from constants we later cannot handle.
15460 2017-09-04 Marek Polacek <polacek@redhat.com>
15463 * doc/invoke.texi: Update -Wtautological-compare documentation.
15465 2017-09-04 Jeff Law <law@redhat.com>
15467 PR tree-optimization/64910
15468 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
15469 swap the first and last operand if the last is a constant.
15471 2017-09-04 Marek Polacek <polacek@redhat.com>
15474 * convert.c (do_narrow): When sanitizing signed integer overflows,
15475 bail out for signed types.
15476 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
15478 2017-09-04 Richard Biener <rguenther@suse.de>
15480 PR tree-optimization/82060
15481 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
15482 Move devirtualization after stmt folding and before EH/AB/noreturn
15483 cleanup to get the stmt refs canonicalized. Use a bool instead
15484 of gimple_modified_p since that doesn't work for NOPs. Schedule
15485 NOPs generated by folding for removal.
15487 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
15488 Alan Hayward <alan.hayward@arm.com>
15489 David Sherwood <david.sherwood@arm.com>
15491 * coretypes.h (pad_direction): New enum.
15492 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
15493 (FUNCTION_ARG_PADDING): Likewise.
15494 * target.def (function_arg_padding): New hook.
15495 * targhooks.h (default_function_arg_padding): Declare.
15496 * targhooks.c (default_function_arg_padding): New function.
15497 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
15498 (TARGET_FUNCTION_ARG_PADDING): ...this.
15499 * doc/tm.texi: Regenerate.
15500 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
15501 instead of direction.
15502 (compute_argument_addresses): Likewise.
15503 (load_register_parameters): Likewise.
15504 (emit_library_call_value_1): Likewise.
15505 (store_one_arg): Use targetm.calls.function_arg_padding instead
15506 of FUNCTION_ARG_PADDING.
15507 (must_pass_in_stack_var_size_or_pad): Likewise.
15508 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
15509 (emit_group_store): Likewise.
15510 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
15511 instead of FUNCTION_ARG_PADDING.
15512 (emit_push_insn): Likewise, and propagate enum change throughout
15514 * function.h (direction): Delete.
15515 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
15517 * function.c (assign_parm_find_stack_rtl): Likewise.
15518 (assign_parm_setup_block_p): Likewise.
15519 (assign_parm_setup_block): Likewise.
15520 (gimplify_parameters): Likewise.
15521 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
15522 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
15524 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
15525 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
15526 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
15527 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
15528 (aarch64_function_arg_padding): ...this new function.
15529 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
15530 (TARGET_FUNCTION_ARG_PADDING): Redefine.
15531 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
15532 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
15533 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
15534 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
15535 (arm_pad_arg_upward): Replace with...
15536 (arm_function_arg_padding): ...this new function.
15537 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
15539 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
15540 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
15541 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
15542 (ia64_hpux_function_arg_padding): Replace with...
15543 (ia64_function_arg_padding): ...this new function. Use pad_direction
15544 instead of direction. Check for TARGET_HPUX.
15545 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
15546 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
15547 (iq2000_function_arg_padding): New function.
15548 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
15549 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
15550 (mips_function_arg_padding): ...this new function.
15551 (mips_pad_reg_upward): Update accordingly.
15552 (TARGET_FUNCTION_ARG_PADDING): Redefine.
15553 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
15554 targetm.calls.function_arg_padding.
15555 (FUNCTION_ARG_PADDING): Delete.
15556 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
15557 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
15558 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
15559 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
15560 (nios2_block_reg_padding): Return pad_direction instead of direction.
15561 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
15562 instead of direction.
15563 (nios2_function_arg_padding): Likewise. Make static.
15564 (TARGET_FUNCTION_ARG_PADDING): Redefine.
15565 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
15566 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
15567 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
15568 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
15569 (pa_function_arg_padding): Make static. Return pad_direction instead
15571 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
15572 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
15573 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
15574 instead of direction. Use targetm.calls.function_arg_padding.
15575 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
15576 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
15577 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
15578 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
15579 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
15581 (function_arg_padding): Rename to...
15582 (rs6000_function_arg_padding): ...this. Make static. Return
15583 pad_direction instead of direction.
15584 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
15585 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
15586 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
15587 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
15588 instead of direction. Use targetm.calls.function_arg_padding.
15589 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
15590 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
15591 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
15592 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
15593 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
15594 (function_arg_padding): Rename to...
15595 (rs6000_function_arg_padding): ...this. Make static. Return
15596 pad_direction instead of direction.
15597 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
15598 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
15599 * config/s390/s390.c (s390_function_arg_padding): New function.
15600 (TARGET_FUNCTION_ARG_PADDING): Redefine.
15601 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
15602 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
15603 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
15604 (function_arg_padding): Rename to...
15605 (sparc_function_arg_padding): ...this. Make static. Return
15606 pad_direction instead of direction.
15607 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
15608 * config/spu/spu.c (spu_function_arg_padding): New function.
15609 (TARGET_FUNCTION_ARG_PADDING): Redefine.
15610 * system.h (FUNCTION_ARG_PADDING): Poison.
15612 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
15613 Alan Hayward <alan.hayward@arm.com>
15614 David Sherwood <david.sherwood@arm.com>
15616 * target.def (modes_tieable_p): New hook.
15617 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
15618 (TARGET_MODES_TIEABLE_P): ...this.
15619 * doc/tm.texi.in: Regenerate.
15620 * hooks.h (hook_bool_mode_mode_true): Declare.
15621 * hooks.c (hook_bool_mode_mode_true): New function.
15622 * combine.c (subst): Use targetm.modes_tieable_p instead of
15624 * dse.c (find_shift_sequence): Likewise.
15625 * expmed.c (extract_low_bits): Likewise.
15626 * lower-subreg.c: Include target.h.
15627 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
15629 * rtlanal.c (rtx_cost): Likewise.
15630 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
15631 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
15632 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
15633 (TARGET_MODES_TIEABLE_P): Redefine.
15634 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
15635 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
15636 (TARGET_MODES_TIEABLE_P): Redefine.
15637 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
15638 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
15639 (arc_modes_tieable_p): New function.
15640 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
15641 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
15642 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
15643 (arm_modes_tieable_p): Make static.
15644 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
15645 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
15646 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
15647 (TARGET_MODES_TIEABLE_P): Redefine.
15648 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
15649 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
15650 (TARGET_MODES_TIEABLE_P): Redefine.
15651 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
15652 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
15653 (cr16_modes_tieable_p): New function.
15654 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
15655 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
15656 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
15657 (TRULY_NOOP_TRUNCATION): Update comment.
15658 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
15659 (TRULY_NOOP_TRUNCATION): Update comment.
15660 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
15661 (frv_modes_tieable_p): New function.
15662 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
15663 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
15664 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
15665 (TARGET_MODES_TIEABLE_P): Redefine.
15666 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
15667 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
15668 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
15669 (TARGET_MODES_TIEABLE_P): Redefine.
15670 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
15671 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
15672 (ia64_modes_tieable_p): New function.
15673 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
15674 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
15675 (iq2000_modes_tieable_p): New function.
15676 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
15677 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
15678 (lm32_modes_tieable_p): New function.
15679 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
15680 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
15681 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
15682 (TARGET_MODES_TIEABLE_P): Redefine.
15683 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
15684 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
15685 (m32r_modes_tieable_p): New function.
15686 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
15687 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
15688 (m68k_modes_tieable_p): New function.
15689 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
15690 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
15691 (mcore_modes_tieable_p): New function.
15692 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
15693 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
15695 (TARGET_MODES_TIEABLE_P): Redefine.
15696 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
15697 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
15698 * config/mips/mips.c (mips_modes_tieable_p): Make static.
15699 (TARGET_MODES_TIEABLE_P): Redefine.
15700 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
15701 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
15702 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
15703 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
15704 (mn10300_modes_tieable_p): ...this and make static.
15705 (TARGET_MODES_TIEABLE_P): Redefine.
15706 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
15707 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
15708 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
15709 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
15710 (msp430_modes_tieable_p): Make static.
15711 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
15712 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
15713 (TARGET_MODES_TIEABLE_P): Redefine.
15714 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
15715 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
15716 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
15717 (TARGET_MODES_TIEABLE_P): Redefine.
15718 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
15719 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
15720 * config/pa/pa.c (pa_modes_tieable_p): Make static.
15721 (TARGET_MODES_TIEABLE_P): Redefine.
15722 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
15723 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
15724 (pdp11_modes_tieable_p): New function.
15725 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
15726 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
15727 (rs6000_modes_tieable_p): New function.
15728 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
15729 * config/powerpcspe/powerpcspe.md: Update comment.
15730 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
15731 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
15732 (TARGET_MODES_TIEABLE_P): Redefine.
15733 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
15734 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
15735 (rl78_modes_tieable_p): New function.
15736 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
15737 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
15738 (rs6000_modes_tieable_p): New function.
15739 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
15740 * config/rs6000/rs6000.md: Update comment.
15741 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
15742 * config/rx/rx.c (rx_modes_tieable_p): New function.
15743 (TARGET_MODES_TIEABLE_P): Redefine.
15744 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
15745 * config/s390/s390.c (s390_modes_tieable_p): New function.
15746 (TARGET_MODES_TIEABLE_P): Redefine.
15747 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
15748 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
15749 (sh_modes_tieable_p): New function.
15750 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
15751 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
15752 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
15753 (sparc_modes_tieable_p): Make static.
15754 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
15755 * config/spu/spu.c (spu_modes_tieable_p): New function.
15756 (TARGET_MODES_TIEABLE_P): Redefine.
15757 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
15758 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
15759 (TARGET_MODES_TIEABLE_P): Redefine.
15760 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
15761 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
15762 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
15763 * config/v850/v850.c (v850_modes_tieable_p): New function.
15764 (TARGET_MODES_TIEABLE_P): Redefine.
15765 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
15766 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
15767 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
15768 (visium_modes_tieable_p): New function.
15769 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
15770 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
15771 (xtensa_modes_tieable_p): New function.
15772 * system.h (MODES_TIEABLE_P): Poison.
15774 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
15775 Alan Hayward <alan.hayward@arm.com>
15776 David Sherwood <david.sherwood@arm.com>
15778 * target.def (hard_regno_mode_ok): New hook.
15779 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
15780 (TARGET_HARD_REGNO_MODE_OK): ...this.
15781 * doc/tm.texi.in: Regenerate.
15782 * hooks.h (hook_bool_uint_mode_true): Declare.
15783 * hooks.c (hook_bool_uint_mode_true): New function.
15784 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
15785 HARD_REGNO_MODE_OK.
15786 * genpreds.c (write_insn_preds_c): Add an include of target.h.
15787 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
15788 instead of HARD_REGNO_MODE_OK.
15789 * caller-save.c: Include target.h.
15790 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
15791 HARD_REGNO_MODE_OK.
15792 * combine.c (can_combine_p): Likewise.
15793 (combinable_i3pat): Likewise.
15794 (can_change_dest_mode): Likewise.
15795 * expr.c (init_expr_target): Likewise.
15796 (convert_move): Likewise.
15797 (convert_modes): Likewise.
15798 * ira.c (setup_prohibited_class_mode_regs): Likewise.
15799 (setup_prohibited_mode_move_regs): Likewise.
15800 * ira.h (target_ira): Likewise.
15801 * lra-assigns.c (find_hard_regno_for_1): Likewise.
15802 * lra-constraints.c (process_alt_operands): Likewise.
15803 (split_reg): Likewise.
15804 * recog.c (peep2_find_free_register): Likewise.
15805 * ree.c (combine_reaching_defs): Likewise.
15806 * regcprop.c (maybe_mode_change): Likewise.
15807 * reginfo.c (init_reg_sets_1): Likewise.
15808 (choose_hard_reg_mode): Likewise.
15809 (simplifiable_subregs): Likewise.
15810 * regrename.c (check_new_reg_p): Likewise.
15811 * reload.c (find_valid_class): Likewise.
15812 (find_valid_class_1): Likewise.
15813 (reload_inner_reg_of_subreg): Likewise.
15814 (push_reload): Likewise.
15815 (combine_reloads): Likewise.
15816 (find_dummy_reload): Likewise.
15817 (find_reloads): Likewise.
15818 * reload1.c (find_reg): Likewise.
15819 (set_reload_reg): Likewise.
15820 (allocate_reload_reg): Likewise.
15821 (choose_reload_regs): Likewise.
15822 (reload_adjust_reg_for_temp): Likewise.
15823 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
15824 (simplify_subreg_regno): Likewise.
15825 * sel-sched.c (init_regs_for_mode): Likewise.
15826 * varasm.c (make_decl_rtl): Likewise.
15827 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
15828 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
15829 HARD_REGNO_MODE_OK.
15830 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
15831 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
15832 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15833 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
15834 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
15835 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15836 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
15837 (arc_mode_class): Delete.
15838 (HARD_REGNO_MODE_OK): Delete.
15839 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15840 (arc_hard_regno_mode_ok): Rename old array to...
15841 (arc_hard_regno_mode_ok_modes): ...this.
15842 (arc_conditional_register_usage): Update accordingly.
15843 (arc_mode_class): Make static.
15844 (arc_hard_regno_mode_ok): New function.
15845 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
15846 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
15847 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15848 (arm_hard_regno_mode_ok): Make static.
15849 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
15850 HARD_REGNO_MODE_OK.
15851 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
15852 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
15853 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
15855 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15856 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
15857 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
15858 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
15859 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
15860 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15861 * config/bfin/predicates.md (valid_reg_operand): Use
15862 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
15863 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
15864 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
15865 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15866 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
15867 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
15868 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15869 (cr16_hard_regno_mode_ok): Make static and return a bool.
15870 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
15871 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15872 (cris_hard_regno_mode_ok): New function.
15873 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
15874 (epiphany_mode_class): Delete.
15875 (HARD_REGNO_MODE_OK): Delete.
15876 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
15877 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15878 (hard_regno_mode_ok): Rename to...
15879 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
15880 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
15881 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
15882 HARD_REGNO_MODE_OK.
15883 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
15884 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
15885 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15886 (frv_hard_regno_mode_ok): Make static and return a bool.
15887 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
15888 HARD_REGNO_MODE_OK.
15889 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
15890 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
15891 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
15892 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
15894 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15895 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
15896 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
15897 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
15899 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15900 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
15901 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15902 (ia64_hard_regno_mode_ok): New function.
15903 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
15904 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15905 (iq2000_hard_regno_mode_ok): New function.
15906 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
15907 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15908 (lm32_hard_regno_mode_ok): New function.
15909 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
15910 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
15911 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
15912 instead of HARD_REGNO_MODE_OK.
15913 (m32c_hard_regno_ok): Rename to...
15914 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
15915 (m32c_cannot_change_mode_class): Update accordingly.
15916 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15917 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
15918 (m32r_mode_class): Delete.
15919 (HARD_REGNO_MODE_OK): Delete.
15920 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15921 (m32r_hard_regno_mode_ok): Rename to...
15922 (m32r_hard_regno_modes): ...this.
15923 (m32r_mode_class): Make static.
15924 (m32r_hard_regno_mode_ok): New function.
15925 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
15926 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
15927 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15928 (m68k_hard_regno_mode_ok): Make static.
15929 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
15930 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15931 (mcore_hard_regno_mode_ok): New function.
15932 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
15933 (HARD_REGNO_MODE_OK): Delete.
15934 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
15936 (microblaze_hard_regno_mode_ok_p): ...this and make static.
15937 (microblaze_hard_regno_mode_ok): New function.
15938 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15939 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
15940 (mips_hard_regno_mode_ok): Delete.
15941 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
15942 (mips_hard_regno_mode_ok_p): ...this and make static.
15943 (mips_hard_regno_mode_ok_p): Rename to...
15944 (mips_hard_regno_mode_ok_uncached): ...this.
15945 (mips_hard_regno_mode_ok): New function.
15946 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
15947 of HARD_REGNO_MODE_OK.
15948 (mips_option_override): Update after above name changes.
15949 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15950 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
15951 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
15952 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
15953 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
15954 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15955 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
15956 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
15957 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
15958 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15959 (msp430_hard_regno_mode_ok): Make static and return a bool.
15960 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
15961 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
15962 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
15964 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15965 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
15966 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
15967 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
15968 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
15969 (PA_HARD_REGNO_MODE_OK): ...this
15970 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
15971 (PA_HARD_REGNO_MODE_OK): ...this.
15972 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15973 (pa_hard_regno_mode_ok): New function.
15974 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
15975 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15976 (pdp11_hard_regno_mode_ok): New function.
15977 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
15978 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
15980 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
15982 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15983 (rs6000_hard_regno_mode_ok): Rename to...
15984 (rs6000_hard_regno_mode_ok_uncached): ...this.
15985 (rs6000_init_hard_regno_mode_ok): Update accordingly.
15986 (rs6000_hard_regno_mode_ok): New function.
15987 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
15988 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
15989 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
15990 (riscv_hard_regno_mode_ok): ...this and make static.
15991 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15992 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
15993 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
15994 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15995 (rl78_hard_regno_mode_ok): Make static and return bool.
15996 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
15997 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
15999 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
16000 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16001 (rs6000_hard_regno_mode_ok): Rename to...
16002 (rs6000_hard_regno_mode_ok_uncached): ...this.
16003 (rs6000_init_hard_regno_mode_ok): Update accordingly.
16004 (rs6000_hard_regno_mode_ok): New function.
16005 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
16006 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
16007 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16008 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
16009 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
16010 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
16011 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16012 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
16013 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
16014 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16015 (sh_hard_regno_mode_ok): Make static.
16016 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
16017 instead of HARD_REGNO_MODE_OK.
16018 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
16019 (sparc_mode_class): Delete.
16020 (HARD_REGNO_MODE_OK): Delete.
16021 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16022 (hard_regno_mode_classes): Make static.
16023 (sparc_mode_class): Likewise.
16024 (sparc_hard_regno_mode_ok): New function.
16025 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
16026 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
16027 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
16029 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16030 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
16031 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
16032 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
16033 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
16034 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16035 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
16036 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
16037 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16038 (visium_hard_regno_mode_ok): New function.
16039 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
16040 instead of HARD_REGNO_MODE_OK.
16041 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
16042 (HARD_REGNO_MODE_OK): Delete.
16043 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
16044 (xtensa_hard_regno_mode_ok_p): ...this and make static.
16045 (xtensa_option_override): Update accordingly.
16046 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16047 (xtensa_hard_regno_mode_ok): New function.
16048 * system.h (HARD_REGNO_MODE_OK): Poison.
16050 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16051 Alan Hayward <alan.hayward@arm.com>
16052 David Sherwood <david.sherwood@arm.com>
16054 * target.def (hard_regno_call_part_clobbered): New hook.
16055 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
16056 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
16057 * doc/tm.texi: Regenerate.
16058 * hooks.h (hook_bool_uint_mode_false): Declare.
16059 * hooks.c (hook_bool_uint_mode_false): New function.
16060 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16061 * cselib.c (cselib_process_insn): Use
16062 targetm.hard_regno_call_part_clobbered instead of
16063 HARD_REGNO_CALL_PART_CLOBBERED.
16064 * ira-conflicts.c (ira_build_conflicts): Likewise.
16065 * ira-costs.c (ira_tune_allocno_costs): Likewise.
16066 * lra-constraints.c (need_for_call_save_p): Likewise.
16067 * lra-lives.c: Include target.h.
16068 (check_pseudos_live_through_calls): Use
16069 targetm.hard_regno_call_part_clobbered instead of
16070 HARD_REGNO_CALL_PART_CLOBBERED.
16071 * regcprop.c: Include target.h.
16072 (copyprop_hardreg_forward_1): Use
16073 targetm.hard_regno_call_part_clobbered instead of
16074 HARD_REGNO_CALL_PART_CLOBBERED.
16075 * reginfo.c (choose_hard_reg_mode): Likewise.
16076 * regrename.c (check_new_reg_p): Likewise.
16077 * reload.c (find_equiv_reg): Likewise.
16078 * reload1.c (emit_reload_insns): Likewise.
16079 * sched-deps.c (deps_analyze_insn): Likewise.
16080 * sel-sched.c (init_regs_for_mode): Likewise.
16081 (mark_unavailable_hard_regs): Likewise.
16082 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
16083 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16084 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
16086 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16087 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16088 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
16090 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
16092 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16093 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16094 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
16096 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16097 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16098 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
16100 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16101 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
16103 * config/powerpcspe/powerpcspe.c
16104 (rs6000_hard_regno_call_part_clobbered): New function.
16105 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16106 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16107 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
16109 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16110 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16111 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
16113 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16114 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16115 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
16117 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16118 Alan Hayward <alan.hayward@arm.com>
16119 David Sherwood <david.sherwood@arm.com>
16121 * rtl.h (subreg_memory_offset): Declare.
16122 * emit-rtl.c (subreg_memory_offset): New function.
16123 * expmed.c (store_bit_field_1): Use it.
16124 * expr.c (undefined_operand_subword_p): Likewise.
16125 * simplify-rtx.c (simplify_subreg): Likewise.
16127 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
16129 PR rtl-optimization/57448
16132 * optabs.c (expand_atomic_load): Place compiler memory barriers if
16133 using atomic_load pattern.
16134 (expand_atomic_store): Likewise.
16136 2017-09-04 Jakub Jelinek <jakub@redhat.com>
16139 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
16140 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
16141 handling. Use replace_call_with_value with NULL instead of
16142 gsi_replace, unlink_stmt_vdef and release_defs.
16144 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
16147 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
16149 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16152 * rtl.h (emit_library_call_value_1): Declare.
16153 (emit_library_call): Replace declaration with a series of overloads.
16154 Remove the parameter count argument.
16155 (emit_library_call_value): Likewise.
16156 * calls.c (emit_library_call_value_1): Make global. Replace varargs
16157 with an "rtx_mode_t *".
16158 (emit_library_call_value): Delete.
16159 (emit_library_call): Likewise.
16160 * asan.c (asan_emit_stack_protection): Update calls accordingly.
16161 (asan_emit_allocas_unpoison): Likewise.
16162 * builtins.c (expand_builtin_powi): Likewise.
16163 (expand_asan_emit_allocas_unpoison): Likewise.
16164 * cfgexpand.c (expand_main_function): Likewise.
16165 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
16166 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
16167 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
16168 * config/arm/arm.c (arm_trampoline_init): Likewise.
16169 (arm_call_tls_get_addr): Likewise.
16170 (arm_expand_divmod_libfunc): Likewise.
16171 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
16172 (smulsi3_highpart): Likewise.
16173 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
16174 (c6x_expand_compare): Likewise.
16175 (c6x_expand_movmem): Likewise.
16176 * config/frv/frv.c (frv_trampoline_init): Likewise.
16177 * config/i386/i386.c (ix86_trampoline_init): Likewise.
16178 (ix86_expand_divmod_libfunc): Likewise.
16179 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
16180 (ia64_expand_compare): Likewise.
16181 (ia64_profile_hook): Likewise.
16182 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
16183 (nonlocal_goto): Likewise.
16184 (restore_stack_nonlocal): Likewise.
16185 * config/m32r/m32r.c (block_move_call): Likewise.
16186 (m32r_trampoline_init): Likewise.
16187 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
16188 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
16189 (m68k_call_m68k_read_tp): Likewise.
16190 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
16191 (microblaze_expand_divide): Likewise.
16192 * config/mips/mips.h (mips_args): Likewise.
16193 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
16194 (MIPS_ICACHE_SYNC): Likewise.
16195 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
16196 (nios2_trampoline_init): Likewise.
16197 * config/pa/pa.c (hppa_tls_call): Likewise.
16198 (pa_trampoline_init): Likewise.
16199 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
16200 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
16201 (expand_strn_compare): Likewise.
16202 (rs6000_generate_compare): Likewise.
16203 (rs6000_expand_float128_convert): Likewise.
16204 (output_profile_hook): Likewise.
16205 (rs6000_trampoline_init): Likewise.
16206 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
16207 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
16208 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
16209 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
16210 (rs6000_generate_compare): Likewise.
16211 (rs6000_expand_float128_convert): Likewise.
16212 (output_profile_hook): Likewise.
16213 (rs6000_trampoline_init): Likewise.
16214 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
16215 * config/sh/sh.c (sh_trampoline_init): Likewise.
16216 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
16217 (sparc_emit_float_lib_cmp): Likewise.
16218 (sparc32_initialize_trampoline): Likewise.
16219 (sparc64_initialize_trampoline): Likewise.
16220 (sparc_profile_hook): Likewise.
16221 * config/spu/spu.c (ea_load_store): Likewise.
16222 * config/spu/spu.md (floatunssidf2): Likewise.
16223 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
16224 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
16225 * config/visium/visium.c (expand_block_move_4): Likewise.
16226 (expand_block_move_2): Likewise.
16227 (expand_block_move_1): Likewise.
16228 (expand_block_set_4): Likewise.
16229 (expand_block_set_2): Likewise.
16230 (expand_block_set_1): Likewise.
16231 (visium_trampoline_init): Likewise.
16232 (visium_profile_hook): Likewise.
16233 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
16234 (xtensa_setup_frame_addresses): Likewise.
16235 (xtensa_trampoline_init): Likewise.
16236 * except.c (sjlj_emit_function_enter): Likewise.
16237 (sjlj_emit_function_exit): Likewise.
16238 * explow.c (allocate_dynamic_stack_space): Likewise.
16239 (probe_stack_range): Likewise.
16240 * expr.c (convert_mode_scalar): Likewise.
16241 * optabs.c (expand_binop): Likewise.
16242 (expand_twoval_binop_libfunc): Likewise.
16243 (expand_unop): Likewise.
16244 (prepare_cmp_insn): Likewise.
16245 (prepare_float_lib_cmp): Likewise.
16246 (expand_float): Likewise.
16247 (expand_fix): Likewise.
16248 (expand_fixed_convert): Likewise.
16249 (maybe_emit_sync_lock_test_and_set): Likewise.
16250 (expand_atomic_compare_and_swap): Likewise.
16251 (expand_mem_thread_fence): Likewise.
16252 (expand_atomic_fetch_op): Likewise.
16254 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
16256 * doc/generic.texi (OpenACC): Adjust URL.
16257 * doc/invoke.texi (C Dialect Options): Ditto.
16259 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
16261 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
16262 predicate for operand 1. Add (m,<S>) constraint.
16263 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
16264 Prevent memory operand 1 with register operand 2.
16266 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
16268 PR rtl-optimization/82024
16269 * combine.c (try_combine): If the combination result is a PARALLEL,
16270 and we only need to retain the SET in there that would be placed
16271 at I2, check that we can place that at I3 instead, before doing so.
16273 2017-09-01 Jakub Jelinek <jakub@redhat.com>
16276 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
16278 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
16279 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
16282 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
16283 Jeff Law <law@redhat.com>
16285 * varasm.c (bss_initializer_p): Do not put constants into .bss
16286 (categorize_decl_for_section): Handle bss_initializer_p returning
16287 false when DECL_INITIAL is NULL.
16289 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16292 * config/s390/s390.c (s390_can_inline_p): New function.
16294 2017-09-01 Jeff Law <law@redhat.com>
16296 PR tree-optimization/82052
16297 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
16298 Always initialize the returned slot after a hash table miss
16299 when INSERT is true.
16301 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
16303 * config/s390/s390.md (mem_signal_fence): Remove.
16304 * doc/md.texi (mem_signal_fence): Remove.
16305 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
16307 * target-insns.def (mem_signal_fence): Remove.
16309 2017-09-01 Jakub Jelinek <jakub@redhat.com>
16312 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
16315 * asan.c (create_odr_indicator): Strip name encoding from assembler
16316 name before appending it after __odr_asan_.
16318 2017-09-01 Martin Liska <mliska@suse.cz>
16320 PR tree-optimization/82059
16321 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
16322 frequency only when an edge is redirected.
16324 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
16326 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
16327 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
16328 (arc_conditional_register_usage): Remove ARC600 lp_count
16330 (arc_file_start): Emit Tag_ARC_CPU_variation.
16331 (arc_can_use_doloop_p): New conditions to use ZOLs.
16332 (hwloop_fail): New function.
16333 (hwloop_optimize): Likewise.
16334 (hwloop_pattern_reg): Likewise.
16335 (arc_doloop_hooks): New struct, to be used with reorg_loops.
16336 (arc_reorg_loops): New function, calls reorg_loops.
16337 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
16338 (arc600_corereg_hazard): Remove ZOL checking, case handled by
16340 (arc_loop_hazard): Remove function, functionality moved into
16342 (arc_hazard): Remove arc_loop_hazard call.
16343 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
16344 into hwloop_optimize.
16345 (arc_label_align): Remove ZOL handling.
16346 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
16347 * config/arc/arc.md (doloop_begin): Remove pattern.
16348 (doloop_begin_i): Likewise.
16349 (doloop_end_i): Likewise.
16350 (doloop_fallback): Likewise.
16351 (doloop_fallback_m): Likewise.
16352 (doloop_end): Reimplement expand.
16353 (arc_lp): New pattern for LP instruction.
16354 (loop_end): New pattern.
16355 (loop_fail): Likewise.
16356 (decrement_and_branch_until_zero): Likewise.
16357 * config/arc/arc.opt (mlpc-width): New option.
16358 * doc/invoke.texi (mlpc-width): Document option.
16360 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
16362 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
16363 (arc_ccfsm_advance): Fix checking for delay slots.
16364 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
16366 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
16368 * config/arc/arc.md (movqi_insn): Add stores to save constant long
16370 (movhi_insn): Update store instruction constraint which are saving
16371 6-bit short immediates.
16372 (movsi_insn): Consider also short scaled load operations.
16373 (zero_extendhisi2_i): Use Usd constraint instead of T.
16374 (extendhisi2_i): Add q constraint.
16375 (arc_clzsi2): Add type and length attributes.
16376 (arc_ctzsi2): Likewise.
16377 * config/arc/constraints.md (Usc): Update constraint, the
16378 assembler can parse two relocations for a single instruction.
16380 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
16382 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
16383 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
16385 2017-08-31 Olivier Hainque <hainque@adacore.com>
16387 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
16388 match as powerpc-wrs-vxworks*.
16390 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
16392 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
16393 register constraint for by-element operand.
16394 (aarch64_mls_elt_merge<mode>): Likewise.
16396 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
16398 * config/arc/arc.c (arc_can_follow_jump): Check for short
16401 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
16403 * config.gcc: Use g.opt for arc.
16404 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
16405 functionality moved to ...
16406 (legitimate_scaled_address_p): New function, ...here.
16407 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
16408 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
16409 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
16411 (arc_override_options): Handle G option.
16412 (arc_output_pic_addr_const): Correct function definition.
16413 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
16414 (arc_decl_anon_ns_mem_p): Delete.
16415 (arc_in_small_data_p): Overhaul this function to take into
16416 consideration the value given via G option.
16417 (arc_rewrite_small_data_1): Renamed and corrected old
16418 arc_rewrite_small_data function.
16419 (arc_rewrite_small_data): New function.
16420 (small_data_pattern): Don't use pic_offset_table_rtx.
16421 * config/arc/arc.h (CC1_SPEC): Recognize G option.
16422 * config/arc/simdext.md (movmisalignv2hi): Use
16423 prepare_move_operands function.
16425 (movmisalign*): Likewise.
16426 * doc/invoke.texi (ARC options): Document -G option.
16428 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
16430 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
16432 * config/arc/arc.c (arc_print_operand): Output scalled address for
16433 sdata whenever is possible.
16434 (arc_in_small_data_p): Allow sdata for 64bit datum when double
16435 load/stores are available.
16436 (compact_sda_memory_operand): Check for the alignment required by
16437 code density instructions.
16438 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
16440 * config/arc/constraints.md (Usd): Update constraint.
16441 (Us0): New constraint.
16442 (Usc): Update constraint.
16444 2017-08-31 Richard Biener <rguenther@suse.de>
16446 PR middle-end/82054
16447 * dwarf2out.c (dwarf2out_early_global_decl): Process each
16448 function only once.
16450 2017-08-31 Tamar Christina <tamar.christina@arm.com>
16452 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
16453 Resize type_signature.
16455 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
16456 Alan Hayward <alan.hayward@arm.com>
16457 David Sherwood <david.sherwood@arm.com>
16459 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
16460 subregs whose inner modes can be stored in GPRs.
16461 (aarch64_classify_index): Likewise.
16463 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
16464 Alan Hayward <alan.hayward@arm.com>
16465 David Sherwood <david.sherwood@arm.com>
16467 * config/aarch64/iterators.md (V_cmp_result): Rename to...
16468 (V_INT_EQUIV): ...this.
16469 (v_cmp_result): Rename to...
16470 (v_int_equiv): ...this.
16471 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
16472 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
16473 (copysign<mode>3): Likewise.
16474 (aarch64_simd_bsl<mode>_internal): Likewise.
16475 (aarch64_simd_bsl<mode>): Likewise.
16476 (vec_cmp<mode><mode>): Likewise.
16477 (vcond<mode><mode>): Likewise.
16478 (vcond<v_cmp_mixed><mode>): Likewise.
16479 (vcondu<mode><v_cmp_mixed>): Likewise.
16480 (aarch64_cm<optab><mode>): Likewise.
16481 (aarch64_cmtst<mode>): Likewise.
16482 (aarch64_fac<optab><mode>): Likewise.
16483 (vec_perm_const<mode>): Likewise.
16484 (vcond_mask_<mode><v_cmp_result>): Rename to...
16485 (vcond_mask_<mode><v_int_equiv>): ...this.
16486 (vec_cmp<mode><v_cmp_result>): Rename to...
16487 (vec_cmp<mode><v_int_equiv>): ...this.
16489 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
16490 Alan Hayward <alan.hayward@arm.com>
16491 David Sherwood <david.sherwood@arm.com>
16493 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
16495 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
16496 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
16497 (UNSPEC_LD4_DREG): New unspecs.
16498 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
16499 (aarch64_ld2<mode>_dreg_be): Replace with...
16500 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
16502 (aarch64_ld3<mode>_dreg_le)
16503 (aarch64_ld3<mode>_dreg_be): Replace with...
16504 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
16506 (aarch64_ld4<mode>_dreg_le)
16507 (aarch64_ld4<mode>_dreg_be): Replace with...
16508 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
16511 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16513 PR tree-optimization/81987
16514 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
16515 insert an initializer in a location not dominated by the stride
16518 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
16520 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
16521 on the entire header of the finally block in the fallthru case.
16523 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
16525 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
16527 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
16529 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
16530 rs6000_emit_move_from_cr and call renamed function.
16531 (rs6000_emit_prologue): Call renamed functions.
16532 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
16533 movesi_from_cr, remove volatile CRs.
16535 2017-08-30 Jon Beniston <jon@beniston.com>
16536 Richard Biener <rguenther@suse.de>
16538 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
16539 of VECTOR_MODE_P check.
16540 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
16541 element vector types.
16543 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16545 * df.h (df_read_modify_subreg_p): Remove in favor of...
16546 * rtl.h (read_modify_subreg_p): ...this new function. Take a
16547 const_rtx instead of an rtx.
16548 * cprop.c (local_cprop_find_used_regs): Update accordingly.
16549 * df-problems.c (df_word_lr_mark_ref): Likewise.
16550 * ira-lives.c (mark_pseudo_reg_live): Likewise.
16551 (mark_pseudo_reg_dead): Likewise.
16552 (mark_ref_dead): Likewise.
16553 * reginfo.c (init_subregs_of_mode): Likewise.
16554 * sched-deps.c (sched_analyze_1): Likewise.
16555 * df-scan.c (df_def_record_1): Likewise.
16556 (df_uses_record): Likewise.
16557 (df_read_modify_subreg_p): Remove in favor of...
16558 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
16559 const_rtx instead of an rtx.
16561 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16562 Alan Hayward <alan.hayward@arm.com>
16563 David Sherwood <david.sherwood@arm.com>
16565 * rtl.h (partial_subreg_p): New function.
16566 * caller-save.c (save_call_clobbered_regs): Use it.
16567 * calls.c (expand_call): Likewise.
16568 * combine.c (combinable_i3pat): Likewise.
16569 (simplify_set): Likewise.
16570 (make_extraction): Likewise.
16571 (make_compound_operation_int): Likewise.
16572 (gen_lowpart_or_truncate): Likewise.
16573 (force_to_mode): Likewise.
16574 (make_field_assignment): Likewise.
16575 (reg_truncated_to_mode): Likewise.
16576 (record_truncated_value): Likewise.
16577 (move_deaths): Likewise.
16578 * cse.c (record_jump_cond): Likewise.
16579 (cse_insn): Likewise.
16580 * cselib.c (cselib_lookup_1): Likewise.
16581 * expmed.c (extract_bit_field_using_extv): Likewise.
16582 * function.c (assign_parm_setup_reg): Likewise.
16583 * ifcvt.c (noce_convert_multiple_sets): Likewise.
16584 * ira-build.c (create_insn_allocnos): Likewise.
16585 * lra-coalesce.c (merge_pseudos): Likewise.
16586 * lra-constraints.c (match_reload): Likewise.
16587 (simplify_operand_subreg): Likewise.
16588 (curr_insn_transform): Likewise.
16589 * lra-lives.c (process_bb_lives): Likewise.
16590 * lra.c (new_insn_reg): Likewise.
16591 (lra_substitute_pseudo): Likewise.
16592 * regcprop.c (mode_change_ok): Likewise.
16593 (maybe_mode_change): Likewise.
16594 (copyprop_hardreg_forward_1): Likewise.
16595 * reload.c (push_reload): Likewise.
16596 (find_reloads): Likewise.
16597 (find_reloads_subreg_address): Likewise.
16598 * reload1.c (alter_reg): Likewise.
16599 (eliminate_regs_1): Likewise.
16600 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
16602 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
16604 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
16605 back to if statements, including unpack.
16607 2017-08-30 Martin Liska <mliska@suse.cz>
16609 PR inline-asm/82001
16610 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
16612 (func_checker::compare_asm_inputs_outputs): ... this function.
16613 (func_checker::compare_gimple_asm): Use the function to compare
16614 also ASM constrains.
16615 * ipa-icf-gimple.h: Rename the function.
16617 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16618 Alan Hayward <alan.hayward@arm.com>
16619 David Sherwood <david.sherwood@arm.com>
16621 * coretypes.h (complex_mode): New type.
16622 * gdbhooks.py (build_pretty_printer): Handle it.
16623 * machmode.h (complex_mode): New class.
16624 (complex_mode::includes_p): New function.
16625 (is_complex_int_mode): Likewise.
16626 (is_complex_float_mode): Likewise.
16627 * genmodes.c (get_mode_class): Handle complex mode classes.
16628 * function.c (expand_function_end): Use is_complex_int_mode.
16630 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16631 Alan Hayward <alan.hayward@arm.com>
16632 David Sherwood <david.sherwood@arm.com>
16634 * coretypes.h (scalar_mode_pod): New typedef.
16635 * gdbhooks.py (build_pretty_printer): Handle it.
16636 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
16637 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
16638 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
16639 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
16640 as_a <scalar_mode>.
16642 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16643 Alan Hayward <alan.hayward@arm.com>
16644 David Sherwood <david.sherwood@arm.com>
16646 * machmode.h (mode_for_vector): Take a scalar_mode instead
16648 * stor-layout.c (mode_for_vector): Likewise.
16649 * explow.c (promote_mode): Use as_a <scalar_mode>.
16650 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
16652 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16653 Alan Hayward <alan.hayward@arm.com>
16654 David Sherwood <david.sherwood@arm.com>
16656 * target.def (preferred_simd_mode): Take a scalar_mode
16657 instead of a machine_mode.
16658 * targhooks.h (default_preferred_simd_mode): Likewise.
16659 * targhooks.c (default_preferred_simd_mode): Likewise.
16660 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
16661 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
16662 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
16663 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
16664 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
16665 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
16666 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
16667 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
16669 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
16670 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
16671 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
16672 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
16673 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
16674 * doc/tm.texi: Regenerate.
16675 * optabs-query.c (can_vec_mask_load_store_p): Return false for
16678 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16679 Alan Hayward <alan.hayward@arm.com>
16680 David Sherwood <david.sherwood@arm.com>
16682 * target.def (scalar_mode_supported_p): Take a scalar_mode
16683 instead of a machine_mode.
16684 * targhooks.h (default_scalar_mode_supported_p): Likewise.
16685 * targhooks.c (default_scalar_mode_supported_p): Likewise.
16686 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
16687 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
16688 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
16689 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
16690 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
16691 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
16692 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
16693 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
16694 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
16695 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
16696 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
16697 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
16699 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
16700 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
16701 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
16702 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
16703 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
16705 * doc/tm.texi: Regenerate.
16707 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16708 Alan Hayward <alan.hayward@arm.com>
16709 David Sherwood <david.sherwood@arm.com>
16711 * coretypes.h (opt_scalar_mode): New typedef.
16712 * gdbhooks.py (build_pretty_printers): Handle it.
16713 * machmode.h (mode_iterator::get_2xwider): Add overload for
16715 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
16717 * expr.c (convert_mode_scalar): Likewise.
16718 * omp-low.c (omp_clause_aligned_alignment): Likewise.
16719 * optabs.c (expand_float): Likewise.
16720 (expand_fix): Likewise.
16721 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
16723 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16724 Alan Hayward <alan.hayward@arm.com>
16725 David Sherwood <david.sherwood@arm.com>
16727 * optabs.c (expand_float): Explicitly check for scalars before
16728 using a branching expansion.
16729 (expand_fix): Likewise.
16731 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16732 Alan Hayward <alan.hayward@arm.com>
16733 David Sherwood <david.sherwood@arm.com>
16735 * expr.c (convert_mode): Split scalar handling out into...
16736 (convert_mode_scalar): ...this new function. Treat the modes
16739 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16740 Alan Hayward <alan.hayward@arm.com>
16741 David Sherwood <david.sherwood@arm.com>
16743 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
16745 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
16747 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16748 Alan Hayward <alan.hayward@arm.com>
16749 David Sherwood <david.sherwood@arm.com>
16751 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
16752 rather than a machine_mode.
16753 (fixed_from_string): Likewise.
16754 (fixed_convert): Likewise.
16755 (fixed_convert_from_int): Likewise.
16756 (fixed_convert_from_real): Likewise.
16757 (real_convert_from_fixed): Likewise.
16758 * fixed-value.c (fixed_from_double_int): Likewise.
16759 (fixed_from_string): Likewise.
16760 (fixed_convert): Likewise.
16761 (fixed_convert_from_int): Likewise.
16762 (fixed_convert_from_real): Likewise.
16763 (real_convert_from_fixed): Likewise.
16764 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
16766 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16767 Alan Hayward <alan.hayward@arm.com>
16768 David Sherwood <david.sherwood@arm.com>
16770 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
16771 of separate mode class checks. Do not allow vector modes here.
16772 (immed_wide_int_const): Use as_a <scalar_mode>.
16773 * explow.c (trunc_int_for_mode): Likewise.
16774 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
16775 (wi::shwi): Likewise.
16776 (wi::min_value): Likewise.
16777 (wi::max_value): Likewise.
16778 * dwarf2out.c (loc_descriptor): Likewise.
16779 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
16780 for CONST_WIDE_INT.
16782 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16783 Alan Hayward <alan.hayward@arm.com>
16784 David Sherwood <david.sherwood@arm.com>
16786 * tree.h (SCALAR_TYPE_MODE): New macro.
16787 * expr.c (expand_expr_addr_expr_1): Use it.
16788 (expand_expr_real_2): Likewise.
16789 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
16790 (fold_convert_const_fixed_from_int): Likewise.
16791 (fold_convert_const_fixed_from_real): Likewise.
16792 (native_encode_fixed): Likewise
16793 (native_encode_complex): Likewise
16794 (native_encode_vector): Likewise.
16795 (native_interpret_fixed): Likewise.
16796 (native_interpret_real): Likewise.
16797 (native_interpret_complex): Likewise.
16798 (native_interpret_vector): Likewise.
16799 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
16800 (simd_clone_adjust_argument_types): Likewise.
16801 (simd_clone_init_simd_arrays): Likewise.
16802 (simd_clone_adjust): Likewise.
16803 * stor-layout.c (layout_type): Likewise.
16804 * tree.c (build_minus_one_cst): Likewise.
16805 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
16806 * tree-inline.c (estimate_move_cost): Likewise.
16807 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
16808 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
16809 (vectorizable_reduction): Likewise.
16810 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
16811 (vect_recog_mixed_size_cond_pattern): Likewise.
16812 (check_bool_pattern): Likewise.
16813 (adjust_bool_pattern): Likewise.
16814 (search_type_for_mask_1): Likewise.
16815 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
16816 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
16817 (vectorizable_load): Likewise.
16818 (vectorizable_store): Likewise.
16819 * ubsan.c (ubsan_encode_value): Likewise.
16820 * varasm.c (output_constant): Likewise.
16822 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16823 Alan Hayward <alan.hayward@arm.com>
16824 David Sherwood <david.sherwood@arm.com>
16826 * coretypes.h (scalar_mode): New class.
16827 * machmode.h (scalar_mode): Likewise.
16828 (scalar_mode::includes_p): New function.
16829 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
16830 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
16831 * genmodes.c (get_mode_class): Handle remaining scalar modes.
16832 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
16833 * expmed.c (store_bit_field_1): Likewise.
16834 (extract_bit_field_1): Likewise.
16835 * expr.c (write_complex_part): Likewise.
16836 (read_complex_part): Likewise.
16837 (emit_move_complex_push): Likewise.
16838 (expand_expr_real_2): Likewise.
16839 * function.c (assign_parm_setup_reg): Likewise.
16840 (assign_parms_unsplit_complex): Likewise.
16841 * optabs.c (expand_binop): Likewise.
16842 * rtlanal.c (subreg_get_info): Likewise.
16843 * simplify-rtx.c (simplify_immed_subreg): Likewise.
16844 * varasm.c (output_constant_pool_2): Likewise.
16846 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16847 Alan Hayward <alan.hayward@arm.com>
16848 David Sherwood <david.sherwood@arm.com>
16850 * expmed.c (extract_high_half): Use scalar_int_mode and remove
16852 (expmed_mult_highpart_optab): Likewise.
16853 (expmed_mult_highpart): Likewise.
16855 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16856 Alan Hayward <alan.hayward@arm.com>
16857 David Sherwood <david.sherwood@arm.com>
16859 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
16860 instead of a machine_mode.
16861 (builtin_memset_read_str): Likewise.
16862 * builtins.c (c_readstr): Likewise.
16863 (builtin_memcpy_read_str): Likewise.
16864 (builtin_strncpy_read_str): Likewise.
16865 (builtin_memset_read_str): Likewise.
16866 (builtin_memset_gen_str): Likewise.
16867 (expand_builtin_signbit): Use scalar_int_mode for local variables.
16868 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
16869 instead of a machine_mode.
16870 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
16872 (make_extraction): Likewise.
16873 (try_widen_shift_mode): Take and return scalar_int_modes instead
16875 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
16876 a scalar_int_mode instead of a machine_mode.
16877 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
16878 (avr_addr_space_pointer_mode): Likewise.
16879 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
16880 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
16881 (msp430_unwind_word_mode): Likewise.
16882 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
16883 (spu_addr_space_pointer_mode): Likewise.
16884 (spu_addr_space_address_mode): Likewise.
16885 (spu_libgcc_cmp_return_mode): Likewise.
16886 (spu_libgcc_shift_count_mode): Likewise.
16887 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
16888 (rl78_addr_space_pointer_mode): Likewise.
16889 (fl78_unwind_word_mode): Likewise.
16890 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
16892 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
16893 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
16894 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
16895 (mips_valid_pointer_mode): Likewise.
16896 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
16897 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
16898 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
16900 (ft32_addr_space_address_mode): Likewise.
16901 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
16902 scalar_int_mode instead of a machine_mode.
16903 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
16905 (m32c_addr_space_address_mode): Likewise.
16906 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
16907 (rs6000_eh_return_filter_mode): Likewise.
16908 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
16909 (rs6000_eh_return_filter_mode): Likewise.
16910 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
16911 (s390_libgcc_shift_count_mode): Likewise.
16912 (s390_unwind_word_mode): Likewise.
16913 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
16915 * target.def (mode_rep_extended): Likewise.
16916 (valid_pointer_mode): Likewise.
16917 (addr_space.valid_pointer_mode): Likewise.
16918 (eh_return_filter_mode): Return a scalar_int_mode rather than
16920 (libgcc_cmp_return_mode): Likewise.
16921 (libgcc_shift_count_mode): Likewise.
16922 (unwind_word_mode): Likewise.
16923 (addr_space.pointer_mode): Likewise.
16924 (addr_space.address_mode): Likewise.
16925 * doc/tm.texi: Regenerate.
16926 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
16928 (do_jump): Use scalar_int_mode for local variables.
16929 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
16930 rather than a machine_mode.
16931 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
16932 (scompare_loc_descriptor_wide): Likewise.
16933 (scompare_loc_descriptor_narrow): Likewise.
16934 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
16936 * except.c (sjlj_emit_dispatch_table): Likewise.
16937 (expand_builtin_eh_copy_values): Likewise.
16938 * explow.c (convert_memory_address_addr_space_1): Likewise.
16939 Take a scalar_int_mode rather than a machine_mode.
16940 (convert_memory_address_addr_space): Take a scalar_int_mode rather
16941 than a machine_mode.
16942 (memory_address_addr_space): Use scalar_int_mode for local variables.
16943 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
16944 rather than a machine_mode.
16945 * expmed.c (mask_rtx): Likewise.
16946 (init_expmed_one_conv): Likewise.
16947 (expand_mult_highpart_adjust): Likewise.
16948 (extract_high_half): Likewise.
16949 (expmed_mult_highpart_optab): Likewise.
16950 (expmed_mult_highpart): Likewise.
16951 (expand_smod_pow2): Likewise.
16952 (expand_sdiv_pow2): Likewise.
16953 (emit_store_flag_int): Likewise.
16954 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
16956 (extract_low_bits): Likewise.
16957 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
16959 * expr.c (pieces_addr::adjust): Likewise.
16960 (can_store_by_pieces): Likewise.
16961 (store_by_pieces): Likewise.
16962 (clear_by_pieces_1): Likewise.
16963 (expand_expr_addr_expr_1): Likewise.
16964 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
16965 (expand_expr_real_1): Likewise.
16966 (try_casesi): Likewise.
16967 * final.c (shorten_branches): Likewise.
16968 * fold-const.c (fold_convert_const_int_from_fixed): Change the
16969 type of "mode" to machine_mode.
16970 * internal-fn.c (expand_arith_overflow_result_store): Take a
16971 scalar_int_mode rather than a machine_mode.
16972 (expand_mul_overflow): Use scalar_int_mode for local variables.
16973 * loop-doloop.c (doloop_modify): Likewise.
16974 (doloop_optimize): Likewise.
16975 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
16976 than a machine_mode.
16977 (expand_doubleword_shift_condmove): Likewise.
16978 (expand_doubleword_shift): Likewise.
16979 (expand_doubleword_clz): Likewise.
16980 (expand_doubleword_popcount): Likewise.
16981 (expand_doubleword_parity): Likewise.
16982 (expand_absneg_bit): Use scalar_int_mode for local variables.
16983 (prepare_float_lib_cmp): Likewise.
16984 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
16985 rather than a machine_mode.
16986 (convert_memory_address_addr_space): Likewise.
16987 (get_mode_bounds): Likewise.
16988 (get_address_mode): Return a scalar_int_mode rather than a
16990 * rtlanal.c (get_address_mode): Likewise.
16991 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
16992 than a machine_mode.
16993 * targhooks.c (default_mode_rep_extended): Likewise.
16994 (default_valid_pointer_mode): Likewise.
16995 (default_addr_space_valid_pointer_mode): Likewise.
16996 (default_eh_return_filter_mode): Return a scalar_int_mode rather
16997 than a machine_mode.
16998 (default_libgcc_cmp_return_mode): Likewise.
16999 (default_libgcc_shift_count_mode): Likewise.
17000 (default_unwind_word_mode): Likewise.
17001 (default_addr_space_pointer_mode): Likewise.
17002 (default_addr_space_address_mode): Likewise.
17003 * targhooks.h (default_eh_return_filter_mode): Likewise.
17004 (default_libgcc_cmp_return_mode): Likewise.
17005 (default_libgcc_shift_count_mode): Likewise.
17006 (default_unwind_word_mode): Likewise.
17007 (default_addr_space_pointer_mode): Likewise.
17008 (default_addr_space_address_mode): Likewise.
17009 (default_mode_rep_extended): Take a scalar_int_mode rather than
17011 (default_valid_pointer_mode): Likewise.
17012 (default_addr_space_valid_pointer_mode): Likewise.
17013 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
17015 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
17016 rather than a machine_mode.
17017 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
17018 for local variables.
17019 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
17020 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
17021 than a machine_mode.
17023 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17024 Alan Hayward <alan.hayward@arm.com>
17025 David Sherwood <david.sherwood@arm.com>
17027 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
17028 the mode argument to scalar_int_mode.
17029 (do_jump_by_parts_zero_rtx): Likewise.
17030 (do_jump_by_parts_equality_rtx): Likewise.
17031 (do_jump_by_parts_greater): Take a mode argument.
17032 (do_jump_by_parts_equality): Likewise.
17033 (do_jump_1): Update calls accordingly.
17035 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17036 Alan Hayward <alan.hayward@arm.com>
17037 David Sherwood <david.sherwood@arm.com>
17039 * is-a.h (safe_dyn_cast): New function.
17040 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
17041 (jump_table_for_label): Likewise.
17042 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
17043 instead of an rtx_insn *.
17044 (shorten_branches): Use dyn_cast instead of LABEL_P and
17045 JUMP_TABLE_DATA_P. Use jump_table_for_label and
17046 rtx_jump_table_data::get_data_mode.
17047 (final_scan_insn): Likewise.
17049 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17050 Alan Hayward <alan.hayward@arm.com>
17051 David Sherwood <david.sherwood@arm.com>
17053 * combine.c (try_combine): Use is_a <scalar_int_mode> when
17054 trying to combine a full-register integer set with a subreg
17057 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17058 Alan Hayward <alan.hayward@arm.com>
17059 David Sherwood <david.sherwood@arm.com>
17061 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
17062 that is always either address_mode or pointer_mode.
17064 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17065 Alan Hayward <alan.hayward@arm.com>
17066 David Sherwood <david.sherwood@arm.com>
17068 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
17069 when the two are known to be equal.
17071 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17072 Alan Hayward <alan.hayward@arm.com>
17073 David Sherwood <david.sherwood@arm.com>
17075 * simplify-rtx.c (simplify_const_unary_operation): Use
17076 is_a <scalar_int_mode> instead of checking for a nonzero
17077 precision. Forcibly convert op_mode to a scalar_int_mode
17078 in that case. More clearly differentiate the operand and
17079 result modes and use the former when deciding what the value
17080 of a count-bits operation should be. Use is_int_mode instead
17081 of checking for a MODE_INT. Remove redundant check for whether
17082 this mode has a zero precision.
17084 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17085 Alan Hayward <alan.hayward@arm.com>
17086 David Sherwood <david.sherwood@arm.com>
17088 * optabs.c (widen_leading): Change the type of the mode argument
17089 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
17090 (widen_bswap): Likewise.
17091 (expand_parity): Likewise.
17092 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
17093 (expand_ffs): Likewise.
17094 (epand_unop): Check for scalar integer modes before calling the
17097 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17098 Alan Hayward <alan.hayward@arm.com>
17099 David Sherwood <david.sherwood@arm.com>
17101 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
17103 (expand_expr_real_1): Update call accordingly.
17105 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17106 Alan Hayward <alan.hayward@arm.com>
17107 David Sherwood <david.sherwood@arm.com>
17109 * expmed.c (store_bit_field_using_insv): Add op0_mode and
17110 value_mode arguments. Use scalar_int_mode internally.
17111 (store_bit_field_1): Rename the new integer mode from imode
17112 to op0_mode and use it instead of GET_MODE (op0). Update calls
17113 to store_split_bit_field, store_bit_field_using_insv and
17114 store_fixed_bit_field.
17115 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
17116 Use scalar_int_mode internally. Use a bit count rather than a mode
17117 when calculating the largest bit size for get_best_mode.
17118 Update calls to store_split_bit_field and store_fixed_bit_field_1.
17119 (store_fixed_bit_field_1): Add mode and value_mode arguments.
17120 Remove assertion that OP0 has a scalar integer mode.
17121 (store_split_bit_field): Add op0_mode and value_mode arguments.
17122 Update calls to extract_fixed_bit_field.
17123 (extract_bit_field_using_extv): Add an op0_mode argument.
17124 Use scalar_int_mode internally.
17125 (extract_bit_field_1): Rename the new integer mode from imode to
17126 op0_mode and use it instead of GET_MODE (op0). Update calls to
17127 extract_split_bit_field, extract_bit_field_using_extv and
17128 extract_fixed_bit_field.
17129 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
17130 to extract_split_bit_field and extract_fixed_bit_field_1.
17131 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
17132 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
17133 on the target mode.
17134 (extract_split_bit_field): Add an op0_mode argument. Update call
17135 to extract_fixed_bit_field.
17137 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17138 Alan Hayward <alan.hayward@arm.com>
17139 David Sherwood <david.sherwood@arm.com>
17141 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
17142 * explow.c (hard_function_value): Likewise.
17143 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
17144 convert_to_mode call outside the loop.
17145 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
17146 for the mode iterator. Require the mode specified by max_pieces
17148 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
17150 (copy_blkmode_to_reg): Likewise.
17151 (set_storage_via_setmem): Likewise.
17152 * optabs.c (prepare_cmp_insn): Likewise.
17153 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
17154 * stor-layout.c (finish_bitfield_representative): Likewise.
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 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
17161 * expr.c (convert_move): Use them.
17162 (convert_modes): Likewise.
17163 (store_expr_with_bounds): Likewise.
17165 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17166 Alan Hayward <alan.hayward@arm.com>
17167 David Sherwood <david.sherwood@arm.com>
17169 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
17170 parameter for the mode of "x". Remove the "known_x", "known_mode"
17171 and "known_ret" arguments. Change the type of the mode argument
17172 to scalar_int_mode.
17173 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
17174 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
17175 (reg_num_sign_bit_copies_for_combine): Likewise.
17176 * rtlanal.c (nonzero_bits1): Likewise.
17177 (num_sign_bit_copies1): Likewise.
17178 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
17179 (reg_num_sign_bit_copies_general): Likewise.
17180 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
17181 (reg_nonzero_bits_general): Likewise.
17183 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17184 Alan Hayward <alan.hayward@arm.com>
17185 David Sherwood <david.sherwood@arm.com>
17187 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
17188 than in subroutines. Return 1 for non-integer modes.
17189 (cached_num_sign_bit_copies): Change the type of the mode parameter
17190 to scalar_int_mode.
17191 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
17192 classes. Handle CONST_INT_P first and then check whether X also
17193 has a scalar integer mode. Check the same thing for inner registers
17194 of a SUBREG and for values that are being extended or truncated.
17196 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17197 Alan Hayward <alan.hayward@arm.com>
17198 David Sherwood <david.sherwood@arm.com>
17200 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
17201 in subroutines. Return the mode mask for non-integer modes.
17202 (cached_nonzero_bits): Change the type of the mode parameter
17203 to scalar_int_mode.
17204 (nonzero_bits1): Likewise. Remove early exit for other mode
17205 classes. Handle CONST_INT_P first and then check whether X
17206 also has a scalar integer mode.
17208 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17209 Alan Hayward <alan.hayward@arm.com>
17210 David Sherwood <david.sherwood@arm.com>
17212 * expr.c (widest_int_mode_for_size): Make the comment match the code.
17213 Return a scalar_int_mode and assert that the size is greater than
17215 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
17216 (op_by_pieces_d::op_by_pieces_d): Likewise.
17217 (op_by_pieces_d::run): Likewise.
17218 (can_store_by_pieces): Likewise.
17220 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17221 Alan Hayward <alan.hayward@arm.com>
17222 David Sherwood <david.sherwood@arm.com>
17224 * combine.c (extract_left_shift): Add a mode argument and update
17226 (make_compound_operation_int): Change the type of the mode parameter
17227 to scalar_int_mode and update the call to extract_left_shift.
17229 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17230 Alan Hayward <alan.hayward@arm.com>
17231 David Sherwood <david.sherwood@arm.com>
17233 * combine.c (simplify_and_const_int): Change the type of the mode
17234 parameter to scalar_int_mode.
17235 (simplify_and_const_int_1): Likewise. Update recursive call.
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 * combine.c (simplify_compare_const): Check that the mode is a
17242 scalar_int_mode (rather than VOIDmode) before testing its
17244 (simplify_comparison): Move COMPARISON_P handling out of the
17245 loop and restrict the latter part of the loop to scalar_int_modes.
17246 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
17247 and when considering SUBREG_REGs. Use is_int_mode instead of
17248 checking GET_MODE_CLASS against MODE_INT.
17250 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17251 Alan Hayward <alan.hayward@arm.com>
17252 David Sherwood <david.sherwood@arm.com>
17254 * combine.c (try_widen_shift_mode): Move check for equal modes to...
17255 (simplify_shift_const_1): ...here. Use scalar_int_mode for
17256 shift_unit_mode and for modes involved in scalar shifts.
17258 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17259 Alan Hayward <alan.hayward@arm.com>
17260 David Sherwood <david.sherwood@arm.com>
17262 * combine.c (force_int_to_mode): New function, split out from...
17263 (force_to_mode): ...here. Keep xmode up-to-date and use it
17264 instead of GET_MODE (x).
17266 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17267 Alan Hayward <alan.hayward@arm.com>
17268 David Sherwood <david.sherwood@arm.com>
17270 * optabs-query.h (extraction_insn::struct_mode): Change type to
17271 opt_scalar_int_mode and update comment.
17272 (extraction_insn::field_mode): Change type to scalar_int_mode.
17273 (extraction_insn::pos_mode): Likewise.
17274 * combine.c (make_extraction): Update accordingly.
17275 * optabs-query.c (get_traditional_extraction_insn): Likewise.
17276 (get_optab_extraction_insn): Likewise.
17277 * recog.c (simplify_while_replacing): Likewise.
17278 * expmed.c (narrow_bit_field_mem): Change the type of the mode
17279 parameter to opt_scalar_int_mode.
17281 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17282 Alan Hayward <alan.hayward@arm.com>
17283 David Sherwood <david.sherwood@arm.com>
17285 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
17286 to a scalar_int_mode instead of a machine_mode.
17287 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
17288 (get_best_mode): Return a boolean and use a pointer argument to store
17289 the selected mode. Replace the limit mode parameter with a bit limit.
17290 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
17291 for the values returned by bit_field_mode_iterator::next_mode.
17292 (store_bit_field): Update call to get_best_mode.
17293 (store_fixed_bit_field): Likewise.
17294 (extract_fixed_bit_field): Likewise.
17295 * expr.c (optimize_bitfield_assignment_op): Likewise.
17296 * fold-const.c (optimize_bit_field_compare): Likewise.
17297 (fold_truth_andor_1): Likewise.
17298 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
17299 Update for new type of m_mode.
17300 (get_best_mode): As above.
17302 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17303 Alan Hayward <alan.hayward@arm.com>
17304 David Sherwood <david.sherwood@arm.com>
17306 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
17307 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
17308 (store_bit_field): Check is_a <scalar_int_mode> before calling
17309 strict_volatile_bitfield_p.
17310 (extract_bit_field): Likewise.
17312 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17313 Alan Hayward <alan.hayward@arm.com>
17314 David Sherwood <david.sherwood@arm.com>
17316 * target.def (cstore_mode): Return a scalar_int_mode.
17317 * doc/tm.texi: Regenerate.
17318 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
17319 * targhooks.h (default_cstore_mode): Likewise.
17320 * targhooks.c (default_cstore_mode): Likewise, using a forced
17322 * expmed.c (emit_cstore): Expect the target of the cstore to be
17325 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17326 Alan Hayward <alan.hayward@arm.com>
17327 David Sherwood <david.sherwood@arm.com>
17329 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
17331 (niter_desc): Likewise mode.
17332 (iv_analyze): Add a mode parameter.
17334 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
17335 and change its type to scalar_int_mode.
17336 * loop-iv.c: Update commentary at head of file.
17337 (iv_constant): Pass the mode paraeter before the rtx it describes
17338 and change its type to scalar_int_mode. Remove VOIDmode handling.
17339 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
17340 (iv_extend): Likewise.
17341 (shorten_into_mode): Likewise.
17342 (iv_add): Use scalar_int_mode.
17343 (iv_mult): Likewise.
17344 (iv_shift): Likewise.
17345 (canonicalize_iv_subregs): Likewise.
17346 (get_biv_step_1): Pass the outer_mode parameter before the rtx
17347 it describes and change its mode to scalar_int_mode. Also change
17348 the type of the returned inner_mode to scalar_int_mode.
17349 (get_biv_step): Likewise, turning outer_mode from a pointer
17350 into a direct parameter. Update call to get_biv_step_1.
17351 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
17352 iv_constant and get_biv_step.
17353 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
17354 and change its type to scalar_int_mode. Don't initialise iv->mode
17355 to VOIDmode and remove later checks for its still being VOIDmode.
17356 Update calls to iv_analyze_op and iv_analyze_expr. Check
17357 is_a <scalar_int_mode> when changing the mode under consideration.
17358 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
17359 Update call to iv_analyze_expr.
17360 (iv_analyze_op): Add a mode parameter. Reject subregs whose
17361 inner register is not also a scalar_int_mode. Update call to
17363 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
17364 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
17365 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
17366 separate mode class checks. Update calls to iv_analyze. Remove
17367 fix-up of VOIDmodes after iv_analyze_biv.
17368 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
17369 don't have a scalar_int_mode. Update call to biv_p.
17371 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17372 Alan Hayward <alan.hayward@arm.com>
17373 David Sherwood <david.sherwood@arm.com>
17375 * cfgexpand.c (convert_debug_memory_address): Use
17376 as_a <scalar_int_mode>.
17377 * combine.c (expand_compound_operation): Likewise.
17378 (make_extraction): Likewise.
17379 (change_zero_ext): Likewise.
17380 (simplify_comparison): Likewise.
17381 * cse.c (cse_insn): Likewise.
17382 * dwarf2out.c (minmax_loc_descriptor): Likewise.
17383 (mem_loc_descriptor): Likewise.
17384 (loc_descriptor): Likewise.
17385 * expmed.c (init_expmed_one_mode): Likewise.
17386 (synth_mult): Likewise.
17387 (emit_store_flag_1): Likewise.
17388 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
17389 of a comparison with size.
17390 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
17391 (reduce_to_bit_field_precision): Likewise.
17392 * function.c (expand_function_end): Likewise.
17393 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
17394 * loop-doloop.c (doloop_modify): Likewise.
17395 * optabs.c (expand_binop): Likewise.
17396 (expand_unop): Likewise.
17397 (expand_copysign_absneg): Likewise.
17398 (prepare_cmp_insn): Likewise.
17399 (maybe_legitimize_operand): Likewise.
17400 * recog.c (const_scalar_int_operand): Likewise.
17401 * rtlanal.c (get_address_mode): Likewise.
17402 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17403 (simplify_cond_clz_ctz): Likewise.
17404 * tree-nested.c (get_nl_goto_field): Likewise.
17405 * tree.c (build_vector_type_for_mode): Likewise.
17406 * var-tracking.c (use_narrower_mode): Likewise.
17408 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17409 Alan Hayward <alan.hayward@arm.com>
17410 David Sherwood <david.sherwood@arm.com>
17412 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
17413 * builtins.c (expand_builtin_signbit): Use it.
17414 * cfgexpand.c (expand_debug_expr): Likewise.
17415 * dojump.c (do_jump): Likewise.
17416 (do_compare_and_jump): Likewise.
17417 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
17418 * expmed.c (make_tree): Likewise.
17419 * expr.c (expand_expr_real_2): Likewise.
17420 (expand_expr_real_1): Likewise.
17421 (try_casesi): Likewise.
17422 * fold-const-call.c (fold_const_call_ss): Likewise.
17423 * fold-const.c (unextend): Likewise.
17424 (extract_muldiv_1): Likewise.
17425 (fold_single_bit_test): Likewise.
17426 (native_encode_int): Likewise.
17427 (native_encode_string): Likewise.
17428 (native_interpret_int): Likewise.
17429 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
17430 * internal-fn.c (expand_addsub_overflow): Likewise.
17431 (expand_neg_overflow): Likewise.
17432 (expand_mul_overflow): Likewise.
17433 (expand_arith_overflow): Likewise.
17434 * match.pd: Likewise.
17435 * stor-layout.c (layout_type): Likewise.
17436 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
17437 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
17438 * tree-ssanames.c (get_range_info): Likewise.
17439 * tree-switch-conversion.c (array_value_type) Likewise.
17440 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
17441 (vect_recog_divmod_pattern): Likewise.
17442 (vect_recog_mixed_size_cond_pattern): Likewise.
17443 * tree-vrp.c (extract_range_basic): Likewise.
17444 (simplify_float_conversion_using_ranges): Likewise.
17445 * tree.c (int_fits_type_p): Likewise.
17446 * ubsan.c (instrument_bool_enum_load): Likewise.
17447 * varasm.c (mergeable_string_section): Likewise.
17448 (narrowing_initializer_constant_valid_p): Likewise.
17449 (output_constant): Likewise.
17451 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17452 Alan Hayward <alan.hayward@arm.com>
17453 David Sherwood <david.sherwood@arm.com>
17455 * machmode.h (NARROWEST_INT_MODE): New macro.
17456 * expr.c (alignment_for_piecewise_move): Use it instead of
17457 GET_CLASS_NARROWEST_MODE (MODE_INT).
17458 (push_block): Likewise.
17459 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
17461 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
17463 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17464 Alan Hayward <alan.hayward@arm.com>
17465 David Sherwood <david.sherwood@arm.com>
17467 * postreload.c (move2add_valid_value_p): Change the type of the
17468 mode parameter to scalar_int_mode.
17469 (move2add_use_add2_insn): Add a mode parameter and use it instead
17471 (move2add_use_add3_insn): Likewise.
17472 (reload_cse_move2add): Update accordingly.
17474 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17475 Alan Hayward <alan.hayward@arm.com>
17476 David Sherwood <david.sherwood@arm.com>
17478 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
17480 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
17481 * optabs.c (expand_unop): Likewise.
17483 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17484 Alan Hayward <alan.hayward@arm.com>
17485 David Sherwood <david.sherwood@arm.com>
17487 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
17488 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
17489 (popcount_loc_descriptor): Likewise.
17490 (bswap_loc_descriptor): Likewise.
17491 (rotate_loc_descriptor): Likewise.
17492 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
17493 calling the functions above.
17495 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17496 Alan Hayward <alan.hayward@arm.com>
17497 David Sherwood <david.sherwood@arm.com>
17499 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
17501 (try_combine): Likewise.
17502 (simplify_if_then_else): Likewise.
17503 * cse.c (cse_insn): Likewise.
17504 * dwarf2out.c (mem_loc_descriptor): Likewise.
17505 * emit-rtl.c (gen_lowpart_common): Likewise.
17506 * simplify-rtx.c (simplify_truncation): Likewise.
17507 (simplify_binary_operation_1): Likewise.
17508 (simplify_const_relational_operation): Likewise.
17509 (simplify_ternary_operation): Likewise.
17510 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
17512 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17513 Alan Hayward <alan.hayward@arm.com>
17514 David Sherwood <david.sherwood@arm.com>
17516 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
17517 * reload.c (push_reload): Likewise.
17518 (find_reloads): Likewise.
17520 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17521 Alan Hayward <alan.hayward@arm.com>
17522 David Sherwood <david.sherwood@arm.com>
17524 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
17525 (make_compound_operation_int): Likewise.
17526 (change_zero_ext): Likewise.
17527 * expr.c (convert_move): Likewise.
17528 (convert_modes): Likewise.
17529 * fwprop.c (forward_propagate_subreg): Likewise.
17530 * loop-iv.c (get_biv_step_1): Likewise.
17531 * optabs.c (widen_operand): Likewise.
17532 * postreload.c (move2add_valid_value_p): Likewise.
17533 * recog.c (simplify_while_replacing): Likewise.
17534 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17535 (simplify_binary_operation_1): Likewise. Remove redundant
17536 mode equality check.
17538 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17539 Alan Hayward <alan.hayward@arm.com>
17540 David Sherwood <david.sherwood@arm.com>
17542 * combine.c (combine_simplify_rtx): Add checks for
17543 is_a <scalar_int_mode>.
17544 (simplify_if_then_else): Likewise.
17545 (make_field_assignment): Likewise.
17546 (simplify_comparison): Likewise.
17547 * ifcvt.c (noce_try_bitop): Likewise.
17548 * loop-invariant.c (canonicalize_address_mult): Likewise.
17549 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17551 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17552 Alan Hayward <alan.hayward@arm.com>
17553 David Sherwood <david.sherwood@arm.com>
17555 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
17556 is_a <scalar_int_mode> instead of != BLKmode.
17558 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17559 Alan Hayward <alan.hayward@arm.com>
17560 David Sherwood <david.sherwood@arm.com>
17562 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
17563 instead of != VOIDmode.
17564 * combine.c (if_then_else_cond): Likewise.
17565 (change_zero_ext): Likewise.
17566 * dwarf2out.c (mem_loc_descriptor): Likewise.
17567 (loc_descriptor): Likewise.
17568 * rtlanal.c (canonicalize_condition): Likewise.
17569 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
17571 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17572 Alan Hayward <alan.hayward@arm.com>
17573 David Sherwood <david.sherwood@arm.com>
17575 * simplify-rtx.c (simplify_binary_operation_1): Use
17576 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
17578 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17579 Alan Hayward <alan.hayward@arm.com>
17580 David Sherwood <david.sherwood@arm.com>
17582 * wide-int.h (int_traits<unsigned char>) New class.
17583 (int_traits<unsigned short>) Likewise.
17584 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
17585 Use GET_MODE_UNIT_PRECISION and remove redundant test for
17587 * combine.c (set_nonzero_bits_and_sign_copies): Use
17588 is_a <scalar_int_mode>.
17589 (find_split_point): Likewise.
17590 (combine_simplify_rtx): Likewise.
17591 (simplify_logical): Likewise.
17592 (expand_compound_operation): Likewise.
17593 (expand_field_assignment): Likewise.
17594 (make_compound_operation): Likewise.
17595 (extended_count): Likewise.
17596 (change_zero_ext): Likewise.
17597 (simplify_comparison): Likewise.
17598 * dwarf2out.c (scompare_loc_descriptor): Likewise.
17599 (ucompare_loc_descriptor): Likewise.
17600 (minmax_loc_descriptor): Likewise.
17601 (mem_loc_descriptor): Likewise.
17602 (loc_descriptor): Likewise.
17603 * expmed.c (init_expmed_one_mode): Likewise.
17604 * lra-constraints.c (lra_constraint_offset): Likewise.
17605 * optabs.c (prepare_libcall_arg): Likewise.
17606 * postreload.c (move2add_note_store): Likewise.
17607 * reload.c (operands_match_p): Likewise.
17608 * rtl.h (load_extend_op): Likewise.
17609 * rtlhooks.c (gen_lowpart_general): Likewise.
17610 * simplify-rtx.c (simplify_truncation): Likewise.
17611 (simplify_unary_operation_1): Likewise.
17612 (simplify_binary_operation_1): Likewise.
17613 (simplify_const_binary_operation): Likewise.
17614 (simplify_const_relational_operation): Likewise.
17615 (simplify_subreg): Likewise.
17616 * stor-layout.c (bitwise_mode_for_mode): Likewise.
17617 * var-tracking.c (adjust_mems): Likewise.
17618 (prepare_call_arguments): Likewise.
17620 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17621 Alan Hayward <alan.hayward@arm.com>
17622 David Sherwood <david.sherwood@arm.com>
17624 * machmode.h (is_int_mode): New fuction.
17625 * combine.c (find_split_point): Use it.
17626 (combine_simplify_rtx): Likewise.
17627 (simplify_if_then_else): Likewise.
17628 (simplify_set): Likewise.
17629 (simplify_shift_const_1): Likewise.
17630 (simplify_comparison): Likewise.
17631 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
17632 * cse.c (notreg_cost): Likewise.
17633 (cse_insn): Likewise.
17634 * cselib.c (cselib_lookup_1): Likewise.
17635 * dojump.c (do_jump_1): Likewise.
17636 (do_compare_rtx_and_jump): Likewise.
17637 * dse.c (get_call_args): Likewise.
17638 * dwarf2out.c (rtl_for_decl_init): Likewise.
17639 (native_encode_initializer): Likewise.
17640 * expmed.c (emit_store_flag_1): Likewise.
17641 (emit_store_flag): Likewise.
17642 * expr.c (convert_modes): Likewise.
17643 (store_field): Likewise.
17644 (expand_expr_real_1): Likewise.
17645 * fold-const.c (fold_read_from_constant_string): Likewise.
17646 * gimple-ssa-sprintf.c (get_format_string): Likewise.
17647 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
17648 * optabs.c (expand_binop): Likewise.
17649 (expand_unop): Likewise.
17650 (expand_abs_nojump): Likewise.
17651 (expand_one_cmpl_abs_nojump): Likewise.
17652 * simplify-rtx.c (mode_signbit_p): Likewise.
17653 (val_signbit_p): Likewise.
17654 (val_signbit_known_set_p): Likewise.
17655 (val_signbit_known_clear_p): Likewise.
17656 (simplify_relational_operation_1): Likewise.
17657 * tree.c (vector_type_mode): Likewise.
17659 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17660 Alan Hayward <alan.hayward@arm.com>
17661 David Sherwood <david.sherwood@arm.com>
17663 * machmode.h (smallest_mode_for_size): Fix formatting.
17664 (smallest_int_mode_for_size): New function.
17665 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
17666 instead of smallest_mode_for_size.
17667 * combine.c (make_extraction): Likewise.
17668 * config/arc/arc.c (arc_expand_movmem): Likewise.
17669 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
17670 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
17671 * config/s390/s390.c (s390_expand_insv): Likewise.
17672 * config/sparc/sparc.c (assign_int_registers): Likewise.
17673 * config/spu/spu.c (spu_function_value): Likewise.
17674 (spu_function_arg): Likewise.
17675 * coverage.c (get_gcov_type): Likewise.
17676 (get_gcov_unsigned_t): Likewise.
17677 * dse.c (find_shift_sequence): Likewise.
17678 * expmed.c (store_bit_field_1): Likewise.
17679 * expr.c (convert_move): Likewise.
17680 (store_field): Likewise.
17681 * internal-fn.c (expand_arith_overflow): Likewise.
17682 * optabs-query.c (get_best_extraction_insn): Likewise.
17683 * optabs.c (expand_twoval_binop_libfunc): Likewise.
17684 * stor-layout.c (layout_type): Likewise.
17685 (initialize_sizetypes): Likewise.
17686 * targhooks.c (default_get_mask_mode): Likewise.
17687 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
17689 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17690 Alan Hayward <alan.hayward@arm.com>
17691 David Sherwood <david.sherwood@arm.com>
17693 * machmode.h (opt_mode::else_blk): New function.
17694 (int_mode_for_mode): Declare.
17695 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
17696 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
17698 * cfgexpand.c (expand_debug_expr): Likewise.
17699 * combine.c (gen_lowpart_or_truncate): Likewise.
17700 (gen_lowpart_for_combine): Likewise.
17701 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
17702 * config/avr/avr.c (avr_to_int_mode): Likewise.
17703 (avr_out_plus_1): Likewise.
17704 (avr_out_plus): Likewise.
17705 (avr_out_round): Likewise.
17706 * config/i386/i386.c (ix86_split_to_parts): Likewise.
17707 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
17708 (s390_expand_vcond): Likewise.
17709 * config/spu/spu.c (spu_split_immediate): Likewise.
17710 (spu_expand_mov): Likewise.
17711 * dse.c (get_stored_val): Likewise.
17712 * expmed.c (store_bit_field_1): Likewise.
17713 (convert_extracted_bit_field): Use int_mode_for_mode instead of
17715 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
17716 (extract_low_bits): Likewise.
17717 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
17718 handling rather than repeating the check.
17719 (emit_group_store): Likewise.
17720 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
17721 * optabs.c (expand_absneg_bit): Likewise.
17722 (expand_copysign_absneg): Likewise.
17723 (expand_copysign_bit): Likewise.
17724 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
17725 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
17726 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
17727 * var-tracking.c (prepare_call_arguments): Likewise.
17728 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
17729 int_mode_for_mode instead of mode_for_size.
17730 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
17732 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17733 Alan Hayward <alan.hayward@arm.com>
17734 David Sherwood <david.sherwood@arm.com>
17736 * machmode.h (int_mode_for_size): New function.
17737 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
17738 instead of mode_for_size.
17739 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
17741 * combine.c (expand_field_assignment): Use int_mode_for_size
17742 instead of mode_for_size.
17743 (make_extraction): Likewise.
17744 (simplify_shift_const_1): Likewise.
17745 (simplify_comparison): Likewise.
17746 * dojump.c (do_jump): Likewise.
17747 * dwarf2out.c (mem_loc_descriptor): Likewise.
17748 * emit-rtl.c (init_derived_machine_modes): Likewise.
17749 * expmed.c (flip_storage_order): Likewise.
17750 (convert_extracted_bit_field): Likewise.
17751 * expr.c (copy_blkmode_from_reg): Likewise.
17752 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
17753 * internal-fn.c (expand_mul_overflow): Likewise.
17754 * lower-subreg.c (simple_move): Likewise.
17755 * optabs-libfuncs.c (init_optabs): Likewise.
17756 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17757 * tree.c (vector_type_mode): Likewise.
17758 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
17759 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
17760 * tree-vect-generic.c (expand_vector_parallel): Likewise.
17761 * tree-vect-stmts.c (vectorizable_load): Likewise.
17762 (vectorizable_store): Likewise.
17764 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17765 Alan Hayward <alan.hayward@arm.com>
17766 David Sherwood <david.sherwood@arm.com>
17768 * coretypes.h (pod_mode): New type.
17769 (scalar_int_mode_pod): New typedef.
17770 * machmode.h (pod_mode): New class.
17771 (int_n_data_t::m): Change type to scalar_int_mode_pod.
17772 * genmodes.c (emit_mode_int_n): Update accordingly.
17773 * lower-subreg.h (target_lower_subreg): Change type to
17774 scalar_int_mode_pod.
17775 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
17776 scalar_int_mode_pod.
17778 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17779 Alan Hayward <alan.hayward@arm.com>
17780 David Sherwood <david.sherwood@arm.com>
17782 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
17783 machine_mode to scalar_int_mode.
17784 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
17785 (rs6000_option_override_internal): Remove cast to int.
17786 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
17787 machine_mode to scalar_int_mode.
17788 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
17789 (rs6000_option_override_internal): Remove cast to int.
17790 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
17791 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
17793 * config/s390/s390.c (s390_expand_builtin): Likewise.
17794 * coretypes.h (scalar_int_mode): New type.
17795 (opt_scalar_int_mode): New typedef.
17796 * machmode.h (scalar_int_mode): New class.
17797 (scalar_int_mode::includes_p): New function.
17798 (byte_mode): Change type to scalar_int_mode.
17799 (word_mode): Likewise.
17800 (ptr_mode): Likewise.
17801 * emit-rtl.c (byte_mode): Likewise.
17802 (word_mode): Likewise.
17803 (ptr_mode): Likewise.
17804 (init_derived_machine_modes): Update accordingly.
17805 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
17806 and MODE_PARTIAL_INT.
17807 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
17808 opt_scalar_int_mode.
17810 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17811 Alan Hayward <alan.hayward@arm.com>
17812 David Sherwood <david.sherwood@arm.com>
17814 * target.def (libgcc_floating_mode_supported_p): Take a
17816 * doc/tm.texi: Regenerate.
17817 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
17819 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
17820 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
17823 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17824 Alan Hayward <alan.hayward@arm.com>
17825 David Sherwood <david.sherwood@arm.com>
17827 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
17828 * doc/tm.texi: Regenerate.
17829 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
17830 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
17831 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
17832 * targhooks.h (default_floatn_mode): Likewise.
17833 * targhooks.c (default_floatn_mode): Likewise.
17834 * tree.c (build_common_tree_nodes): Update accordingly.
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 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
17841 (mode_iterator::iterate_p): Likewise.
17842 (mode_iterator::get_wider): Likewise.
17843 * expr.c (init_expr_target): Use opt_scalar_float_mode.
17845 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17846 Alan Hayward <alan.hayward@arm.com>
17847 David Sherwood <david.sherwood@arm.com>
17849 * coretypes.h (opt_scalar_float_mode): New typedef.
17850 * machmode.h (float_mode_for_size): New function.
17851 * emit-rtl.c (double_mode): Delete.
17852 (init_emit_once): Use float_mode_for_size.
17853 * stor-layout.c (layout_type): Likewise.
17854 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
17856 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17857 Alan Hayward <alan.hayward@arm.com>
17858 David Sherwood <david.sherwood@arm.com>
17860 * output.h (assemble_real): Take a scalar_float_mode.
17861 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
17862 * config/arm/arm.md (consttable_4): Likewise.
17863 (consttable_8): Likewise.
17864 (consttable_16): Likewise.
17865 * config/mips/mips.md (consttable_float): Likewise.
17866 * config/s390/s390.c (s390_output_pool_entry): Likewise.
17867 * varasm.c (assemble_real): Take a scalar_float_mode.
17868 (output_constant_pool_2): Update accordingly.
17869 (output_constant): Likewise.
17871 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17872 Alan Hayward <alan.hayward@arm.com>
17873 David Sherwood <david.sherwood@arm.com>
17875 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
17876 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
17877 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
17878 (native_encode_real): Likewise.
17879 (native_interpret_real): Likewise.
17880 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
17881 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
17883 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17884 Alan Hayward <alan.hayward@arm.com>
17885 David Sherwood <david.sherwood@arm.com>
17887 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
17888 <scalar_float_mode>. Simplify.
17889 (gen_extend_conv_libfunc): Likewise.
17891 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17892 Alan Hayward <alan.hayward@arm.com>
17893 David Sherwood <david.sherwood@arm.com>
17895 * coretypes.h (scalar_float_mode): New type.
17896 * machmode.h (mode_traits::from_int): Use machine_mode if
17897 USE_ENUM_MODES is defined.
17898 (is_a): New function.
17900 (dyn_cast): Likewise.
17901 (scalar_float_mode): New class.
17902 (scalar_float_mode::includes_p): New function.
17903 (is_float_mode): Likewise.
17904 * gdbhooks.py (MachineModePrinter): New class.
17905 (build_pretty_printer): Use it for scalar_float_mode.
17906 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
17907 (format_helper::format_helper): Turn into a template.
17908 * genmodes.c (get_mode_class): New function.
17909 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
17910 or machine_mode if none.
17911 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
17912 as_a <scalar_float_mode>.
17913 * dwarf2out.c (mem_loc_descriptor): Likewise.
17914 (insert_float): Likewise.
17915 (add_const_value_attribute): Likewise.
17916 * simplify-rtx.c (simplify_immed_subreg): Likewise.
17917 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
17918 (expand_unop): Update accordingly.
17919 (expand_abs_nojump): Likewise.
17920 (expand_copysign_absneg): Take a scalar_float_mode.
17921 (expand_copysign_bit): Likewise.
17922 (expand_copysign): Update accordingly.
17924 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17925 Alan Hayward <alan.hayward@arm.com>
17926 David Sherwood <david.sherwood@arm.com>
17928 * coretypes.h (opt_mode): New class.
17929 * machmode.h (opt_mode): Likewise.
17930 (opt_mode::else_void): New function.
17931 (opt_mode::require): Likewise.
17932 (opt_mode::exists): Likewise.
17933 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
17934 (GET_MODE_2XWIDER_MODE): Likewise.
17935 (mode_iterator::get_wider): Update accordingly.
17936 (mode_iterator::get_2xwider): Likewise.
17937 (mode_iterator::get_known_wider): Likewise, turning into a template.
17938 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
17939 forcing a wider mode to exist.
17940 * config/cr16/cr16.h (LONG_REG_P): Likewise.
17941 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
17942 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
17943 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
17944 * lower-subreg.c (init_lower_subreg): Likewise.
17945 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
17946 on the final iteration.
17947 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
17948 a wider mode exists before asking for a move pattern.
17949 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
17950 forcing a wider mode to exist.
17951 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
17952 returning false if no such mode exists.
17953 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
17954 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
17955 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
17956 Avoid checking for a MODE_INT if we already know the mode is not a
17958 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
17959 forcing a wider mode to exist.
17960 (expmed_mult_highpart_optab): Likewise.
17961 (expmed_mult_highpart): Likewise.
17962 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
17964 * lto-streamer-in.c (lto_input_mode_table): Likewise.
17965 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
17966 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
17967 * internal-fn.c (expand_mul_overflow): Update use of
17968 GET_MODE_2XWIDER_MODE.
17969 * omp-low.c (omp_clause_aligned_alignment): Likewise.
17970 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
17971 GET_MODE_WIDER_MODE.
17972 (convert_plusminus_to_widen): Likewise.
17973 * tree-switch-conversion.c (array_value_type): Likewise.
17974 * var-tracking.c (emit_note_insn_var_location): Likewise.
17975 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
17976 Return false inside rather than outside the loop if no wider mode
17978 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
17979 and GET_MODE_2XWIDER_MODE
17980 (can_compare_p): Use else_void.
17981 * gdbhooks.py (OptMachineModePrinter): New class.
17982 (build_pretty_printer): Use it for opt_mode.
17984 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17985 Alan Hayward <alan.hayward@arm.com>
17986 David Sherwood <david.sherwood@arm.com>
17988 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
17989 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
17991 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17992 Alan Hayward <alan.hayward@arm.com>
17993 David Sherwood <david.sherwood@arm.com>
17995 * machmode.h (mode_traits): New structure.
17996 (get_narrowest_mode): New function.
17997 (mode_iterator::start): Likewise.
17998 (mode_iterator::iterate_p): Likewise.
17999 (mode_iterator::get_wider): Likewise.
18000 (mode_iterator::get_known_wider): Likewise.
18001 (mode_iterator::get_2xwider): Likewise.
18002 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
18003 (FOR_EACH_MODE): Likewise.
18004 (FOR_EACH_MODE_FROM): Likewise.
18005 (FOR_EACH_MODE_UNTIL): Likewise.
18006 (FOR_EACH_WIDER_MODE): Likewise.
18007 (FOR_EACH_2XWIDER_MODE): Likewise.
18008 * builtins.c (expand_builtin_strlen): Use new mode iterators.
18009 * combine.c (simplify_comparison): Likewise
18010 * config/i386/i386.c (type_natural_mode): Likewise.
18011 * cse.c (cse_insn): Likewise.
18012 * dse.c (find_shift_sequence): Likewise.
18013 * emit-rtl.c (init_derived_machine_modes): Likewise.
18014 (init_emit_once): Likewise.
18015 * explow.c (hard_function_value): Likewise.
18016 * expmed.c (extract_fixed_bit_field_1): Likewise.
18017 (extract_bit_field_1): Likewise.
18018 (expand_divmod): Likewise.
18019 (emit_store_flag_1): Likewise.
18020 * expr.c (init_expr_target): Likewise.
18021 (convert_move): Likewise.
18022 (alignment_for_piecewise_move): Likewise.
18023 (widest_int_mode_for_size): Likewise.
18024 (emit_block_move_via_movmem): Likewise.
18025 (copy_blkmode_to_reg): Likewise.
18026 (set_storage_via_setmem): Likewise.
18027 (compress_float_constant): Likewise.
18028 * omp-low.c (omp_clause_aligned_alignment): Likewise.
18029 * optabs-query.c (get_best_extraction_insn): Likewise.
18030 * optabs.c (expand_binop): Likewise.
18031 (expand_twoval_unop): Likewise.
18032 (expand_twoval_binop): Likewise.
18033 (widen_leading): Likewise.
18034 (widen_bswap): Likewise.
18035 (expand_parity): Likewise.
18036 (expand_unop): Likewise.
18037 (prepare_cmp_insn): Likewise.
18038 (prepare_float_lib_cmp): Likewise.
18039 (expand_float): Likewise.
18040 (expand_fix): Likewise.
18041 (expand_sfix_optab): Likewise.
18042 * postreload.c (move2add_use_add2_insn): Likewise.
18043 * reg-stack.c (reg_to_stack): Likewise.
18044 * reginfo.c (choose_hard_reg_mode): Likewise.
18045 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
18046 * stor-layout.c (mode_for_size): Likewise.
18047 (smallest_mode_for_size): Likewise.
18048 (mode_for_vector): Likewise.
18049 (finish_bitfield_representative): Likewise.
18050 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
18051 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
18052 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
18053 * var-tracking.c (prepare_call_arguments): Likewise.
18055 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18056 Alan Hayward <alan.hayward@arm.com>
18057 David Sherwood <david.sherwood@arm.com>
18059 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
18060 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
18061 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
18062 * machmode.h (mode_size): Move earlier in file.
18063 (mode_precision): Likewise.
18064 (mode_inner): Likewise.
18065 (mode_nunits): Likewise.
18066 (mode_unit_size): Likewise.
18067 (unit_unit_precision): Likewise.
18068 (mode_wider): Likewise.
18069 (mode_2xwider): Likewise.
18070 (machine_mode): New class.
18071 (mode_to_bytes): New function.
18072 (mode_to_bits): Likewise.
18073 (mode_to_precision): Likewise.
18074 (mode_to_inner): Likewise.
18075 (mode_to_unit_size): Likewise.
18076 (mode_to_unit_precision): Likewise.
18077 (mode_to_nunits): Likewise.
18078 (GET_MODE_SIZE): Use mode_to_bytes.
18079 (GET_MODE_BITSIZE): Use mode_to_bits.
18080 (GET_MODE_PRECISION): Use mode_to_precision.
18081 (GET_MODE_INNER): Use mode_to_inner.
18082 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
18083 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
18084 (GET_MODE_NUNITS): Use mode_to_nunits.
18085 * system.h (ALWAYS_INLINE): New macro.
18086 * config/powerpcspe/powerpcspe-c.c
18087 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
18088 int for arg1_mode and arg2_mode.
18090 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18091 Alan Hayward <alan.hayward@arm.com>
18092 David Sherwood <david.sherwood@arm.com>
18094 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
18095 Prefix mode names with E_ in case statements.
18096 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18097 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
18098 (aarch64_split_simd_move): Likewise.
18099 (aarch64_gen_storewb_pair): Likewise.
18100 (aarch64_gen_loadwb_pair): Likewise.
18101 (aarch64_gen_store_pair): Likewise.
18102 (aarch64_gen_load_pair): Likewise.
18103 (aarch64_get_condition_code_1): Likewise.
18104 (aarch64_constant_pool_reload_icode): Likewise.
18105 (get_rsqrte_type): Likewise.
18106 (get_rsqrts_type): Likewise.
18107 (get_recpe_type): Likewise.
18108 (get_recps_type): Likewise.
18109 (aarch64_gimplify_va_arg_expr): Likewise.
18110 (aarch64_simd_container_mode): Likewise.
18111 (aarch64_emit_load_exclusive): Likewise.
18112 (aarch64_emit_store_exclusive): Likewise.
18113 (aarch64_expand_compare_and_swap): Likewise.
18114 (aarch64_gen_atomic_cas): Likewise.
18115 (aarch64_emit_bic): Likewise.
18116 (aarch64_emit_atomic_swap): Likewise.
18117 (aarch64_emit_atomic_load_op): Likewise.
18118 (aarch64_evpc_trn): Likewise.
18119 (aarch64_evpc_uzp): Likewise.
18120 (aarch64_evpc_zip): Likewise.
18121 (aarch64_evpc_ext): Likewise.
18122 (aarch64_evpc_rev): Likewise.
18123 (aarch64_evpc_dup): Likewise.
18124 (aarch64_gen_ccmp_first): Likewise.
18125 (aarch64_gen_ccmp_next): Likewise.
18126 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
18127 (alpha_emit_xfloating_libcall): Likewise.
18128 (emit_insxl): Likewise.
18129 (alpha_arg_type): Likewise.
18130 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
18131 (arc_preferred_simd_mode): Likewise.
18132 (arc_secondary_reload): Likewise.
18133 (get_arc_condition_code): Likewise.
18134 (arc_print_operand): Likewise.
18135 (arc_legitimate_constant_p): Likewise.
18136 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18137 * config/arc/arc.md (casesi_load): Likewise.
18138 (casesi_compact_jump): Likewise.
18139 * config/arc/predicates.md (proper_comparison_operator): Likewise.
18140 (cc_use_register): Likewise.
18141 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18142 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
18143 (arm_init_iwmmxt_builtins): Likewise.
18144 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
18145 (neon_expand_vector_init): Likewise.
18146 (arm_attr_length_move_neon): Likewise.
18147 (maybe_get_arm_condition_code): Likewise.
18148 (arm_emit_vector_const): Likewise.
18149 (arm_preferred_simd_mode): Likewise.
18150 (arm_output_iwmmxt_tinsr): Likewise.
18151 (thumb1_output_casesi): Likewise.
18152 (thumb2_output_casesi): Likewise.
18153 (arm_emit_load_exclusive): Likewise.
18154 (arm_emit_store_exclusive): Likewise.
18155 (arm_expand_compare_and_swap): Likewise.
18156 (arm_evpc_neon_vuzp): Likewise.
18157 (arm_evpc_neon_vzip): Likewise.
18158 (arm_evpc_neon_vrev): Likewise.
18159 (arm_evpc_neon_vtrn): Likewise.
18160 (arm_evpc_neon_vext): Likewise.
18161 (arm_validize_comparison): Likewise.
18162 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
18163 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
18164 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
18165 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
18166 (c6x_preferred_simd_mode): Likewise.
18167 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
18168 (epiphany_rtx_costs): Likewise.
18169 * config/epiphany/predicates.md (proper_comparison_operator):
18171 * config/frv/frv.c (condexec_memory_operand): Likewise.
18172 (frv_emit_move): Likewise.
18173 (output_move_single): Likewise.
18174 (output_condmove_single): Likewise.
18175 (frv_hard_regno_mode_ok): Likewise.
18176 (frv_matching_accg_mode): Likewise.
18177 * config/h8300/h8300.c (split_adds_subs): Likewise.
18178 (h8300_rtx_costs): Likewise.
18179 (h8300_print_operand): Likewise.
18180 (compute_mov_length): Likewise.
18181 (output_logical_op): Likewise.
18182 (compute_logical_op_length): Likewise.
18183 (compute_logical_op_cc): Likewise.
18184 (h8300_shift_needs_scratch_p): Likewise.
18185 (output_a_shift): Likewise.
18186 (compute_a_shift_length): Likewise.
18187 (compute_a_shift_cc): Likewise.
18188 (expand_a_rotate): Likewise.
18189 (output_a_rotate): Likewise.
18190 * config/i386/i386.c (classify_argument): Likewise.
18191 (function_arg_advance_32): Likewise.
18192 (function_arg_32): Likewise.
18193 (function_arg_64): Likewise.
18194 (function_value_64): Likewise.
18195 (ix86_gimplify_va_arg): Likewise.
18196 (ix86_legitimate_constant_p): Likewise.
18197 (put_condition_code): Likewise.
18198 (split_double_mode): Likewise.
18199 (ix86_avx256_split_vector_move_misalign): Likewise.
18200 (ix86_expand_vector_logical_operator): Likewise.
18201 (ix86_split_idivmod): Likewise.
18202 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
18203 (ix86_build_const_vector): Likewise.
18204 (ix86_build_signbit_mask): Likewise.
18205 (ix86_match_ccmode): Likewise.
18206 (ix86_cc_modes_compatible): Likewise.
18207 (ix86_expand_branch): Likewise.
18208 (ix86_expand_sse_cmp): Likewise.
18209 (ix86_expand_sse_movcc): Likewise.
18210 (ix86_expand_int_sse_cmp): Likewise.
18211 (ix86_expand_vec_perm_vpermi2): Likewise.
18212 (ix86_expand_vec_perm): Likewise.
18213 (ix86_expand_sse_unpack): Likewise.
18214 (ix86_expand_int_addcc): Likewise.
18215 (ix86_split_to_parts): Likewise.
18216 (ix86_vectorize_builtin_gather): Likewise.
18217 (ix86_vectorize_builtin_scatter): Likewise.
18218 (avx_vpermilp_parallel): Likewise.
18219 (inline_memory_move_cost): Likewise.
18220 (ix86_tieable_integer_mode_p): Likewise.
18221 (x86_maybe_negate_const_int): Likewise.
18222 (ix86_expand_vector_init_duplicate): Likewise.
18223 (ix86_expand_vector_init_one_nonzero): Likewise.
18224 (ix86_expand_vector_init_one_var): Likewise.
18225 (ix86_expand_vector_init_concat): Likewise.
18226 (ix86_expand_vector_init_interleave): Likewise.
18227 (ix86_expand_vector_init_general): Likewise.
18228 (ix86_expand_vector_set): Likewise.
18229 (ix86_expand_vector_extract): Likewise.
18230 (emit_reduc_half): Likewise.
18231 (ix86_emit_i387_round): Likewise.
18232 (ix86_mangle_type): Likewise.
18233 (ix86_expand_round_sse4): Likewise.
18234 (expand_vec_perm_blend): Likewise.
18235 (canonicalize_vector_int_perm): Likewise.
18236 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
18237 (expand_vec_perm_1): Likewise.
18238 (expand_vec_perm_interleave3): Likewise.
18239 (expand_vec_perm_even_odd_pack): Likewise.
18240 (expand_vec_perm_even_odd_1): Likewise.
18241 (expand_vec_perm_broadcast_1): Likewise.
18242 (ix86_vectorize_vec_perm_const_ok): Likewise.
18243 (ix86_expand_vecop_qihi): Likewise.
18244 (ix86_expand_mul_widen_hilo): Likewise.
18245 (ix86_expand_sse2_abs): Likewise.
18246 (ix86_expand_pextr): Likewise.
18247 (ix86_expand_pinsr): Likewise.
18248 (ix86_preferred_simd_mode): Likewise.
18249 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
18250 * config/i386/sse.md (*andnot<mode>3): Likewise.
18251 (<mask_codefor><code><mode>3<mask_name>): Likewise.
18252 (*<code><mode>3): Likewise.
18253 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
18254 (ia64_expand_atomic_op): Likewise.
18255 (ia64_arg_type): Likewise.
18256 (ia64_mode_to_int): Likewise.
18257 (ia64_scalar_mode_supported_p): Likewise.
18258 (ia64_vector_mode_supported_p): Likewise.
18259 (expand_vec_perm_broadcast): Likewise.
18260 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
18261 (iq2000_function_arg_advance): Likewise.
18262 (iq2000_function_arg): Likewise.
18263 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
18264 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
18265 (m68k_libcall_value): Likewise.
18266 (m68k_function_value): Likewise.
18267 (sched_attr_op_type): Likewise.
18268 * config/mcore/mcore.c (mcore_output_move): Likewise.
18269 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
18271 (microblaze_function_arg): Likewise.
18272 * config/mips/mips.c (mips16_build_call_stub): Likewise.
18273 (mips_print_operand): Likewise.
18274 (mips_mode_ok_for_mov_fmt_p): Likewise.
18275 (mips_vector_mode_supported_p): Likewise.
18276 (mips_preferred_simd_mode): Likewise.
18277 (mips_expand_vpc_loongson_even_odd): Likewise.
18278 (mips_expand_vec_unpack): Likewise.
18279 (mips_expand_vi_broadcast): Likewise.
18280 (mips_expand_vector_init): Likewise.
18281 (mips_expand_vec_reduc): Likewise.
18282 (mips_expand_msa_cmp): Likewise.
18283 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
18284 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
18285 (cc_flags_for_mode): Likewise.
18286 * config/msp430/msp430.c (msp430_print_operand): Likewise.
18287 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
18288 (nds32_output_casesi_pc_relative): Likewise.
18289 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18290 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
18291 (nvptx_gen_unpack): Likewise.
18292 (nvptx_gen_pack): Likewise.
18293 (nvptx_gen_shuffle): Likewise.
18294 (nvptx_gen_wcast): Likewise.
18295 (nvptx_preferred_simd_mode): Likewise.
18296 * config/pa/pa.c (pa_secondary_reload): Likewise.
18297 * config/pa/predicates.md (base14_operand): Likewise.
18298 * config/powerpcspe/powerpcspe-c.c
18299 (altivec_resolve_overloaded_builtin): Likewise.
18300 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
18302 (rs6000_preferred_simd_mode): Likewise.
18303 (output_vec_const_move): Likewise.
18304 (rs6000_expand_vector_extract): Likewise.
18305 (rs6000_split_vec_extract_var): Likewise.
18306 (reg_offset_addressing_ok_p): Likewise.
18307 (rs6000_legitimate_offset_address_p): Likewise.
18308 (rs6000_legitimize_address): Likewise.
18309 (rs6000_emit_set_const): Likewise.
18310 (rs6000_const_vec): Likewise.
18311 (rs6000_emit_move): Likewise.
18312 (spe_build_register_parallel): Likewise.
18313 (rs6000_darwin64_record_arg_recurse): Likewise.
18314 (swap_selector_for_mode): Likewise.
18315 (spe_init_builtins): Likewise.
18316 (paired_init_builtins): Likewise.
18317 (altivec_init_builtins): Likewise.
18318 (do_load_for_compare): Likewise.
18319 (rs6000_generate_compare): Likewise.
18320 (rs6000_expand_float128_convert): Likewise.
18321 (emit_load_locked): Likewise.
18322 (emit_store_conditional): Likewise.
18323 (rs6000_output_function_epilogue): Likewise.
18324 (rs6000_handle_altivec_attribute): Likewise.
18325 (rs6000_function_value): Likewise.
18326 (emit_fusion_gpr_load): Likewise.
18327 (emit_fusion_p9_load): Likewise.
18328 (emit_fusion_p9_store): Likewise.
18329 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
18330 (fusion_gpr_mem_load): Likewise.
18331 (fusion_addis_mem_combo_load): Likewise.
18332 (fusion_addis_mem_combo_store): Likewise.
18333 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
18334 (fusion_gpr_mem_load): Likewise.
18335 (fusion_addis_mem_combo_load): Likewise.
18336 (fusion_addis_mem_combo_store): Likewise.
18337 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18339 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
18340 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
18341 (rs6000_preferred_simd_mode): Likewise.
18342 (output_vec_const_move): Likewise.
18343 (rs6000_expand_vector_extract): Likewise.
18344 (rs6000_split_vec_extract_var): Likewise.
18345 (reg_offset_addressing_ok_p): Likewise.
18346 (rs6000_legitimate_offset_address_p): Likewise.
18347 (rs6000_legitimize_address): Likewise.
18348 (rs6000_emit_set_const): Likewise.
18349 (rs6000_const_vec): Likewise.
18350 (rs6000_emit_move): Likewise.
18351 (rs6000_darwin64_record_arg_recurse): Likewise.
18352 (swap_selector_for_mode): Likewise.
18353 (paired_init_builtins): Likewise.
18354 (altivec_init_builtins): Likewise.
18355 (rs6000_expand_float128_convert): Likewise.
18356 (emit_load_locked): Likewise.
18357 (emit_store_conditional): Likewise.
18358 (rs6000_output_function_epilogue): Likewise.
18359 (rs6000_handle_altivec_attribute): Likewise.
18360 (rs6000_function_value): Likewise.
18361 (emit_fusion_gpr_load): Likewise.
18362 (emit_fusion_p9_load): Likewise.
18363 (emit_fusion_p9_store): Likewise.
18364 * config/rx/rx.c (rx_gen_move_template): Likewise.
18365 (flags_from_mode): Likewise.
18366 * config/s390/predicates.md (s390_alc_comparison): Likewise.
18367 (s390_slb_comparison): Likewise.
18368 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
18369 (s390_vector_mode_supported_p): Likewise.
18370 (s390_cc_modes_compatible): Likewise.
18371 (s390_match_ccmode_set): Likewise.
18372 (s390_canonicalize_comparison): Likewise.
18373 (s390_emit_compare_and_swap): Likewise.
18374 (s390_branch_condition_mask): Likewise.
18375 (s390_rtx_costs): Likewise.
18376 (s390_secondary_reload): Likewise.
18377 (__SECONDARY_RELOAD_CASE): Likewise.
18378 (s390_expand_cs): Likewise.
18379 (s390_preferred_simd_mode): Likewise.
18380 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
18381 * config/sh/sh.c (sh_print_operand): Likewise.
18382 (dump_table): Likewise.
18383 (sh_secondary_reload): Likewise.
18384 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18385 * config/sh/sh.md (casesi_worker_1): Likewise.
18386 (casesi_worker_2): Likewise.
18387 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
18388 (fcc_comparison_operator): Likewise.
18389 * config/sparc/sparc.c (sparc_expand_move): Likewise.
18390 (emit_soft_tfmode_cvt): Likewise.
18391 (sparc_preferred_simd_mode): Likewise.
18392 (output_cbranch): Likewise.
18393 (sparc_print_operand): Likewise.
18394 (sparc_expand_vec_perm_bmask): Likewise.
18395 (vector_init_bshuffle): Likewise.
18396 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
18397 (spu_vector_mode_supported_p): Likewise.
18398 (spu_expand_insv): Likewise.
18399 (spu_emit_branch_or_set): Likewise.
18400 (spu_handle_vector_attribute): Likewise.
18401 (spu_builtin_splats): Likewise.
18402 (spu_builtin_extract): Likewise.
18403 (spu_builtin_promote): Likewise.
18404 (spu_expand_sign_extend): Likewise.
18405 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
18406 (tilegx_simd_int): Likewise.
18407 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
18408 (tilepro_simd_int): Likewise.
18409 * config/v850/v850.c (const_double_split): Likewise.
18410 (v850_print_operand): Likewise.
18411 (ep_memory_offset): Likewise.
18412 * config/vax/vax.c (vax_rtx_costs): Likewise.
18413 (vax_output_int_move): Likewise.
18414 (vax_output_int_add): Likewise.
18415 (vax_output_int_subtract): Likewise.
18416 * config/visium/predicates.md (visium_branch_operator): Likewise.
18417 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
18418 (visium_print_operand_address): Likewise.
18419 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18420 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
18421 (xtensa_expand_conditional_branch): Likewise.
18422 (xtensa_copy_incoming_a7): Likewise.
18423 (xtensa_output_literal): Likewise.
18424 * dfp.c (decimal_real_maxval): Likewise.
18425 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
18427 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18428 Alan Hayward <alan.hayward@arm.com>
18429 David Sherwood <david.sherwood@arm.com>
18431 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
18432 (mode_nunits_inline): Likewise.
18433 (mode_inner_inline): Likewise.
18434 (mode_unit_size_inline): Likewise.
18435 (mode_unit_precision_inline): Likewise.
18436 (emit_insn_modes_h): Likewise. Also emit a #define of the
18438 (emit_mode_wider): Add an E_ prefix to mode names.
18439 (emit_mode_complex): Likewise.
18440 (emit_mode_inner): Likewise.
18441 (emit_mode_adjustments): Likewise.
18442 (emit_mode_int_n): Likewise.
18443 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
18444 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
18445 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
18446 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
18447 (CRC32_BUILTIN, ENTRY): Likewise.
18448 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
18449 (aarch64_pop_regs): Likewise.
18450 (aarch64_process_components): Likewise.
18451 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
18452 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
18453 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
18454 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
18455 * config/arm/arm.c (arm_init_libfuncs): Likewise.
18456 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
18458 * config/i386/i386-builtin.def (pcmpestr): Likewise.
18459 (pcmpistr): Likewise.
18460 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
18461 * config/mmix/mmix.c (mmix_output_condition): Likewise.
18462 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
18464 * config/rl78/rl78.c (mduc_regs): Likewise.
18465 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
18466 (htm_expand_builtin): Likewise.
18467 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
18468 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
18469 * config/xtensa/xtensa.c (print_operand): Likewise.
18470 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
18471 (NUM_MODE_VECTOR_INT): Likewise.
18472 * genoutput.c (null_operand): Likewise.
18473 (output_operand_data): Likewise.
18474 * genrecog.c (print_parameter_value): Likewise.
18475 * lra.c (debug_operand_data): Likewise.
18477 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18479 * dwarf2out.c (scompare_loc_descriptor_wide)
18480 (scompare_loc_descriptor_narrow): New functions, split out from...
18481 (scompare_loc_descriptor): ...here.
18482 * expmed.c (emit_store_flag_int): New function, split out from...
18483 (emit_store_flag): ...here.
18485 2017-08-30 Richard Biener <rguenther@suse.de>
18487 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
18488 (dwarf2out_early_finish): Move setting of AT_pubnames from
18489 early debug output to early finish.
18491 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
18493 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
18494 and -mdata-region to the assembler.
18496 2017-08-30 Richard Biener <rguenther@suse.de>
18498 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
18500 (gen_subprogram_die): Add DW_AT_object_pointer only early.
18501 (dwarf2out_early_global_decl): Only generate a DIE for the
18502 abstract origin if it doesn't already exist or is a declaration DIE.
18503 (resolve_addr): Do not add the linkage name twice when
18504 generating a stub DIE for the DW_TAG_GNU_call_site target.
18506 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18508 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18509 Use machine_mode rather than int for arg1_mode.
18511 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
18514 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
18515 that the second argument of the built-in functions to unpack
18516 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
18517 switch statement instead a lot of if statements.
18518 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
18519 Allow 64-bit values to be in Altivec registers as well as
18520 traditional floating point registers.
18521 (pack<mode>, FMOVE128_VSX iterator): Likewise.
18523 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
18525 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
18526 MAX_REGS_PER_ADDRESS == 1.
18528 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
18530 * config/i386/i386.opt (flag_fentry): Do not init to -1.
18531 * config/i386/i386.c (ix86_option_override_internal): Simplify
18532 setting of opts->x_flag_entry.
18534 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18535 Jakub Jelinek <jakub@redhat.com>
18536 Richard Biener <rguenther@suse.de>
18538 PR tree-optimization/81503
18539 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
18540 folded constant fits in the target type; reorder tests for clarity.
18542 2017-08-29 Martin Liska <mliska@suse.cz>
18544 * passes.def: Include pass_lower_switch.
18545 * stmt.c (dump_case_nodes): Remove and move to
18546 tree-switch-conversion.
18547 (case_values_threshold): Likewise.
18548 (expand_switch_as_decision_tree_p): Likewise.
18549 (emit_case_decision_tree): Likewise.
18550 (expand_case): Likewise.
18551 (balance_case_nodes): Likewise.
18552 (node_has_low_bound): Likewise.
18553 (node_has_high_bound): Likewise.
18554 (node_is_bounded): Likewise.
18555 (emit_case_nodes): Likewise.
18556 (struct simple_case_node): New struct.
18557 (add_case_node): Remove.
18558 (emit_case_dispatch_table): Use vector instead of case_list.
18559 (reset_out_edges_aux): Remove.
18560 (compute_cases_per_edge): Likewise.
18561 (expand_case): Build list of simple_case_node.
18562 (expand_sjlj_dispatch_table): Use it.
18563 * tree-switch-conversion.c (struct case_node): Moved from
18564 stmt.c and adjusted.
18565 (emit_case_nodes): Likewise.
18566 (node_has_low_bound): Likewise.
18567 (node_has_high_bound): Likewise.
18568 (node_is_bounded): Likewise.
18569 (case_values_threshold): Likewise.
18570 (reset_out_edges_aux): Likewise.
18571 (compute_cases_per_edge): Likewise.
18572 (add_case_node): Likewise.
18573 (dump_case_nodes): Likewise.
18574 (balance_case_nodes): Likewise.
18575 (expand_switch_as_decision_tree_p): Likewise.
18576 (emit_jump): Likewise.
18577 (emit_case_decision_tree): Likewise.
18578 (try_switch_expansion): Likewise.
18579 (do_jump_if_equal): Likewise.
18580 (emit_cmp_and_jump_insns): Likewise.
18581 (fix_phi_operands_for_edge): New function.
18582 (record_phi_operand_mapping): Likewise.
18583 (class pass_lower_switch): New pass.
18584 (pass_lower_switch::execute): New function.
18585 (make_pass_lower_switch): Likewise.
18586 (conditional_probability):
18587 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
18588 * tree-pass.h: Add make_pass_lower_switch.
18590 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
18593 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
18596 2017-08-29 Richard Biener <rguenther@suse.de>
18598 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
18599 we do not add a DW_AT_inline attribute twice.
18600 (gen_subprogram_die): Remove code setting DW_AT_inline on
18601 DECL_ABSTRACT_P nodes.
18603 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
18605 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
18606 calls to internal functions.
18607 (gimplify_modify_expr): Likewise.
18608 * tree-call-cdce.c (use_internal_fn): Likewise.
18609 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
18610 (convert_to_divmod): Set the nothrow flag.
18611 * tree-if-conv.c (predicate_mem_writes): Likewise.
18612 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
18613 (vectorizable_call): Likewise.
18614 (vectorizable_store): Likewise.
18615 (vectorizable_load): Likewise.
18616 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
18617 (vect_recog_mask_conversion_pattern): Likewise.
18619 2017-08-29 Martin Liska <mliska@suse.cz>
18622 * gcc.c (driver_handle_option): Add new argument.
18623 * opts-common.c (handle_option): Pass
18624 target_option_override_hook.
18625 * opts-global.c (lang_handle_option): Add new option.
18626 (set_default_handlers): Add new argument.
18627 (decode_options): Likewise.
18628 * opts.c (target_handle_option): Likewise.
18629 (common_handle_option): Call target_option_override_hook.
18630 * opts.h (struct cl_option_handler_func): Add hook for
18631 target option override.
18632 (struct cl_option_handlers): Likewise.
18633 (set_default_handlers): Add new argument.
18634 (decode_options): Likewise.
18635 (common_handle_option): Likewise.
18636 (target_handle_option): Likewise.
18637 * toplev.c (toplev::main): Pass targetm.target_option.override
18640 2017-08-29 Richard Biener <rguenther@suse.de>
18641 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
18643 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
18644 life to the active subtree.
18646 2017-08-28 Jeff Law <law@redhat.com>
18648 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
18649 derive_equivalences.
18650 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
18651 Code moved into....
18652 (edge_info::derive_equivalences): New private member function
18654 * tree-ssa-dom.c (class edge_info): Changed from a struct
18655 to a class. Add ctor/dtor, methods and data members.
18656 (edge_info::edge_info): Renamed from allocate_edge_info.
18657 Initialize additional members.
18658 (edge_info::~edge_info): New.
18659 (free_dom_edge_info): Delete the edge info.
18660 (record_edge_info): Use new class & associated member functions.
18661 Tighten forms for testing for edge equivalences.
18662 (record_temporary_equivalences): Iterate over the simple
18663 equivalences rather than assuming there's only one per edge.
18664 (cprop_into_successor_phis): Iterate over the simple
18665 equivalences rather than assuming there's only one per edge.
18666 (optimize_stmt): Use operand_equal_p rather than pointer
18667 equality for mini-DSE code.
18669 2017-08-28 Nathan Sidwell <nathan@acm.org>
18671 * gcc.c (execute): Fold SIGPIPE handling into switch
18672 statement. Adjust internal error message.
18674 2017-08-28 Richard Biener <rguenther@suse.de>
18677 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
18678 Do nothing for removed DIEs.
18680 2017-08-28 Richard Biener <rguenther@suse.de>
18682 PR tree-optimization/81977
18683 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
18686 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
18689 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
18690 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
18691 using targetm.gen_mem_thread_fence.
18693 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
18696 * config/i386/i386.md (*<btsc><mode>): Change operand 2
18697 predicate to register_operand. Reorder operands.
18698 (*btr<mode>): Ditto.
18699 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
18700 (*btr<mode>_mask): Ditto.
18702 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
18704 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
18705 * config/rs6000/xmmintrin.h: New file.
18706 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
18708 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18711 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
18712 parameter and_insn and return it.
18713 (recombine_lvx_pattern): Insert a copy to ensure availability of
18714 the base register of the copied masking operation at the point of
18715 the instruction replacement.
18716 (recombine_stvx_pattern): Likewise.
18718 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
18720 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
18721 undocumented switches.
18722 (-mpower9-dform-vector): Likewise.
18723 (-mpower9-dform): Likewise.
18724 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
18725 comments to delete references to -mpower9-dform* switches.
18726 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
18727 Delete reference to -mpower9-dform* switches, test for
18728 -mpower9-vector instead.
18729 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
18730 (OTHER_P9_VECTOR_MASKS): Likewise.
18731 (POWERPC_MASKS): Likewise.
18732 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
18733 tests against -mpower9-dform* to -mpower9-vector. Delete code
18734 that checked for -mpower9-dform* consistancy with other options.
18735 Add test for -mpower9-misc to enable other power9 switches.
18736 (rs6000_init_hard_regno_mode_ok): Likewise.
18737 (rs6000_option_override_internal): Likewise.
18738 (rs6000_emit_prologue): Likewise.
18739 (rs6000_emit_epilogue): Likewise.
18740 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
18741 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
18742 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
18744 (emit_fusion_p9_store): Likewise.
18745 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
18746 resetting these macros if the assembler does not support ISA 3.0
18748 (TARGET_P9_DFORM_VECTOR): Likewise.
18749 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
18750 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
18752 2017-08-25 Alan Modra <amodra@gmail.com>
18754 PR rtl-optimization/81747
18755 * cse.c (cse_extended_basic_block): Don't attempt to record
18756 equivalences for degenerate conditional branches that branch
18757 to their fall-through.
18759 2017-08-24 Martin Sebor <msebor@redhat.com>
18761 PR middle-end/81908
18762 * gimple-fold.c (size_must_be_zero_p): New function.
18763 (gimple_fold_builtin_memory_op): Call it.
18765 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
18767 * config/rs6000/mm_malloc.h: New file.
18769 2017-08-24 Bin Cheng <bin.cheng@arm.com>
18771 PR tree-optimization/81913
18772 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
18773 analysis when either IVs in condition can wrap.
18775 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
18777 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
18778 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
18780 2017-08-24 Richard Biener <rguenther@suse.de>
18783 * targhooks.c (default_target_can_inline_p): Properly
18784 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
18785 is present and always compare.
18786 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
18787 infer -mfpmath=sse from TARGET_SSE_P.
18788 (ix86_can_inline_p): Properly use target_option_default_node when
18789 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
18791 2017-08-24 Richard Biener <rguenther@suse.de>
18794 * dwarf2out.c (output_die): Handle flag_generate_offload like
18796 (output_comp_unit): Likewise.
18797 (gen_array_type_die): Likewise.
18798 (dwarf2out_early_finish): Likewise.
18799 (note_variable_value_in_expr): Likewise.
18800 (dwarf2out_finish): Likewise. Adjust assert.
18801 * cgraphunit.c (symbol_table::compile): Move setting of
18802 flag_generate_offload earlier ...
18803 (symbol_table::finalize_compilation_unit): ... here, before
18804 early debug finalization.
18806 2017-08-24 Richard Biener <rguenther@suse.de>
18808 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
18809 and ipa-fnsummary.h.
18810 (ix86_can_inline_p): When ix86_fpmath flags do not match
18811 check whether the callee uses FP math at all.
18813 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
18815 PR middle-end/81931
18816 * tree-ssanames.c (get_nonzero_bits): Use element_precision
18817 instead of TYPE_PRECISION.
18819 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
18820 Alan Hayward <alan.hayward@arm.com>
18821 David Sherwood <david.sherwood@arm.com>
18823 * combine.c (make_extraction): Use subreg_offset_from_lsb.
18825 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
18827 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
18829 (ix86_frame::stack_realign_allocate): New field.
18830 (struct machine_frame_state): Modify comments.
18831 (machine_frame_state::sp_realigned_fp_end): New field.
18832 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
18833 layout calculation.
18834 (sp_valid_at): Add assertion to assure no attempt to access invalid
18835 offset of a realigned stack.
18836 (fp_valid_at): Likewise.
18837 (choose_baseaddr): Modify comments.
18838 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
18839 ix86_expand_prologue.
18840 (ix86_expand_prologue): Modify stack realignment and allocation.
18841 (ix86_expand_epilogue): Modify comments.
18842 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
18843 avx2_runtime, avx512f, and avx512f_runtime.
18845 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
18847 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
18848 (mstackrealign): Do not init to -1.
18849 * config/i386/i386.c (ix86_option_override_internal):
18850 Check opts_set, not opts when setting default value of
18851 opts->x_ix86_force_align_arg_pointer.
18853 2017-08-23 Richard Biener <rguenther@suse.de>
18855 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
18856 lang_hooks.decl_printable_name.
18857 * print-rtl-function.c (print_rtx_function): Likewise.
18858 * tree-pretty-print.c (dump_function_header): Likewise.
18860 2017-08-23 Richard Biener <rguenther@suse.de>
18863 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
18864 -g0 at compile-time.
18866 2017-08-23 Tamar Christina <tamar.christina@arm.com>
18868 PR middle-end/19706
18869 * doc/sourcebuild.texi (Other hardware attributes):
18872 2017-08-23 Tamar Christina <tamar.christina@arm.com>
18874 PR middle-end/19706
18875 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
18876 Fix single-use check.
18878 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
18880 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
18882 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
18884 * doc/install.texi: Modify to add more details on running selected
18887 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
18889 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
18890 is combined with -mabi=ms.
18891 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
18894 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18896 PR tree-optimization/81488
18897 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
18898 and cached_basis fields.
18899 (MAX_SPREAD): New constant.
18900 (alloc_cand_and_find_basis): Initialize new fields.
18901 (clear_visited): New function.
18902 (create_phi_basis_1): Rename from create_phi_basis, set visited
18903 and cached_basis fields.
18904 (create_phi_basis): New wrapper function.
18905 (phi_add_costs_1): Rename from phi_add_costs, add spread
18906 parameter, set visited field, short-circuit when limits reached.
18907 (phi_add_costs): New wrapper function.
18908 (record_phi_increments_1): Rename from record_phi_increments, set
18910 (record_phi_increments): New wrapper function.
18911 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
18912 (phi_incr_cost): New wrapper function.
18913 (all_phi_incrs_profitable_1): Rename from
18914 all_phi_incrs_profitable, set visited field.
18915 (all_phi_incrs_profitable): New wrapper function.
18917 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
18918 Alan Hayward <alan.hayward@arm.com>
18919 David Sherwood <david.sherwood@arm.com>
18921 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
18922 that takes the outer and inner modes.
18923 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
18924 comparison as the canonical test for a paradoxical subreg.
18925 * combine.c (simplify_set): Use paradoxical_subreg_p.
18926 (make_extraction): Likewise.
18927 (force_to_mode): Likewise.
18928 (rtx_equal_for_field_assignment_p): Likewise.
18929 (gen_lowpart_for_combine): Likewise.
18930 (simplify_comparison): Likewise.
18931 * cse.c (equiv_constant): Likewise.
18932 * expmed.c (store_bit_field_1): Likewise.
18933 * final.c (alter_subreg): Likewise.
18934 * fwprop.c (propagate_rtx): Likewise.
18935 (forward_propagate_subreg): Likewise.
18936 * ira-conflicts.c (ira_build_conflicts): Likewise.
18937 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
18938 * lra-constraints.c (curr_insn_transform): Likewise.
18939 (split_reg): Likewise.
18940 * lra-eliminations.c (move_plus_up): Likewise.
18941 (lra_eliminate_regs_1): Likewise.
18942 * recog.c (general_operand): Likewise.
18943 * ree.c (combine_reaching_defs): Likewise.
18944 * reload.c (push_reload): Likewise.
18945 (find_reloads): Likewise.
18946 * reload1.c (elimination_effects): Likewise.
18947 (compute_reload_subreg_offset): Likewise.
18948 (choose_reload_regs): Likewise.
18949 * rtlanal.c (subreg_lsb_1): Likewise.
18950 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18951 (simplify_subreg): Likewise.
18952 * var-tracking.c (track_loc_p): Likewise.
18953 * emit-rtl.c (byte_lowpart_offset): Likewise.
18954 (paradoxical_subreg_p): Delete out-of-line definition.
18956 2017-08-22 Jeff Law <law@redhat.com>
18958 PR tree-optimization/81741
18959 PR tree-optimization/71947
18960 * tree-ssa-dom.c: Include tree-inline.h.
18961 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
18962 equivalences if one is more expensive to compute than the other.
18963 * tree-ssa-scopedtables.h (class const_or_copies): Make
18964 record_const_or_copy_raw method private.
18965 (class avail_exprs_stack): New method simplify_binary_operation.
18966 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
18967 avail_exprs_stack::simplify_binary_operation as needed.
18968 (avail_exprs_stack::simplify_binary_operation): New function.
18970 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
18972 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
18973 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
18974 (DOT_SYMBOLS): Likewise.
18975 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
18976 (RELOCATABLE_NEEDS_FIXUP): Likewise.
18977 (RS6000_ABI_NAME): Likewise.
18978 (TARGET_CMODEL): Likewise.
18979 (TOC_SECTION_ASM_OP): Likewise.
18980 (SET_CMODEL): New macro.
18981 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
18983 2017-08-22 Richard Biener <rguenther@suse.de>
18985 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
18986 to allow for free-lang-data replacements similar to verify_type_variant.
18988 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
18989 Alan Hayward <alan.hayward@arm.com>
18990 David Sherwood <david.sherwood@arm.com>
18992 * config/aarch64/aarch64.md (casesi): Use DImode rather than
18993 VOIDmode for the LABEL_REF.
18995 2017-08-22 Richard Biener <rguenther@suse.de>
18997 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
18999 2017-08-22 Richard Biener <rguenther@suse.de>
19001 * common.opt (feliminate-dwarf2-dups): Ignore.
19002 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
19003 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
19004 same_die_p_wrap, compute_section_prefix,
19005 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
19006 (comdat_symbol_id, comdat_symbol_number): Likewise.
19007 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
19009 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
19010 (output_die): Mark unreachable path unreachable.
19011 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
19012 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
19013 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
19014 (dwarf2out_early_finish): Likewise.
19016 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
19018 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
19020 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
19023 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
19024 not VAR_P. Filter attribute warnings with OPT_Wattributes.
19025 (avr_attribute_table) <io, io_low, address>: Initialize
19026 .decl_required with true.
19028 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
19030 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
19031 undocumented debugging options.
19032 (-mvsx-scalar-double): Likewise.
19033 (-mallow-df-permute): Likewise.
19034 (-mvectorize-builtins): Likewise.
19035 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
19036 (rs6000_builtin_vectorized_function): Likewise.
19037 (rs6000_builtin_md_vectorized_function): Likewise.
19038 (rs6000_opt_vars): Likewise.
19040 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
19043 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
19044 (*btrq_imm): Rename from *btrq.
19045 (*btcq_imm): Rename from *btcq.
19046 (btsc): New code attribute.
19047 (*<btsc><mode>): New insn pattern.
19048 (*btr<mode>): Ditto.
19049 (*<btsc><mode>_mask): New insn_and_split pattern.
19050 (*btr<mode>_mask): Ditto.
19052 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19053 Alan Hayward <alan.hayward@arm.com>
19054 David Sherwood <david.sherwood@arm.com>
19056 * function.c (pad_below): Simplify padding calculation.
19058 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19059 Alan Hayward <alan.hayward@arm.com>
19060 David Sherwood <david.sherwood@arm.com>
19062 * target.def (function_prologue): Remove frame size argument.
19063 (function_epilogue): Likewise.
19064 * doc/tm.texi: Regenerate.
19065 * final.c (final_start_function): Update call to function_prologue.
19066 (final_end_function): Update call to function_epilogue.
19067 (default_function_pro_epilogue): Remove frame size argument.
19068 * output.h (default_function_pro_epilogue): Likewise.
19069 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
19070 (arm_output_function_prologue): Likewise.
19071 * config/frv/frv.c (frv_function_prologue): Likewise.
19072 (frv_function_epilogue): Likewise.
19073 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
19074 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
19075 (ia64_output_function_epilogue): Likewise.
19076 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
19077 (m32r_output_function_epilogue): Likewise.
19078 * config/microblaze/microblaze.c (microblaze_function_prologue)
19079 (microblaze_function_epilogue): Likewise.
19080 * config/mips/mips.c (mips_output_function_prologue): Likewise.
19081 (mips_output_function_epilogue): Likewise.
19082 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
19083 (mmix_target_asm_function_epilogue): Likewise.
19084 * config/msp430/msp430.c (msp430_start_function): Likewise.
19085 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
19086 (nds32_asm_function_epilogue): Likewise.
19087 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
19088 * config/pa/pa.c (pa_output_function_prologue): Likewise.
19089 (pa_output_function_epilogue): Likewise.
19090 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
19091 (rs6000_output_function_epilogue): Likewise.
19092 * config/rl78/rl78.c (rl78_start_function): Likewise.
19093 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
19094 (rs6000_output_function_epilogue): Likewise.
19095 * config/rx/rx.c (rx_output_function_prologue): Likewise.
19096 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
19097 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
19098 (sparc_asm_function_epilogue): Likewise.
19100 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19102 * tree.h (type_has_mode_precision_p): New function.
19103 * convert.c (convert_to_integer_1): Use it.
19104 * expr.c (expand_expr_real_2): Likewise.
19105 (expand_expr_real_1): Likewise.
19106 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
19107 * match.pd: Likewise.
19108 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
19109 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
19110 * tree-tailcall.c (process_assignment): Likewise.
19111 * tree-vect-loop.c (vectorizable_reduction): Likewise.
19112 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
19113 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
19114 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
19115 (vectorizable_assignment): Likewise.
19116 (vectorizable_shift): Likewise.
19117 (vectorizable_operation): Likewise.
19118 * tree-vrp.c (register_edge_assert_for_2): Likewise.
19120 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
19122 * match.pd: Add pow (C, x) simplification.
19124 2017-08-21 Richard Biener <rguenther@suse.de>
19126 PR tree-optimization/81900
19127 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
19128 for blocks with abnormal predecessors.
19129 (compute_antic): Do not set visited flag prematurely.
19131 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
19134 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
19136 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19138 * stor-layout.h (vector_type_mode): Move to...
19139 * tree.h (vector_type_mode): ...here.
19140 * stor-layout.c (vector_type_mode): Move to...
19141 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
19143 2017-08-21 Richard Biener <rguenther@suse.de>
19145 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
19146 register_external_die hooks.
19147 (debug_false_tree_charstarstar_uhwistar): Declare.
19148 (debug_nothing_tree_charstar_uhwi): Likewise.
19149 * debug.c (do_nothing_debug_hooks): Adjust.
19150 (debug_false_tree_charstarstar_uhwistar): New do nothing.
19151 (debug_nothing_tree_charstar_uhwi): Likewise.
19152 * dbxout.c (dbx_debug_hooks): Adjust.
19153 (xcoff_debug_hooks): Likewise.
19154 * sdbout.c (sdb_debug_hooks): Likewise.
19155 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
19156 * dwarf2out.c (macinfo_label_base): New global.
19157 (dwarf2out_register_external_die): New function for the
19158 register_external_die hook.
19159 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
19160 (dwarf2_debug_hooks): Use them.
19161 (dwarf2_lineno_debug_hooks): Adjust.
19162 (struct die_struct): Add with_offset flag.
19163 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
19164 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
19165 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
19166 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
19167 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
19168 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
19169 defining section names for the early LTO debug variants.
19170 (reset_indirect_string): New helper.
19171 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
19172 (print_dw_val): Add support for offsetted symbol references.
19173 (get_ultimate_context): Split out from is_cxx.
19174 (is_cxx): Use get_ultimate_context.
19175 (is_fortran): Add decl overload.
19176 (compute_comp_unit_symbol): Split out worker from
19177 compute_section_prefix.
19178 (compute_section_prefix): Call compute_comp_unit_symbol and
19179 set comdat_type_p here.
19180 (output_die): Skip DIE symbol output for the LTO added one.
19181 Handle DIE symbol references with offset.
19182 (output_comp_unit): Guard section name mangling properly.
19183 For LTO debug sections emit a symbol at the section beginning
19184 which we use to refer to its DIEs.
19185 (add_abstract_origin_attribute): For DIEs registered via
19186 dwarf2out_register_external_die directly refer to the early
19187 DIE rather than indirectly through the shadow one we created.
19188 Remove obsolete call to dwarf2out_abstract_function for
19189 non-function/block origins.
19190 (gen_array_type_die): When generating early LTO debug do
19191 not emit DW_AT_string_length.
19192 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
19193 late when in LTO. As suggested place a gcc_unreachable for
19194 the DECL_ABSTRACT_P case.
19195 (gen_subprogram_die): Avoid another specification DIE
19196 for early built declarations/definitions for the late LTO case.
19197 (gen_variable_die): Add type references for late duplicated VLA dies
19199 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
19200 we have the abstract instance already.
19201 (process_scope_var): Adjust decl DIE contexts in LTO which
19202 first puts them in limbo.
19203 (gen_decl_die): Do not generate type DIEs late apart from
19204 types for VLAs or for decls we do not yet have a DIE. Do not
19205 call dwarf2out_abstract_function late.
19206 (dwarf2out_early_global_decl): Make sure to create DIEs
19207 for abstract instances of a decl first.
19208 (dwarf2out_late_global_decl): Adjust comment.
19209 (output_macinfo_op): With multiple macro sections use
19210 macinfo_label_base to distinguish labels.
19211 (output_macinfo): Likewise. Update macinfo_label_base.
19212 Pass in the line info label.
19213 (note_variable_value_in_expr): When generating LTO resolve
19214 all variable values here by generating DIEs as needed.
19215 (init_sections_and_labels): Add early LTO debug flag parameter
19216 and generate different sections and names if set. Add generation
19217 counter for the labels so we can have multiple of them.
19218 (reset_dies): Helper to allow DIEs to be output multiple times.
19219 (dwarf2out_finish): When outputting DIEs to the fat part of an
19220 LTO object first reset DIEs.
19221 (dwarf2out_early_finish): Output early DIEs when generating LTO.
19222 (modified_type_die): Check for decl_ultimate_origin being self
19224 (gen_type_die_with_usage): Likewise.
19225 (gen_typedef_die): Allow decl_ultimate_origin being self.
19226 (set_decl_abstract_flags): Remove.
19227 (set_block_abstract_flags): Likewise.
19228 (dwarf2out_abstract_function): Treat the early generated DIEs
19229 as the abstract copy and only add DW_AT_inline and
19230 DW_AT_artificial here and call set_decl_origin_self.
19231 If the DIE has an abstract origin don't do anything.
19232 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
19233 if we have none yet (Go fails to build one, PR78628).
19234 (variably_modified_type_p): Prevent endless recursion for Ada
19235 cyclic pointer types.
19236 * lto-streamer-in.c: Include debug.h.
19237 (dref_queue): New global.
19238 (lto_read_tree_1): Stream in DIE references.
19239 (lto_input_tree): Register DIE references.
19240 (input_function): Stream DECL_DEBUG_ARGS.
19241 * lto-streamer-out.c: Include debug.h.
19242 (lto_write_tree_1): Output DIE references.
19243 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
19244 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
19245 (output_function): Stream DECL_DEBUG_ARGS.
19246 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
19247 Stream DECL_ABSTRACT_ORIGIN.
19248 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
19249 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
19250 DECL_CONTEXT for file-scope decls.
19251 * lto-streamer.h (struct dref_entry): Declare.
19252 (dref_queue): Likewise.
19253 * cfgexpand.c (pass_expand::execute): Do not call the
19254 outlining_inline_function hook here.
19255 * lto-wrapper.c (debug_obj): New global.
19256 (tool_cleanup): Unlink it if required.
19257 (debug_objcopy): New function.
19258 (run_gcc): Handle early debug sections in the IL files by
19259 extracting them to separate files, partially linkin them and
19260 feeding the result back as result to the linker.
19261 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
19262 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
19263 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
19264 sections into a separate segment.
19265 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
19267 (darwin_asm_dwarf_section): Likewise.
19268 (darwin_asm_output_dwarf_offset): Likewise.
19269 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
19271 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19272 Alan Hayward <alan.hayward@arm.com>
19273 David Sherwood <david.sherwood@arm.com>
19275 * read-md.h (md_reader::record_potential_iterator_use): Replace
19276 pointer argument with an rtx and an index.
19277 * read-rtl.c (iterator_group::apply_iterator): Likewise.
19278 (apply_mode_iterator): Likewise.
19279 (apply_code_iterator): Likewise.
19280 (apply_int_iterator): Likewise.
19281 (apply_subst_iterator): Likewise.
19282 (record_iterator_use): Likewise.
19283 (record_attribute_use): Likewise.
19284 (md_reader::record_potential_iterator_use): Likewise. Update calls
19285 to record_iterator_use and apply_iterator.
19286 (iterator_use): Replace ptr with x and index.
19287 (attribute_use): Likewise.
19288 (apply_attribute_uses): Update calls to apply_iterator.
19289 (apply_iterators): Likewise. Update initialization of iterator_use.
19290 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
19291 and record_potential_iterator_use.
19292 (rtx_reader::read_rtx_operand): Likewise.
19294 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19295 Alan Hayward <alan.hayward@arm.com>
19296 David Sherwood <david.sherwood@arm.com>
19298 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
19301 2017-08-21 Richard Biener <rguenther@suse.de>
19303 PR middle-end/81884
19304 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
19305 at struct end conservatively when comparing common bases.
19307 2017-08-21 Richard Biener <rguenther@suse.de>
19309 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
19310 (mem_ref_in_stmt): Remove.
19311 (determine_max_movement): Use ref index to get at the reference.
19312 (invariantness_dom_walker::before_dom_children): Deal with
19313 lim data already initialized.
19314 (gather_mem_refs_stmt): Initialize lim data and record ref index.
19316 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
19318 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
19319 (TARGET_ISA_ROUND): Ditto.
19320 (TARGET_ROUND): Ditto.
19321 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
19322 * config/i386/i386.md: Ditto.
19323 * config/i386/sse.md: Ditto.
19324 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
19325 with OPTION_MASK_ISA_SSE4_1.
19327 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
19330 * doc/extend.texi (x86 Built-in Functions): Correct the name of
19331 __builtin_ia32_lzcnt_u16.
19333 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
19336 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
19337 (rs6000_set_current_function): Rewrite function to use it.
19339 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
19342 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
19343 and TYPE_WARN_IF_NOT_ALIGN.
19344 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
19345 (handle_warn_if_not_align): New.
19346 (place_union_field): Call handle_warn_if_not_align.
19347 (place_field): Call handle_warn_if_not_align.
19348 Copy TYPE_WARN_IF_NOT_ALIGN.
19349 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
19350 (layout_type): Likewise.
19351 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
19353 (tree_decl_common): Add warn_if_not_align.
19354 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
19355 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
19356 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
19357 (DECL_WARN_IF_NOT_ALIGN): Likewise.
19358 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
19359 * doc/extend.texi: Document warn_if_not_aligned attribute.
19360 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
19362 2017-08-17 Martin Liska <mliska@suse.cz>
19365 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
19366 (get_data_dependence): Use it as pointer type.
19367 (distribute_loop): Likewise.
19369 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19371 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
19372 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
19373 (p8_vmrgow_<mode>_direct): New define_insn.
19374 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
19375 handle endianness for vmrgew and vmrgow permute patterns.
19377 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
19379 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
19380 * config/rs6000/rs6000-cpus.def: Remove comment.
19381 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
19382 (POWERPC_MASKS): Likewise.
19383 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
19384 use of TARGET_VSX_TIMODE.
19385 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
19386 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
19387 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
19388 (rs6000_option_override_internal): Remove dead code.
19389 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
19390 (rs6000_legitimize_reload_address): Likewise.
19391 (rs6000_legitimate_address_p): Likewise.
19392 (rs6000_opt_masks): Delete "vsx-timode".
19393 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
19394 from function comment.
19395 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
19396 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
19397 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
19399 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
19400 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
19401 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
19402 (VSX_TI): Likewise.
19404 (define_peephole2): Likewise.
19406 2017-08-17 Martin Sebor <msebor@redhat.com>
19409 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
19410 past the end of an array.
19411 (test_pp_format): Add test cases.
19413 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
19415 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
19416 missing ECF_NOTHROW flags.
19418 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
19421 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
19422 operands residing in integer registers.
19423 (*vsx_le_perm_load_<mode>): Likewise.
19424 (*vsx_le_perm_store_<mode>): Likewise.
19425 (define_peephole2): Add peepholes to optimize the above.
19427 2017-08-17 Marek Polacek <polacek@redhat.com>
19429 PR middle-end/81814
19430 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
19431 to mimic what shorten_compare did. Change the return type to bool.
19432 (fold_cond_expr_with_comparison): Update call to
19433 operand_equal_for_comparison_p.
19434 (fold_ternary_loc): Likewise.
19436 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
19438 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
19440 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
19442 2017-08-17 Richard Biener <rguenther@suse.de>
19444 * tree-ssa-structalias.c (solve_graph): When propagating
19445 to successors update the graphs succ edges and avoid duplicate work.
19447 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
19450 * config/i386/i386.c (ix86_option_override_internal): Save target
19451 specific options after ix86_stack_protector_guard_reg was changed.
19453 2017-08-17 Richard Biener <rguenther@suse.de>
19455 PR tree-optimization/81827
19456 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
19457 (new_var_info): Initialize it conservatively.
19458 (get_call_vi): Mark register vars.
19459 (new_scalar_tmp_constraint_exp): Likewise.
19460 (handle_rhs_call): Likewise.
19461 (handle_const_call): Likewise.
19462 (create_function_info_for): Likewise.
19463 (solve_constraints): Sort varinfos to separate register from
19464 non-register vars to pack points-to solution bitmaps during
19467 2017-08-17 Marek Polacek <polacek@redhat.com>
19469 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
19471 2017-08-17 Richard Biener <rguenther@suse.de>
19473 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
19474 to true when overflow is undefined and we saturated the result.
19476 2017-08-17 Alan Modra <amodra@gmail.com>
19479 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
19480 Don't use store multiple if only one reg needs saving.
19481 (interesting_frame_related_regno): New function.
19482 (rs6000_frame_related): Don't emit frame info for regs that
19484 (rs6000_emit_epilogue): Likewise.
19486 2017-08-16 Nathan Sidwell <nathan@acm.org>
19488 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
19489 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
19490 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
19491 * tree.c (free_lang_data_in_type): Use else-if chain. Always
19492 clear TYPE_LANG_1. Remove obsolete member-function stripping.
19493 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
19494 (verify_type): Adjust for TYPE_BINFO move.
19495 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
19496 process TYPE_BINFO directly.
19497 (hash_tree): Likewise.
19498 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
19500 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
19503 2017-08-16 David Malcolm <dmalcolm@redhat.com>
19505 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
19507 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
19510 * config/i386/i386.md (*anddi_1_btr): Change predicates of
19511 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
19512 Add ix86_binary_operator_ok to insn constraint.
19513 (*iordi_1_bts): Ditto.
19514 (*xordi_1_btc): Ditto.
19515 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
19516 Update corresponding peephole2 pattern.
19520 2017-08-16 Bin Cheng <bin.cheng@arm.com>
19522 PR tree-optimization/81832
19523 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
19524 copy loop header which has IFN_LOOP_DIST_ALIAS call.
19526 2017-08-16 Marek Polacek <polacek@redhat.com>
19529 * fold-const.c (fold_indirect_ref_1): Restore original behavior
19530 regarding size_zero_node.
19532 2017-08-16 Martin Liska <mliska@suse.cz>
19535 * config.gcc: Respect previously set extra_objs in case
19538 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
19540 PR tree-optimization/81835
19541 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
19542 the phi SSA_NAME. Check that the condition in a COND_EXPR does
19543 not depend on the phi.
19545 2017-08-16 Alan Modra <amodra@gmail.com>
19547 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
19550 2017-08-16 Alan Modra <amodra@gmail.com>
19552 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
19553 (save_reg_p): ..into this. Update all callers.
19554 (first_reg_to_save): Simplify.
19556 2017-08-16 Alan Modra <amodra@gmail.com>
19558 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
19561 2017-08-15 Joseph Myers <joseph@codesourcery.com>
19565 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
19566 constant count if that count is less than 32.
19568 2017-08-15 Nathan Sidwell <nathan@acm.org>
19570 * gcc.c (execute): Emit friendlier message if inferior is killed
19571 by an external cause.
19573 2017-08-15 Richard Biener <rguenther@suse.de>
19575 PR tree-optimization/81790
19576 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
19577 CONSTRUCTORs from simplifying and VN.
19579 2017-08-14 Martin Sebor <msebor@redhat.com>
19581 * builtin-attrs.def: Add comments.
19583 2017-08-14 Martin Sebor <msebor@redhat.com>
19586 * doc/extend.texi (attribute nonstring): Document new attribute.
19588 2017-08-14 Martin Sebor <msebor@redhat.com>
19591 * tree-diagnostic.c (default_tree_printer): Handle %G.
19592 * gimple-pretty-print.h (percent_G_format): Declare new function.
19593 * gimple-pretty-print.c (percent_G_format): Define.
19594 * tree-pretty-print.c (percent_K_format): Add argument.
19596 2017-08-14 Martin Sebor <msebor@redhat.com>
19598 PR translation/79998
19599 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
19600 Remove a stray space.
19602 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
19605 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
19606 (*iordi_1_bts): Ditto.
19607 (*xordi_1_btc): Ditto.
19609 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19612 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
19613 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19615 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
19616 quoted strings, and make more translator-friendly.
19617 (darwin_rs6000_override_options): Likewise.
19618 (rs6000_option_override_internal): Likewise.
19619 (rs6000_return_in_memory): Fix overlong line.
19620 (init_cmulative_args): Use quoted strings, and make more
19621 translator-friendly.
19622 (rs6000_pass_by_reference): Fix overlong line.
19623 (def_builtin): Use quoted strings.
19624 (altivec_expand_predicate_builtin): Use quoted strings, and make
19625 more translator-friendly.
19626 (htm_expand_builtin): Use quoted strings.
19627 (cpu_expand_builtin): Use quoted strings, and make more
19628 translator-friendly.
19629 (altivec_expand_builtin): Likewise.
19630 (paired_expand_predicate_builtin): Likewise.
19631 (rs6000_invalid_builtin): Likewise.
19632 (builtin_function_type): Use quoted strings.
19633 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
19634 more translator-friendly.
19635 (rs6000_trampoline_init): Likewise.
19636 (rs6000_handle_altivec_attribute): Likewise.
19637 (rs6000_inner_target_options): Use quoted strings.
19638 (rs6000_disable_incompatible_switches): Likewise.
19639 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
19640 strings, and make more translator-friendly.
19641 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
19643 2017-08-14 Bin Cheng <bin.cheng@arm.com>
19645 PR tree-optimization/81799
19646 * tree-loop-distribution.c (version_loop_by_alias_check): Force
19647 cond_expr to simple gimple operand.
19649 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
19651 PR middle-end/46932
19652 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
19654 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
19658 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
19659 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
19660 TARGET_GASISR_PROLOGUES.
19661 * config/avr/avr.c (avr_option_override): Same.
19662 (avr_pass_pre_proep::execute): Same.
19664 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
19667 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
19668 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
19670 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
19672 * config/i386/i386.md (*load_tp_<mode>): Redefine as
19673 define_insn_and_split. Split to a memory load from 0 in
19674 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
19675 using PTR mode iterator.
19676 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
19677 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
19678 (*add_tp_<mode>): Redefine as define_insn_and_split.
19679 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
19680 address space. Merge with *add_tp_x32 using PTR mode iterator.
19681 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
19682 Split to an add with a memory load from 0 in
19683 DEFAULT_TLS_SEG_REG address space.
19685 2017-08-12 Andrew Pinski <apinski@cavium.com>
19687 * config/aarch64/aarch64-option-extensions.def (rdma):
19688 Fix feature string to what Linux prints out in /proc/cpuinfo.
19690 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
19693 * dwarf2out.c (modified_type_die): For C typedef types that have
19694 an ultimate origin, process the ultimate origin instead of the
19696 (gen_typedef_die): Assert that input DECLs have no ultimate
19698 (gen_type_die_with_usage): For typedef variants that have an
19699 ultimate origin, just call gen_decl_die on the original DECL.
19700 (process_scope_var): Avoid creating DIEs for local typedefs and
19701 concrete static variables.
19703 2017-08-12 Alan Modra <amodra@gmail.com>
19707 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
19708 match gnu-user.h startfile.
19709 (ENDFILE_LINUX_SPEC): Similarly.
19711 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
19714 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
19716 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
19718 2017-08-11 Tamar Christina <tamar.christina@arm.com>
19719 * config/aarch64/aarch64.md (mov<mode>): Change.
19720 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
19721 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
19722 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
19724 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
19726 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
19727 for storage order barriers.
19729 2017-08-11 Martin Liska <mliska@suse.cz>
19732 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
19733 variables of void type.
19735 2017-08-11 Martin Liska <mliska@suse.cz>
19737 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
19738 TARGET_SUPPORTS_ALIASES.
19739 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
19740 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
19741 (optimize_weakref): Likewise.
19742 * symtab.c (symtab_node::noninterposable_alias): Likewise.
19743 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
19744 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
19746 2017-08-11 Martin Liska <mliska@suse.cz>
19749 * config/i386/i386.c (make_resolver_func): Do complete
19750 refactoring of the function.
19752 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
19755 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
19756 * config/i386/i386.c (ix86_stack_protect_guard): Use
19757 ix86_stack_protect_guard_symbol_str to generate varible declaration.
19758 * doc/invoke.texi (x86 Options): Document
19759 -mstack-protector-guard-symbol= option.
19761 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
19763 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
19764 * config/i386/i386.c (ix86_split_stack_guard): New function.
19765 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
19766 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
19767 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
19768 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
19769 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
19770 (split_stack_space_check): Call ix86_split_stack_guard.
19772 2017-08-10 Martin Sebor <msebor@redhat.com>
19774 * print-tree.c (print_node): Print location using the established
19776 Replace spaces with colons.
19777 (debug_raw, debug): Ditto.
19779 2017-08-10 Martin Sebor <msebor@redhat.com>
19782 * pretty-print.c (pp_format): Correct the handling of %s precision.
19784 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
19787 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
19789 (ix86_finalize_stack_frame_flags): This. Also clear
19790 frame_pointer_needed if -fno-omit-frame-pointer is used without
19792 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
19793 with ix86_finalize_stack_frame_flags.
19794 (ix86_expand_epilogue): Likewise.
19795 (ix86_expand_split_stack_prologue): Likewise.
19796 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
19798 2017-08-10 Martin Liska <mliska@suse.cz>
19801 * c-attribs.c (handle_target_attribute):
19802 Report warning for an empty string argument of target attribute.
19804 2017-08-09 Jakub Jelinek <jakub@redhat.com>
19807 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
19809 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
19810 or DECL_NONLOCAL labels.
19811 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
19812 or DECL_NONLOCAL labels here.
19814 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
19816 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
19817 to indicate when early gimple folding has been disabled.
19818 (rs6000_gimple_fold_builtin): Add debug content.
19819 (rs6000_invalid_builtin): Fix whitespace.
19820 (rs6000_expand_builtin): Fix whitespace.
19821 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
19823 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
19826 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
19827 SAVE_MULTIPLE if not all the registers that saves, should be saved.
19829 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
19831 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
19832 (qdf24xx): Likewise.
19833 * config/aarch64/aarch64.md: Include falkor.md.
19834 * config/aarch64/falkor.md: New.
19836 2017-08-09 Marek Polacek <polacek@redhat.com>
19839 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
19840 * diagnostic.c (emit_diagnostic): Add a comment.
19841 (emit_diagnostic_valist): New function.
19843 2017-08-09 Marek Polacek <polacek@redhat.com>
19846 * input.c (make_location): New overload.
19847 * input.h (make_location): Declare.
19849 2017-08-08 Alan Modra <amodra@gmail.com>
19850 H.J. Lu <hongjiu.lu@intel.com>
19853 * gcc.c (NO_PIE_SPEC): Delete.
19854 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
19856 (LINK_PIE_SPEC): ..to here.
19857 (LINK_COMMAND_SPEC): Support -no-pie.
19858 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
19859 chain of crtbegin*.o selection, update for PIE_SPEC changes and
19861 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
19862 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
19863 (ENDFILE_CRTEND_SPEC): Similarly.
19865 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
19868 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
19869 (mstack-protector-guard-offset=): Ditto.
19870 * config/i386/i386.c (ix86_option_override): Handle
19871 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
19873 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
19874 ix86_stack_protect_guard_offset variables.
19875 (TARGET_STACK_PROTECT_GUARD): Always define.
19876 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
19877 and -mstack-protector-guard-offset= options.
19879 2017-08-08 Bin Cheng <bin.cheng@arm.com>
19881 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
19882 boundary case for the last candidate.
19884 2017-08-08 Bin Cheng <bin.cheng@arm.com>
19886 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
19887 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
19889 2017-08-08 Tamar Christina <tamar.christina@arm.com>
19891 PR middle-end/19706
19892 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
19893 * config/aarch64/aarch64-builtins.c
19894 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
19895 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
19896 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
19898 2017-08-08 Tamar Christina <tamar.christina@arm.com>
19899 Andrew Pinski <pinskia@gmail.com>
19901 PR middle-end/19706
19902 * internal-fn.def (XORSIGN): New.
19903 * optabs.def (xorsign_optab): New.
19904 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
19905 (convert_expand_mult_copysign): New.
19906 (pass_optimize_widening_mul::execute): Call
19907 convert_expand_mult_copysign.
19909 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19911 PR tree-optimization/81354
19912 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
19913 Insert on edges rather than explicitly creating landing pads.
19914 (analyze_candidates_and_replace): Commit edge inserts.
19916 2017-08-08 Richard Biener <rguenther@suse.de>
19918 PR middle-end/81719
19919 * tree-ssa-loop-niter.c: Include tree-dfa.h.
19920 (expand_simple_operations): Also look through ADDR_EXPRs with
19921 MEM_REF bases treating them as POINTER_PLUS_EXPR.
19923 2017-08-08 Richard Biener <rguenther@suse.de>
19925 PR tree-optimization/81723
19926 * tree-vect-slp.c (struct bst_traits): New hash traits.
19927 (bst_fail): New global.
19928 (vect_build_slp_tree_2): New worker, split out from ...
19929 (vect_build_slp_tree): ... this now wrapping it with using
19930 bst_fail set to cache SLP tree build fails. Properly handle
19932 (vect_analyze_slp_instance): Allocate and free bst_fail.
19934 2017-08-08 Martin Liska <mliska@suse.cz>
19937 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
19938 LABEL_DECLs that can be from a different function.
19940 2017-08-08 Bin Cheng <bin.cheng@arm.com>
19942 PR tree-optimization/81744
19943 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
19944 loop's number of iterations.
19946 2017-08-08 Martin Liska <mliska@suse.cz>
19948 * asan.c: Include header files.
19949 * attribs.c (build_decl_attribute_variant): New function moved
19951 (build_type_attribute_qual_variant): Likewise.
19952 (cmp_attrib_identifiers): Likewise.
19953 (simple_cst_list_equal): Likewise.
19954 (omp_declare_simd_clauses_equal): Likewise.
19955 (attribute_value_equal): Likewise.
19956 (comp_type_attributes): Likewise.
19957 (build_type_attribute_variant): Likewise.
19958 (lookup_ident_attribute): Likewise.
19959 (remove_attribute): Likewise.
19960 (merge_attributes): Likewise.
19961 (merge_type_attributes): Likewise.
19962 (merge_decl_attributes): Likewise.
19963 (merge_dllimport_decl_attributes): Likewise.
19964 (handle_dll_attribute): Likewise.
19965 (attribute_list_equal): Likewise.
19966 (attribute_list_contained): Likewise.
19967 * attribs.h (lookup_attribute): New function moved from tree.[ch].
19968 (lookup_attribute_by_prefix): Likewise.
19969 * bb-reorder.c: Include header files.
19970 * builtins.c: Likewise.
19971 * calls.c: Likewise.
19972 * cfgexpand.c: Likewise.
19973 * cgraph.c: Likewise.
19974 * cgraphunit.c: Likewise.
19975 * convert.c: Likewise.
19976 * dwarf2out.c: Likewise.
19977 * final.c: Likewise.
19978 * fold-const.c: Likewise.
19979 * function.c: Likewise.
19980 * gimple-expr.c: Likewise.
19981 * gimple-fold.c: Likewise.
19982 * gimple-pretty-print.c: Likewise.
19983 * gimple.c: Likewise.
19984 * gimplify.c: Likewise.
19985 * hsa-common.c: Likewise.
19986 * hsa-gen.c: Likewise.
19987 * internal-fn.c: Likewise.
19988 * ipa-chkp.c: Likewise.
19989 * ipa-cp.c: Likewise.
19990 * ipa-devirt.c: Likewise.
19991 * ipa-fnsummary.c: Likewise.
19992 * ipa-inline.c: Likewise.
19993 * ipa-visibility.c: Likewise.
19995 * lto-cgraph.c: Likewise.
19996 * omp-expand.c: Likewise.
19997 * omp-general.c: Likewise.
19998 * omp-low.c: Likewise.
19999 * omp-offload.c: Likewise.
20000 * omp-simd-clone.c: Likewise.
20001 * opts-global.c: Likewise.
20002 * passes.c: Likewise.
20003 * predict.c: Likewise.
20004 * sancov.c: Likewise.
20005 * sanopt.c: Likewise.
20006 * symtab.c: Likewise.
20007 * toplev.c: Likewise.
20008 * trans-mem.c: Likewise.
20009 * tree-chkp.c: Likewise.
20010 * tree-eh.c: Likewise.
20011 * tree-into-ssa.c: Likewise.
20012 * tree-object-size.c: Likewise.
20013 * tree-parloops.c: Likewise.
20014 * tree-profile.c: Likewise.
20015 * tree-ssa-ccp.c: Likewise.
20016 * tree-ssa-live.c: Likewise.
20017 * tree-ssa-loop.c: Likewise.
20018 * tree-ssa-sccvn.c: Likewise.
20019 * tree-ssa-structalias.c: Likewise.
20020 * tree-ssa.c: Likewise.
20021 * tree-streamer-in.c: Likewise.
20022 * tree-vectorizer.c: Likewise.
20023 * tree-vrp.c: Likewise.
20024 * tsan.c: Likewise.
20025 * ubsan.c: Likewise.
20026 * varasm.c: Likewise.
20027 * varpool.c: Likewise.
20028 * tree.c: Remove functions moved to attribs.[ch].
20029 * tree.h: Likewise.
20030 * config/aarch64/aarch64.c: Add attrs.h header file.
20031 * config/alpha/alpha.c: Likewise.
20032 * config/arc/arc.c: Likewise.
20033 * config/arm/arm.c: Likewise.
20034 * config/avr/avr.c: Likewise.
20035 * config/bfin/bfin.c: Likewise.
20036 * config/c6x/c6x.c: Likewise.
20037 * config/cr16/cr16.c: Likewise.
20038 * config/cris/cris.c: Likewise.
20039 * config/darwin.c: Likewise.
20040 * config/epiphany/epiphany.c: Likewise.
20041 * config/fr30/fr30.c: Likewise.
20042 * config/frv/frv.c: Likewise.
20043 * config/ft32/ft32.c: Likewise.
20044 * config/h8300/h8300.c: Likewise.
20045 * config/i386/winnt.c: Likewise.
20046 * config/ia64/ia64.c: Likewise.
20047 * config/iq2000/iq2000.c: Likewise.
20048 * config/lm32/lm32.c: Likewise.
20049 * config/m32c/m32c.c: Likewise.
20050 * config/m32r/m32r.c: Likewise.
20051 * config/m68k/m68k.c: Likewise.
20052 * config/mcore/mcore.c: Likewise.
20053 * config/microblaze/microblaze.c: Likewise.
20054 * config/mips/mips.c: Likewise.
20055 * config/mmix/mmix.c: Likewise.
20056 * config/mn10300/mn10300.c: Likewise.
20057 * config/moxie/moxie.c: Likewise.
20058 * config/msp430/msp430.c: Likewise.
20059 * config/nds32/nds32-isr.c: Likewise.
20060 * config/nds32/nds32.c: Likewise.
20061 * config/nios2/nios2.c: Likewise.
20062 * config/nvptx/nvptx.c: Likewise.
20063 * config/pa/pa.c: Likewise.
20064 * config/pdp11/pdp11.c: Likewise.
20065 * config/powerpcspe/powerpcspe.c: Likewise.
20066 * config/riscv/riscv.c: Likewise.
20067 * config/rl78/rl78.c: Likewise.
20068 * config/rx/rx.c: Likewise.
20069 * config/s390/s390.c: Likewise.
20070 * config/sh/sh.c: Likewise.
20071 * config/sol2.c: Likewise.
20072 * config/sparc/sparc.c: Likewise.
20073 * config/spu/spu.c: Likewise.
20074 * config/stormy16/stormy16.c: Likewise.
20075 * config/tilegx/tilegx.c: Likewise.
20076 * config/tilepro/tilepro.c: Likewise.
20077 * config/v850/v850.c: Likewise.
20078 * config/vax/vax.c: Likewise.
20079 * config/visium/visium.c: Likewise.
20080 * config/xtensa/xtensa.c: Likewise.
20082 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
20085 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
20086 constraints since the -mupper-regs-* switches have been
20088 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
20089 into a vector from a double word element that was extracted from
20090 another vector, and eliminate extra XXPERMDI instructions.
20091 (vsx_concat_<mode>_2): Likewise.
20092 (vsx_concat_<mode>_3): Likewise.
20093 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
20094 concat to allow optimizing inserts from previous extracts.
20096 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
20098 * config/i386/i386.c (ix86_stack_protect_guard): Generate
20099 memory reference to a SSP offset in TLS address space.
20100 (ix86_print_operand) <case '@'>: Remove.
20101 (ix86_print_operand_punct_valid_p): Remove '@' code.
20102 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
20103 UNSPEC_SP_TLS_TEST.
20104 (stack_tls_protect_set_<mode>): Remove.
20105 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
20106 (stack_tls_protect_test_<mode>): Remove.
20107 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
20109 2017-08-07 Olivier Hainque <hainque@adacore.com>
20112 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
20114 2017-08-07 Douglas Rupp <rupp@adacore.com>
20116 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
20117 variable was referenced as multidir in command.
20119 2017-08-07 Jakub Jelinek <jakub@redhat.com>
20122 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
20125 2017-08-07 Martin Liska <mliska@suse.cz>
20127 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
20128 * config/rl78/rl78.c: Add include of attribs.h.
20129 * config/sh/sh.c: Likewise.
20130 * config/v850/v850.c: Likewise.
20132 2017-08-07 Tom de Vries <tom@codesourcery.com>
20134 PR middle-end/78266
20135 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
20137 2017-08-07 Martin Liska <mliska@suse.cz>
20139 * config/mips/mips.c: Include attribs.h.
20141 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
20144 * doc/invoke.texi: Document change in behvaior for -Ofast for
20147 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
20149 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
20151 (aarch64_pop_regs): Likewise.
20152 (aarch64_gen_load_pair): Likewise.
20153 (aarch64_save_callee_saves): Likewise.
20154 (aarch64_restore_callee_saves): Likewise.
20156 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
20158 * config/i386/i386.c: Revert the last change.
20160 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
20163 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
20165 (ix86_finalize_stack_frame_flags): This. Also clear
20166 frame_pointer_needed if -fno-omit-frame-pointer is used without
20168 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
20169 with ix86_finalize_stack_frame_flags.
20170 (ix86_expand_epilogue): Likewise.
20171 (ix86_expand_split_stack_prologue): Likewise.
20173 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
20176 * config/i386/i386.c (get_builtin_code_for_version): Set priority
20177 to P_AES for Westmere.
20179 2017-08-07 Jonathan Yong <10walls@gmail.com>
20181 * config/i386/mingw.opt (fset-stack-executable): Removed.
20182 * config/i386/cygming.opt (fset-stack-executable): Moved
20184 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
20186 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
20188 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
20190 2017-08-07 Marek Polacek <polacek@redhat.com>
20192 PR middle-end/81737
20193 * fold-const.c (fold_indirect_ref_1): Check type_domain.
20195 2017-08-07 Martin Liska <mliska@suse.cz>
20197 * attribs.h (canonicalize_attr_name): New function.
20198 (cmp_attribs): Move from c-format.c and adjusted.
20199 (is_attribute_p): Moved from tree.h.
20200 * tree-inline.c: Add new includes.
20201 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
20202 (private_is_attribute_p): Remove.
20203 (private_lookup_attribute): Likewise.
20204 (private_lookup_attribute_by_prefix): Simplify.
20205 (remove_attribute): Use is_attribute_p.
20206 * tree.h: Remove removed declarations.
20208 2017-08-07 Jakub Jelinek <jakub@redhat.com>
20210 PR middle-end/81698
20211 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
20212 instead of computing it in the function. Formatting fix.
20213 (expand_case): Don't rely on default_edge being the first edge,
20214 clear it if removing it, pass default_edge to
20215 emit_case_dispatch_table.
20216 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
20219 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
20221 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
20222 insn in the function, emit NOP after the insn.
20224 2017-08-06 Tom de Vries <tom@codesourcery.com>
20226 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
20229 2017-08-06 Tom de Vries <tom@codesourcery.com>
20231 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
20234 2017-08-04 Yury Gribov <tetra2005@gmail.com>
20236 PR tree-optimization/57371
20237 * match.pd: New pattern.
20239 2017-08-04 Marek Polacek <polacek@redhat.com>
20241 PR middle-end/81695
20242 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
20243 perform the computation in offset_int.
20245 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
20247 PR tree-optimization/81136
20248 * tree-vectorizer.h: Include tree-hash-traits.h.
20249 (vec_base_alignments): New typedef.
20250 (vec_info): Add a base_alignments field.
20251 (vect_record_base_alignments): Declare.
20252 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
20254 (DR_IS_CONDITIONAL_IN_STMT): New macro.
20255 (create_data_ref): Add an is_conditional_in_stmt argument.
20256 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
20257 the is_conditional_in_stmt field.
20258 (data_ref_loc): Add an is_conditional_in_stmt field.
20259 (get_references_in_stmt): Set the is_conditional_in_stmt field.
20260 (find_data_references_in_stmt): Update call to create_data_ref.
20261 (graphite_find_data_references_in_stmt): Likewise.
20262 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
20263 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
20264 (vect_record_base_alignment): New function.
20265 (vect_record_base_alignments): Likewise.
20266 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
20267 for nested statements even if we fail to compute a misalignment.
20268 Use pooled base alignments for unconditional references.
20269 (vect_find_same_alignment_drs): Compare base addresses instead
20271 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
20272 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
20274 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
20276 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
20277 Add an explicit name for the enum. Use auto_vec for slp_instances
20278 and grouped_stores.
20279 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
20281 (_bb_vec_info): Add a constructor and destructor.
20282 (vinfo_for_stmt): Return NULL for uids of -1 as well.
20283 (destroy_loop_vec_info): Delete.
20284 (vect_destroy_datarefs): Likewise.
20285 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
20286 (vec_info::vec_info): New function.
20287 (vec_info::~vec_info): Likewise.
20288 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
20289 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
20290 destroy_loop_vec_info.
20291 * tree-vect-loop.c (new_loop_vec_info): Replace with...
20292 (_loop_vec_info::_loop_vec_info): ...this.
20293 (destroy_loop_vec_info): Replace with...
20294 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
20295 the stmt_vec_infos. Leave handling of vec_info information to its
20296 destructor. Remove explicit vector releases.
20297 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
20298 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
20299 * tree-vect-slp.c (new_bb_vec_info): Replace with...
20300 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
20301 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
20302 (destroy_bb_vec_info): Replace with...
20303 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
20304 information to its destructor.
20305 (vect_slp_analyze_bb_1): Use new and delete instead of
20306 new_bb_vec_info and destroy_bb_vec_info.
20307 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
20310 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
20312 * tree-data-ref.h (subscript): Add access_fn field.
20313 (data_dependence_relation): Add could_be_independent_p.
20314 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
20315 (same_access_functions): Move to tree-data-ref.c.
20316 * tree-data-ref.c (ref_contains_union_access_p): New function.
20317 (access_fn_component_p): Likewise.
20318 (access_fn_components_comparable_p): Likewise.
20319 (dr_analyze_indices): Add a reference to access_fn_component_p.
20320 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
20322 (constant_access_functions): Likewise.
20323 (add_other_self_distances): Likewise.
20324 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
20325 (initialize_data_dependence_relation): Use XCNEW and remove
20326 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
20327 of access functions that have the same type. Allow the
20328 subsequence to end with different bases in some circumstances.
20329 Record the chosen access functions in SUB_ACCESS_FN.
20330 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
20331 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
20332 (subscript_dependence_tester_1): Likewise dra and drb.
20333 (build_classic_dist_vector): Update calls accordingly.
20334 (subscript_dependence_tester): Likewise.
20335 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
20336 DDR_COULD_BE_INDEPENDENT_P.
20337 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
20338 comp_alias_ddrs instead of may_alias_ddrs.
20339 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
20341 (vect_analyze_data_ref_dependence): Use it if
20342 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
20343 distance vectors if that fails.
20344 (dependence_distance_ge_vf): New function.
20345 (vect_prune_runtime_alias_test_list): Use it. Don't clear
20346 LOOP_VINFO_MAY_ALIAS_DDRS.
20348 2017-08-04 Richard Biener <rguenther@suse.de>
20350 PR middle-end/81705
20351 * fold-const.c (fold_binary_loc): Properly restrict
20352 minus_var0 && minus_var1 case when associating undefined overflow
20355 2017-08-04 Tom de Vries <tom@codesourcery.com>
20357 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
20359 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20361 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20362 Don't start diagnostic messages with a capital letter.
20363 * config/rs6000/rs6000.c (rs6000_option_override_internal):
20365 (rs6000_invalid_builtin): Likewise.
20366 (rs6000_trampoline_init): Likewise.
20368 2017-08-03 Jakub Jelinek <jakub@redhat.com>
20371 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
20372 after setting changeable df flags.
20374 2017-08-03 Richard Biener <rguenther@suse.de>
20376 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
20377 up if the use is in USE - X.
20379 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
20381 * toplev.c (dumpfile.h): New include.
20382 (internal_error_reentered): New static function. Use it...
20383 (internal_error_function): ...here to handle reentered internal_error.
20385 2017-08-03 Richard Biener <rguenther@suse.de>
20387 PR middle-end/81148
20388 * fold-const.c (split_tree): Add minus_var and minus_con
20389 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
20390 here but always use minus_*.
20391 (associate_trees): Assert we never associate with MINUS_EXPR
20392 and NULL first operand. Do not recurse for PLUS_EXPR operands
20393 when associating as MINUS_EXPR either.
20394 (fold_binary_loc): Track minus_var and minus_con.
20396 2017-08-03 Tom de Vries <tom@codesourcery.com>
20399 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
20400 ACCEL_COMPILER, apply finish_options on
20401 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
20403 2017-08-03 Tom de Vries <tom@codesourcery.com>
20406 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
20407 function_entry_patch_area_size > 0.
20409 2017-08-03 Jakub Jelinek <jakub@redhat.com>
20412 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
20413 instead of 10??LU, perform unit multiplication in wide_int,
20414 don't change alloc_object_size_limit if the limit is larger
20417 PR tree-optimization/81655
20418 PR tree-optimization/81588
20419 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
20420 the case when ranges[i].low and high are 1 for unsigned type with
20423 PR middle-end/81052
20424 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
20425 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
20427 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20429 * tree-vrp.h: Add include guard.
20431 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
20434 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
20435 (ud2): New insn pattern.
20436 * config/i386/i386.c (ix86_expand_epilogue):
20437 For naked functions, generate ud2 instead of trap insn.
20439 2017-08-02 Marek Polacek <polacek@redhat.com>
20442 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
20444 2017-08-02 Tom de Vries <tom@codesourcery.com>
20445 Cesar Philippidis <cesar@codesourcery.com>
20447 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
20448 Add missing edge probabilities.
20450 2017-08-02 Tamar Christina <tamar.christina@arm.com>
20452 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
20453 Correct endianness.
20455 2017-08-02 Jakub Jelinek <jakub@redhat.com>
20457 PR middle-end/79499
20458 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
20459 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
20460 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
20461 prologue_seq sequences - if any.
20463 2017-08-02 Richard Biener <rguenther@suse.de>
20465 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
20466 via vectors if supported, integer extracts via punning if supported
20467 or otherwise vector extracts.
20469 2017-08-02 Richard Biener <rguenther@suse.de>
20471 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
20473 (bitmap_insert_into_set): ... this.
20475 2017-08-02 Richard Biener <rguenther@suse.de>
20477 PR tree-optimization/81633
20479 2015-08-17 Alan Hayward <alan.hayward@arm.com>
20481 PR tree-optimization/71752
20482 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
20484 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
20486 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
20487 (machine_function::call_ms2sysv_pad_out): Remove field.
20488 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
20489 (ix86_compute_frame_layout): Likewise.
20491 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
20494 * config/i386/i386.c (ix86_set_func_type): Disallow naked
20495 attribute with interrupt attribute.
20497 2017-08-01 Andrew Pinski <apinski@cavium.com>
20499 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
20500 BIT_INSERT_EXPR's operand 1
20501 to see if the types precision matches.
20503 2017-08-01 Martin Liska <mliska@suse.cz>
20505 PR middle-end/70140
20506 * builtins.c (expand_builtin_memcpy_args): Remove.
20507 (expand_builtin_memcpy): Call newly added function
20508 expand_builtin_memory_copy_args.
20509 (expand_builtin_memcpy_with_bounds): Likewise.
20510 (expand_builtin_mempcpy): Remove last argument.
20511 (expand_builtin_mempcpy_with_bounds): Likewise.
20512 (expand_builtin_memory_copy_args): New function created from
20513 expand_builtin_mempcpy_args with small modifications.
20514 (expand_builtin_mempcpy_args): Remove.
20515 (expand_builtin_stpcpy): Remove unused argument.
20516 (expand_builtin): Likewise.
20517 (expand_builtin_with_bounds): Likewise.
20519 2017-08-01 Martin Liska <mliska@suse.cz>
20522 Make mempcpy more optimal (PR middle-end/70140).
20524 2017-08-01 Jakub Jelinek <jakub@redhat.com>
20527 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
20528 __builtin_vec_cmpne verify both arguments are compatible vectors
20529 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
20530 verify arg1_type is a pointer or array type. For __builtin_vec_st,
20531 move computation of aligned to after checking the argument types.
20535 * config/rs6000/vsx.md (vextract_fp_from_shorth,
20536 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
20539 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
20540 Doug Rupp <rupp@adacore.com>
20541 Olivier Hainque <hainque@adacore.com>
20543 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
20544 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
20546 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
20547 for TARGET_OS_CPP_BUILTIN.
20548 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
20549 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
20551 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
20552 passing required abi options to the assembler for EABI configurations.
20553 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
20554 of .text.hot and .text.unlikely sections for kernel modules when
20555 using ARM style exceptions.
20556 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
20557 options. Add EXTRA_CC1_SPEC.
20558 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
20560 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
20562 (ARM_TARGET2_DWARF_FORMAT): Define.
20563 * config/arm/t-vxworks: Adjust multilib control to removal of the
20564 Diab command line options.
20566 2017-08-01 Martin Liska <mliska@suse.cz>
20568 PR gcov-profile/81561
20569 * gcov.c (unblock): Make unblocking safe as we need to preserve
20570 index correspondence of blocks and block_lists.
20572 2017-08-01 Richard Biener <rguenther@suse.de>
20574 PR tree-optimization/81181
20575 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
20576 (compute_antic): ... end of iteration here.
20578 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
20580 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
20581 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
20582 (ftree-slp-vectorize): Likewise.
20583 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
20584 can no longer be set independent of flag_tree_loop_vectorize.
20585 * omp-general.c (emp_max_vf): Likewise.
20586 * opts.c (enable_fdo_optimizations): Remove references to
20587 flag_tree_vectorize, these are now implicit.
20588 (common_handle_option): Remove handling for OPT_ftree_vectorize,
20589 and leave it for the options machinery.
20591 2017-08-01 Martin Liska <mliska@suse.cz>
20593 PR middle-end/70140
20594 * builtins.c (expand_builtin_memcpy_args): Remove.
20595 (expand_builtin_memcpy): Call newly added function
20596 expand_builtin_memory_copy_args.
20597 (expand_builtin_memcpy_with_bounds): Likewise.
20598 (expand_builtin_mempcpy): Remove last argument.
20599 (expand_builtin_mempcpy_with_bounds): Likewise.
20600 (expand_builtin_memory_copy_args): New function created from
20601 expand_builtin_mempcpy_args with small modifications.
20602 (expand_builtin_mempcpy_args): Remove.
20603 (expand_builtin_stpcpy): Remove unused argument.
20604 (expand_builtin): Likewise.
20605 (expand_builtin_with_bounds): Likewise.
20607 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
20610 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
20611 print "ds:" only for immediates in generic address space.
20613 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
20616 * config/i386/i386.c (ix86_funciton_naked): New prototype.
20617 (ix86_function_ok_for_sibcall): Return false for naked functions.
20619 2017-08-01 Richard Biener <rguenther@suse.de>
20621 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
20622 (compute_antic): Seed worklist with exit block predecessors.
20623 * cfganal.c (dfs_find_deadend): For a cycle return the source
20624 of the edge closing it.
20626 2017-08-01 Tamar Christina <tamar.christina@arm.com>
20628 * config/aarch64/aarch64.c
20629 (aarch64_can_const_movi_rtx_p): Move 0 check.
20631 2017-08-01 Bin Cheng <bin.cheng@arm.com>
20633 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
20634 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
20636 * match.pd: Ditto in address comparison pattern.
20638 2017-08-01 Bin Cheng <bin.cheng@arm.com>
20640 PR tree-optimization/81627
20641 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
20642 closed ssa form for store-store chain.
20644 2017-08-01 Bin Cheng <bin.cheng@arm.com>
20646 PR tree-optimization/81620
20647 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
20648 for store-store chain.
20650 2017-08-01 Jakub Jelinek <jakub@redhat.com>
20652 PR tree-optimization/81588
20653 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
20654 ranges[i].in_p, invert comparison code ccode. For >/>=,
20655 swap rhs1 and rhs2 and comparison code unconditionally,
20656 for </<= don't do that. Don't swap rhs1/rhs2 again if
20657 ranges[i].in_p, instead invert comparison code ccode if
20658 opcode or oe->rank is BIT_IOR_EXPR.
20661 * optabs.def (vec_extract_optab, vec_init_optab): Change from
20662 a direct optab to conversion optab.
20663 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
20664 with GET_MODE_INNER as last argument instead of optab_handler.
20665 * expmed.c (extract_bit_field_1): Likewise. Use vector from
20666 vector extraction if possible and optab is available.
20667 * expr.c (store_constructor): Use convert_optab_handler instead
20668 of optab_handler. Use vector initialization from smaller
20669 vectors if possible and optab is available.
20670 * tree-vect-stmts.c (vectorizable_load): Likewise.
20671 * doc/md.texi (vec_extract, vec_init): Document that the optabs
20672 now have two modes.
20673 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
20674 of vec_init from half-sized vectors with the same element mode.
20675 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
20676 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
20677 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
20678 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
20679 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
20680 after mode in gen_vec_extract* calls.
20681 (vec_extract<mode>): Renamed to ...
20682 (vec_extract<mode><ssescalarmodelower>): ... this.
20683 (vec_extract<mode><ssehalfvecmodelower>): New expander.
20684 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
20685 element mode after mode in gen_vec_init* calls.
20686 (VEC_INIT_HALF_MODE): New mode iterator.
20687 (vec_init<mode>): Renamed to ...
20688 (vec_init<mode><ssescalarmodelower>): ... this.
20689 (vec_init<mode><ssehalfvecmodelower>): New expander.
20690 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
20691 (vec_extractv2sfsf): ... this.
20692 (vec_initv2sf): Renamed to ...
20693 (vec_initv2sfsf): ... this.
20694 (vec_extractv2si): Renamed to ...
20695 (vec_extractv2sisi): ... this.
20696 (vec_initv2si): Renamed to ...
20697 (vec_initv2sisi): ... this.
20698 (vec_extractv4hi): Renamed to ...
20699 (vec_extractv4hihi): ... this.
20700 (vec_initv4hi): Renamed to ...
20701 (vec_initv4hihi): ... this.
20702 (vec_extractv8qi): Renamed to ...
20703 (vec_extractv8qiqi): ... this.
20704 (vec_initv8qi): Renamed to ...
20705 (vec_initv8qiqi): ... this.
20706 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
20707 (vec_init<mode>): Renamed to ...
20708 (vec_init<mode><VEC_base_l>): ... this.
20709 (vec_extract<mode>): Renamed to ...
20710 (vec_extract<mode><VEC_base_l>): ... this.
20711 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
20712 (vec_initv2sfsf): ... this.
20713 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
20714 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
20715 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
20716 element mode after mode in gen_vec_init* calls.
20717 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
20718 (vec_init<mode><Vel>): ... this.
20719 (vec_extract<mode>): Renamed to ...
20720 (vec_extract<mode><Vel>): ... this.
20721 * config/aarch64/iterators.md (Vel): New mode attribute.
20722 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
20723 Add element mode after mode in gen_vec_extract* calls.
20724 * config/s390/vector.md (non_vec_l): New mode attribute.
20725 (vec_extract<mode>): Renamed to ...
20726 (vec_extract<mode><non_vec_l>): ... this.
20727 (vec_init<mode>): Renamed to ...
20728 (vec_init<mode><non_vec_l>): ... this.
20729 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
20730 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
20732 * config/arm/iterators.md (V_elem_l): New mode attribute.
20733 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
20734 (vec_extract<mode><V_elem_l>): ... this.
20735 (vec_extractv2di): Renamed to ...
20736 (vec_extractv2didi): ... this.
20737 (vec_init<mode>): Renamed to ...
20738 (vec_init<mode><V_elem_l>): ... this.
20739 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
20740 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
20741 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
20742 Add element mode after gen_vec_extract* calls.
20743 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
20744 (vec_init<mode><unitmode>): ... this.
20745 (vec_extract<mode>): Renamed to ...
20746 (vec_extract<mode><unitmode>): ... this.
20747 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
20748 (vec_init<mode><unitmode>): ... this.
20749 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
20750 (vec_initv2sfsf): ... this.
20751 (vec_extractv2sf): Renamed to ...
20752 (vec_extractv2sfsf): ... this.
20753 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
20754 Add element mode after gen_vec_extract* calls.
20755 * config/mips/mips.md (unitmode): New mode iterator.
20756 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
20757 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
20758 * config/spu/spu.md (inner_l): New mode attribute.
20759 (vec_init<mode>): Renamed to ...
20760 (vec_init<mode><inner_l>): ... this.
20761 (vec_extract<mode>): Renamed to ...
20762 (vec_extract<mode><inner_l>): ... this.
20763 * config/sparc/sparc.md (veltmode): New mode iterator.
20764 (vec_init<VMALL:mode>): Renamed to ...
20765 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
20766 * config/ia64/vect.md (vec_initv2si): Renamed to ...
20767 (vec_initv2sisi): ... this.
20768 (vec_initv2sf): Renamed to ...
20769 (vec_initv2sfsf): ... this.
20770 (vec_extractv2sf): Renamed to ...
20771 (vec_extractv2sfsf): ... this.
20772 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
20773 (vec_init<mode>): Renamed to ...
20774 (vec_init<mode><VEC_base_l>): ... this.
20775 (vec_extract<mode>): Renamed to ...
20776 (vec_extract<mode><VEC_base_l>): ... this.
20777 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
20778 (vec_initv2sfsf): ... this.
20779 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
20780 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
20781 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
20782 gen_vec_init* calls.
20784 2017-08-01 Richard Biener <rguenther@suse.de>
20786 PR tree-optimization/81297
20787 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
20788 TREE_OVERFLOW from INTEGER_CSTs.
20790 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
20792 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
20794 2017-07-31 Carl Love <cel@us.ibm.com>
20796 * config/rs6000/rs6000-c: Add support for built-in functions
20797 vector signed char vec_xl_be (signed long long, signed char *);
20798 vector unsigned char vec_xl_be (signed long long, unsigned char *);
20799 vector signed int vec_xl_be (signed long long, signed int *);
20800 vector unsigned int vec_xl_be (signed long long, unsigned int *);
20801 vector signed long long vec_xl_be (signed long long, signed long long *);
20802 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
20803 vector signed short vec_xl_be (signed long long, signed short *);
20804 vector unsigned short vec_xl_be (signed long long, unsigned short *);
20805 vector double vec_xl_be (signed long long, double *);
20806 vector float vec_xl_be (signed long long, float *);
20807 * config/rs6000/altivec.h (vec_xl_be): Add #define.
20808 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
20809 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
20811 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
20812 (altivec_expand_builtin): Add switch statement to call
20813 altivec_expand_xl_be for each builtin.
20814 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
20815 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
20816 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
20817 __builtin_vsx_le_be_v16qi.
20818 * doc/extend.texi: Update the built-in documentation file for the
20819 new built-in functions.
20821 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
20824 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
20826 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
20828 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20830 * config.gcc: Add z14.
20831 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
20832 CPU model numbers for z13s and z14.
20833 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
20835 * config/s390/s390-opts.h (enum processor_type): Rename
20836 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
20837 * config/s390/s390.c (processor_table): Add field for CPU name to
20838 be passed to Binutils.
20839 (s390_asm_output_machine_for_arch): Use the new field in
20840 processor_table for Binutils.
20841 (s390_expand_builtin): Replace arch12 with z14.
20842 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
20843 (s390_get_sched_attrmask): Likewise.
20844 (s390_get_unit_mask): Likewise.
20845 * config/s390/s390.opt: Add z14 to processor_type enum.
20847 2017-07-31 Martin Jambor <mjambor@suse.cz>
20850 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
20851 regardless of optimization level.
20853 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
20854 Martin Liska <mliska@suse.cz>
20856 * predict.def: Remove old comment and adjust probability.
20857 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
20858 PREDICT statements.
20860 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
20863 * config/i386/i386.c (ix86_function_naked): New function.
20864 (ix86_can_use_return_insn_p): Return false for naked functions.
20865 (ix86_expand_prologue): Skip prologue for naked functions.
20866 (ix86_expand_epilogue): Skip epilogue for naked functions
20867 and emit trap instruction.
20868 (ix86_warn_func_return): New function.
20869 (ix86_attribute_table): Add "naked" attribute specification.
20870 (TARGET_WARN_FUNC_RETURN): Define.
20871 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
20873 2017-07-31 Martin Liska <mliska@suse.cz>
20875 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
20876 (dump_gimple_bb_header): Always dump BB info.
20877 (pp_cfg_jump): Do not append info about BB when dumping a jump.
20879 2017-07-31 Martin Liska <mliska@suse.cz>
20882 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
20883 also with current_function_decl non-null equality.
20885 2017-07-31 Jakub Jelinek <jakub@redhat.com>
20888 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
20889 change type to the element type, instead add eltype variable and
20890 use it where we are interested in the element type.
20892 PR tree-optimization/81603
20893 * ipa-polymorphic-call.c
20894 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
20895 offset arithmetic in offset_int, bail out if the resulting bit offset
20896 doesn't fit into shwi.
20898 2017-07-31 Martin Liska <mliska@suse.cz>
20900 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
20901 (gimplify_save_expr): Fix comment.
20903 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
20906 * config/i386/i386.c (ix86_function_arg): Update arguments for
20908 (ix86_compute_frame_layout): Set the initial stack offset to
20909 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
20910 INCOMING_FRAME_SP_OFFSET.
20911 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
20912 stack before exception handler returns.
20913 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
20914 the 'ERROR_CODE' for exception handler.
20916 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
20918 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
20919 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
20920 (ASM_OUTPUT_REG_POP): Ditto.
20921 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
20922 instead of asm_fprintf to output pure string.
20924 2017-07-29 Jakub Jelinek <jakub@redhat.com>
20926 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
20927 to imported_module_or_decl hook.
20928 (debug_nothing_tree_tree_tree_bool): Remove.
20929 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
20930 * debug.c (do_nothing_debug_hooks): Use
20931 debug_nothing_tree_tree_tree_bool_bool instead of
20932 debug_nothing_tree_tree_tree_bool.
20933 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
20934 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
20935 * sdbout.c (sdb_debug_hooks): Likewise.
20936 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
20937 (gen_namespace_die): Add DW_AT_export_symbols attribute if
20939 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
20940 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
20941 attribute, don't add anything.
20943 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20945 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
20946 (fold_build2_stat_loc): Likewise.
20947 (fold_build3_stat_loc): Likewise.
20948 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
20949 (fold_build1_loc): Remove macro.
20950 (fold_build2_loc): Likewise.
20951 (fold_build3_loc): Likewise.
20953 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20955 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
20956 (gimple_build_debug_bind_source_stat): Likewise.
20957 * gimple.h (gimple_build_debug_bind): Remove macro.
20958 (gimple_build_debug_bind_source): Likewise.
20960 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20962 * bitmap.c (bitmap_alloc): Adjust.
20963 (bitmap_gc_alloc): Likewise.
20964 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
20966 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20968 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
20969 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
20970 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
20971 (bitmap_gc_alloc_stat): Likewise.
20972 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
20974 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20976 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
20977 * rtl.h (shallow_copy_rtx): Remove macro.
20979 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20981 * emit-rtl.c (gen_raw_REG): Adjust.
20982 * gengenrtl.c (gendef): Likewise.
20983 * rtl.c (rtx_alloc_stat): Remove _stat from name.
20984 * rtl.h (rtx_alloc): Remove macro.
20986 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20988 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
20989 (build_tree_list_stat): Likewise.
20990 * tree.h (build_tree_list): Remove macro.
20991 (build_tree_list_vec): Likewise.
20993 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20995 * tree.c (make_vector_stat): Remove _stat from name.
20996 (build_vector_stat): Likewise.
20997 * tree.h (make_vector_stat): Remove macro.
20998 (build_vector_stat): Likewise.
21000 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21002 * tree.h (build_var_debug_value): Remove prototype.
21004 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21006 * tree.c (tree_cons_stat): Remove _stat from name.
21007 * tree.h (tree_cons): Remove macro.
21009 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21011 * tree.c (build_vl_exp_stat): Remove _stat from name.
21012 * tree.h (build_vl_exp): Remove macro.
21014 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21016 * tree.c (build_decl_stat): Remove _stat from name.
21017 * tree.h (build_decl): Remove macro.
21019 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21021 * gimple.c (gimple_build_with_ops_stat): Adjust.
21022 (gimple_alloc_stat): Remove _stat from name.
21023 * gimple.h (gimple_alloc): Remove macro.
21025 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21027 * tree.c (make_tree_vec_stat): Remove _stat from name.
21028 (grow_tree_vec_stat): Likewise.
21029 * tree.h (make_tree_vec_stat): Adjust prototype.
21030 (grow_tree_vec_stat): Likewise.
21031 (make_tree_vec): Remove macro.
21032 (grow_tree_vec): Likewise.
21034 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21036 * fold-const.c (fold_build1_stat_loc): Adjust.
21037 (fold_build2_stat_loc): Likewise.
21038 (fold_build3_stat_loc): Likewise.
21039 * tree.c (build0_stat): Remove _stat from name.
21040 (build1_stat): Likewise.
21041 (build2_stat): Likewise.
21042 (build3_stat): Likewise.
21043 (build4_stat): Likewise.
21044 (build5_stat): Likewise.
21045 * tree.h (build1_loc): Remove macro, and rename _stat function
21047 (build2_loc): Likewise.
21048 (build3_loc): Likewise.
21049 (build4_loc): Likewise.
21050 (build5_loc): Likewise.
21052 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21054 * tree.c (make_int_cst_stat): Remove _stat from name.
21055 * tree.h (make_int_cst_stat): Adjust prototype.
21056 (make_int_cst): Remove macro.
21058 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21060 * tree.c (make_tre_binfo_stat): Remove _stat from name.
21061 * tree.h (make_tree_binfo_stat): Adjust prototype.
21062 (make_tree_binfo): Remove.
21064 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21066 * tree.c (copy_node_stat): Rename to copy_node.
21067 (build_distinct_type_copy): Adjust.
21068 * tree.h (copy_node_stat): Adjust prototype.
21069 (copy_node): Remove macro.
21071 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21073 * tree.c (make_node_stat): rename to make_node.
21074 (build_tree_list_stat): Adjust.
21075 (build0_stat): Likewise.
21076 (build2_stat): Likewise.
21077 (build3_stat): Likewise.
21078 (build4_stat): Likewise.
21079 (build5_stat): Likewise.
21080 (build_decl_stat): Likewise.
21081 * tree.h (make_node_stat): Adjust prototype.
21082 (make_node): remove macro.
21084 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
21086 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
21087 (PPC_FEATURE2_SCV): Likewise.
21088 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
21090 2017-07-28 Tamar Christina <tamar.christina@arm.com>
21092 * config/aarch64/aarch64.c
21093 (aarch64_internal_mov_immediate): Add new special pattern.
21094 * config/aarch64/aarch64.md (*movdi_aarch64):
21095 Add reg/32bit const mov case.
21097 2017-07-28 Tamar Christina <tamar.christina@arm.com>
21098 Richard Sandiford <richard.sandiford@linaro.org>
21100 * config/aarch64/aarch64.md (mov<mode>): Generalize.
21101 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
21102 Add integer and movi cases.
21103 (movi-split-hf-df-sf split, fp16): New.
21104 (enabled): Added TARGET_FP_F16INST.
21105 * config/aarch64/iterators.md (GPF_HF): New.
21106 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
21108 2017-07-28 Tamar Christina <tamar.christina@arm.com>
21110 * config/aarch64/aarch64.c
21111 (aarch64_simd_container_mode): Add prototype.
21112 (aarch64_expand_mov_immediate): Add HI support.
21113 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
21114 (aarch64_can_const_movi_rtx_p): New.
21115 (aarch64_preferred_reload_class):
21116 Remove restrictions of using FP registers for certain SIMD operations.
21117 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
21118 (aarch64_valid_floating_const): Add integer move validation.
21119 (aarch64_simd_imm_scalar_p): Remove.
21120 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
21121 (aarch64_legitimate_constant_p): Expand list of supported cases.
21122 * config/aarch64/aarch64-protos.h
21123 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
21124 (aarch64_reinterpret_float_as_int): New.
21125 (aarch64_simd_imm_scalar_p): Remove.
21126 * config/aarch64/constraints.md (Uvi): New.
21127 (Dd): Split into Ds and new Dd.
21128 * config/aarch64/aarch64.md (*movsi_aarch64):
21130 (*movdi_aarch64): Add SIMD mov case.
21132 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21134 * tree-predcom.c: (struct chain): Handle store-store chain in which
21135 stores for elimination only store loop invariant values.
21136 (execute_pred_commoning_chain): Ditto.
21137 (prepare_initializers_chain_store_elim): Ditto.
21138 (prepare_finalizers): Ditto.
21139 (is_inv_store_elimination_chain): New function.
21140 (initialize_root_vars_store_elim_1): New function.
21142 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21144 * tree-predcom.c: Revise general description of the pass.
21145 (enum chain_type): New enum type for store elimination.
21146 (struct chain): New field supporting store elimination.
21147 (struct component): Ditto.
21148 (dump_chain): Dump store-stores chain.
21149 (release_chain): Release resources.
21150 (split_data_refs_to_components): Compute and create component
21151 contains only stores for elimination.
21152 (get_chain_last_ref_at): New function.
21153 (make_invariant_chain): Initialization.
21154 (make_rooted_chain): Specify chain type in parameter and record it.
21155 (add_looparound_copies): Skip for store-stores chain.
21156 (determine_roots_comp): Compute type of chain and pass it to
21158 (initialize_root_vars_store_elim_2): New function.
21159 (finalize_eliminated_stores): New function.
21160 (remove_stmt): Handle store for elimination.
21161 (execute_pred_commoning_chain): Execute predictive commoning on
21162 store-store chains.
21163 (determine_unroll_factor): Skip unroll for store-stores chain.
21164 (prepare_initializers_chain_store_elim): New function.
21165 (prepare_initializers_chain): Hanlde store-store chain.
21166 (prepare_finalizers_chain, prepare_finalizers): New function.
21167 (tree_predictive_commoning_loop): Return integer value indicating
21168 if loop is unrolled or lcssa form is corrupted.
21169 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
21171 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21173 * tree-predcom.c (initialize_root): Delete.
21174 (execute_pred_commoning_chain): Initialize root vars and replace
21175 reference of non-combined chain directly, rather than call above
21178 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21180 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
21181 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
21183 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21185 * tree-predcom.c (struct chain): New field init_seq.
21186 (release_chain): Release init_seq.
21187 (prepare_initializers_chain): Record intialization stmts in above
21189 (insert_init_seqs): New function.
21190 (tree_predictive_commoning_loop): Call insert_init_seqs.
21192 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21194 * tree-predcom.c (determine_roots_comp): Skip trivial components.
21196 2017-07-28 Richard Biener <rguenther@suse.de>
21198 * match.pd: Remove superfluous :c.
21199 * genmatch.c (simplify::id): Add member.
21200 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
21202 (current_id): New global.
21203 (dt_node::parent): Move from ...
21204 (dt_operand::parent): ... here. Add for_id member.
21205 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
21206 (decision_tree::find_node): Relax order requirement when
21207 merging DT_TRUE nodes to ones inbetween the current simplify
21208 and the one we try to merge with. Add diagnostic whenever
21209 we need to enforce pattern order by not merging.
21210 (decision_tree::insert): Set current_id.
21211 (decision_tree::print_node): Dump parent node and for_id.
21212 (parser::last_id): Add member.
21213 (parser::push_simplify): Assign unique id.
21214 (parser::parser): Initialize last_id.
21216 2017-07-28 Martin Liska <mliska@suse.cz>
21219 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
21220 gimple_build_debug_bind.
21222 2017-07-28 Richard Biener <rguenther@suse.de>
21224 PR tree-optimization/81502
21225 * match.pd: Add pattern combining BIT_INSERT_EXPR with
21227 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
21229 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
21230 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
21231 for BIT_FIELD_REF args.
21232 * fold-const.c (make_bit_field_ref): Likewise.
21233 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
21235 2017-07-28 Jakub Jelinek <jakub@redhat.com>
21238 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
21239 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
21240 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
21241 Or it into SANITIZER_UNDEFINED.
21242 * ubsan.c: Include gimple-fold.h and varasm.h.
21243 (ubsan_expand_ptr_ifn): New function.
21244 (instrument_pointer_overflow): New function.
21245 (maybe_instrument_pointer_overflow): New function.
21246 (instrument_object_size): Formatting fix.
21247 (pass_ubsan::execute): Call instrument_pointer_overflow
21248 and maybe_instrument_pointer_overflow.
21249 * internal-fn.c (expand_UBSAN_PTR): New function.
21250 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
21251 * sanitizer.def (__ubsan_handle_pointer_overflow,
21252 __ubsan_handle_pointer_overflow_abort): New builtins.
21253 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
21254 * internal-fn.def (UBSAN_PTR): New internal function.
21255 * opts.c (sanitizer_opts): Add pointer-overflow.
21256 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
21257 * fold-const.c (build_range_check): Compute pointer range check in
21258 integral type if pointer arithmetics would be needed. Formatting
21261 2017-07-28 Martin Liska <mliska@suse.cz>
21264 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
21265 parameters that are of a variable-length.
21267 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
21269 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
21271 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
21272 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
21273 (CRT_CALL_STATIC_FUNCTION): Likewise.
21274 (ASM_DEFAULT_SPEC): New define.
21275 (ASM_SPEC32): Likewise.
21276 (ASM_SPEC64): Likewise.
21277 (ASM_SPEC_COMMON): Likewise.
21278 (ASM_SPEC): Likewise.
21279 (INVALID_64BIT): Likewise.
21280 (LINK_OS_DEFAULT_SPEC): Likewise.
21281 (LINK_OS_SPEC32): Likewise.
21282 (LINK_OS_SPEC64): Likewise.
21283 (POWERPC_LINUX): Likewise.
21284 (PTRDIFF_TYPE): Likewise.
21285 (RESTORE_FP_PREFIX): Likewise.
21286 (RESTORE_FP_SUFFIX): Likewise.
21287 (SAVE_FP_PREFIX): Likewise.
21288 (SAVE_FP_SUFFIX): Likewise.
21289 (SIZE_TYPE): Likewise.
21290 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
21291 (TARGET_64BIT): Likewise.
21292 (TARGET_64BIT): Likewise.
21293 (TARGET_AIX): Likewise.
21294 (WCHAR_TYPE_SIZE): Likewise.
21295 (WCHAR_TYPE): Undefine.
21296 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
21297 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
21298 (CPP_OS_RTEMS_SPEC): Delete.
21299 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
21300 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
21302 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
21304 2017-07-28 Jakub Jelinek <jakub@redhat.com>
21306 PR tree-optimization/81578
21307 * tree-parloops.c (build_new_reduction): Bail out if
21308 reduction_code isn't one of the standard OpenMP reductions.
21309 Move the details printing after that decision.
21311 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
21313 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
21314 related to reload_in_progress.
21315 (splat_input_operand): Likewise.
21316 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
21318 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
21320 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
21321 (TARGET_INSTANTIATE_DECLS): Likewise.
21322 (legitimate_indexed_address_p): Delete reload_in_progress code.
21323 (rs6000_debug_legitimate_address_p): Likewise.
21324 (rs6000_eliminate_indexed_memrefs): Likewise.
21325 (rs6000_emit_le_vsx_store): Likewise.
21326 (rs6000_emit_move_si_sf_subreg): Likewise.
21327 (rs6000_emit_move): Likewise.
21328 (register_to_reg_type): Likewise.
21329 (rs6000_pre_atomic_barrier): Likewise.
21330 (rs6000_machopic_legitimize_pic_address): Likewise.
21331 (rs6000_allocate_stack_temp): Likewise.
21332 (rs6000_address_for_fpconvert): Likewise.
21333 (rs6000_address_for_altivec): Likewise.
21334 (rs6000_secondary_memory_needed_rtx): Delete function.
21335 (rs6000_check_sdmode): Likewise.
21336 (rs6000_alloc_sdmode_stack_slot): Likewise.
21337 (rs6000_instantiate_decls): Likewise.
21338 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
21339 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
21340 Delete reload_in_progress.
21341 (*vec_reload_and_plus_<mptrsize>): Likewise.
21342 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
21343 (vsx_div_v2di): Likewise.
21344 (vsx_udiv_v2di): Likewise.
21346 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
21348 * config/rs6000/rs6000.opt (mlra): Replace with stub.
21349 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
21350 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
21351 (rs6000_debug_reg_global): Delete print of LRA status.
21352 (rs6000_option_override_internal): Delete dead LRA related code.
21353 (rs6000_lra_p): Delete function.
21354 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
21356 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
21358 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
21359 * config/riscv/rtems.h: New file.
21361 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21362 Sudakshina Das <sudi.das@arm.com>
21364 * config/aarch64/aarch64.md
21365 (define_split for and<mode>3nr_compare): Move
21366 non aarch64_logical_operand to a register.
21367 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
21368 register immediate operand to a register.
21369 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
21371 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
21373 PR middle-end/81564
21374 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
21376 2017-07-27 Richard Biener <rguenther@suse.de>
21378 PR tree-optimization/81573
21379 PR tree-optimization/81494
21380 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
21381 multi defuse cycle case.
21383 2017-07-27 Richard Biener <rguenther@suse.de>
21385 PR tree-optimization/81571
21386 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
21389 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
21391 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
21392 earlier and only if MASK_FPU is set. Adjust formatting.
21394 2017-07-27 Martin Liska <mliska@suse.cz>
21396 * opt-functions.awk: Add validation of value of Init.
21397 * optc-gen.awk: Pass new argument.
21399 2017-07-27 Martin Liska <mliska@suse.cz>
21401 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
21402 Fix wrong condition.
21404 2017-07-27 Martin Liska <mliska@suse.cz>
21406 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
21407 BBs and edges seen by autoFDO.
21409 2017-07-27 Richard Biener <rguenther@suse.de>
21411 PR tree-optimization/81502
21412 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
21413 with incompatible but same sized type.
21414 (execute_update_addresses_taken): Likewise.
21416 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
21418 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
21419 flag_tree_loop_vectorize rather than flag_tree_vectorize.
21421 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21424 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
21425 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
21426 Change s_operand to memory_operand.
21428 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
21430 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
21431 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
21432 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
21433 Emit instructions rather than returning an expression. Handle TFmode
21434 and KFmode by casting to TImode.
21435 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
21436 (rs6000_emit_le_vsx_store): Likewise.
21437 * config/rs6000/vsx.md (VSX_TI): New iterator.
21438 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
21439 (*vsx_le_undo_permute_<mode>): Likewise.
21440 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
21441 emit the split sequence.
21442 (*vsx_le_perm_store_<mode>): Likewise.
21444 2017-07-27 Jakub Jelinek <jakub@redhat.com>
21446 PR tree-optimization/81555
21447 PR tree-optimization/81556
21448 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
21449 if true, force CHANGED for the recursive invocation.
21450 (reassociate_bb): Remember original length of ops array, pass
21451 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
21453 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
21454 attributes for noipa attribute. For naked attribute use
21455 lookup_attribute first before lookup_attribute_spec.
21456 * final.c (rest_of_handle_final): Disable IPA RA for functions with
21458 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
21459 for functions with noipa attribute.
21460 (cgraph_externally_visible_p): Return true for functions with noipa
21462 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
21463 for functions with noipa attribute.
21464 * doc/extend.texi: Document noipa function attribute.
21465 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
21466 also for functions with noipa attribute.
21467 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
21469 2017-07-26 Andrew Pinski <apinski@cavium.com>
21471 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
21472 vec_unalign_load_cost and vec_unalign_store_cost.
21474 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
21476 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
21477 -mvsx-small-integer option.
21478 (ISA_3_0_MASKS_IEEE): Likewise.
21479 (OTHER_VSX_VECTOR_MASKS): Likewise.
21480 (POWERPC_MASKS): Likewise.
21481 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
21482 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
21483 code, only testing for DImode being allowed in non-VSX floating
21485 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
21486 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
21488 (rs6000_option_override_internal): Delete -mvsx-small-integer.
21489 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
21490 TARGET_P8_VECTOR test.
21491 (rs6000_secondary_reload_simple_move): Likewise.
21492 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
21493 since TARGET_P9_VECTOR was already tested.
21494 (rs6000_opt_masks): Remove -mvsx-small-integer.
21495 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
21496 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
21498 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
21499 test for TARGET_VEXTRACTUB was used, and that uses
21501 (p9 extract splitter): Likewise.
21502 (vsx_extract_<mode>_di_p9): Likewise.
21503 (vsx_extract_<mode>_store_p9): Likewise.
21504 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
21505 for TARGET_P9_VECTOR was used. Delete code that is now dead with
21506 the elimination of TARGET_VSX_SMALL_INTEGER.
21507 (vsx_extract_<mode>_p8): Likewise.
21508 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
21509 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
21510 (vsx_set_<mode>_p9): Likewise.
21511 (vsx_set_v4sf_p9): Likewise.
21512 (vsx_set_v4sf_p9_zero): Likewise.
21513 (vsx_insert_extract_v4sf_p9): Likewise.
21514 (vsx_insert_extract_v4sf_p9_2): Likewise.
21515 * config/rs6000/rs6000.md (sign extend splitter): Change
21516 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
21517 (floatsi<mode>2_lfiwax_mem): Likewise.
21518 (floatunssi<mode>2_lfiwzx_mem): Likewise.
21519 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
21520 since a test for TARGET_P9_VECTOR was used.
21521 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
21522 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
21523 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
21524 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
21525 TARGET_P8_VECTOR test.
21526 (fix_trunc<mode>si2_stfiwx): Likewise.
21527 (fix_trunc<mode>si2_internal): Likewise.
21528 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
21529 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
21531 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
21532 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
21533 TARGET_P8_VECTOR test.
21534 (fixuns_trunc<mode>si2_stfiwx): Likewise.
21535 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
21536 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
21538 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
21539 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
21540 since a test for TARGET_P9_VECTOR was used.
21541 (splitter for loading small constants): Likewise.
21543 2017-07-26 Andrew Pinski <apinski@cavium.com>
21545 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
21548 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
21551 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
21552 (fp_valid_at): Likewise.
21554 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
21556 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
21557 (qdf24xx_addrcost_table): Likewise.
21558 (cortexa57_tunings): Update to use generic_branch_cost.
21559 (cortexa72_tunings): Likewise.
21560 (cortexa73_tunings): Likewise.
21561 (qdf24xx_tunings): Likewise.
21563 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
21565 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
21566 (thunderx2t99_branch_cost): Likewise.
21567 (cortexa35_tunings): Update to use generic_branch_cost.
21568 (cortexa53_tunings): Likewise.
21569 (cortexa57_tunings): Likewise.
21570 (cortexa72_tunings): Likewise.
21571 (cortexa73_tunings): Likewise.
21572 (thunderx2t99_tunings): Likewise.
21574 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
21576 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
21577 (sparc_option_override): Honour MASK_FSMULD.
21578 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
21579 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
21580 * config/sparc/sparc.opt (mfsmuld): New option.
21581 * doc/invoke.texi (mfsmuld): Document option.
21583 2017-07-26 Marek Polacek <polacek@redhat.com>
21585 PR middle-end/70992
21586 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
21588 2017-07-26 Richard Biener <rguenther@suse.de>
21590 * gimple-match-head.c (do_valueize): Return OP if valueize
21592 (get_def): New helper to get at the def stmt of a SSA name
21593 if valueize allows.
21594 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
21595 do_valueize to get at the def stmt.
21596 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
21598 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
21600 PR middle-end/46932
21601 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
21603 2017-07-26 Martin Liska <mliska@suse.cz>
21606 * function.c (expand_function_start): Make expansion of
21607 nonlocal_goto_save_area after parm_birth_insn.
21609 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
21611 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
21612 from all CPU target flags enable members.
21614 2017-07-26 Richard Biener <rguenther@suse.de>
21616 * genmatch.c (dt_simplify::gen): Make iterator vars const.
21617 (decision_tree::gen): Make 'type' const.
21618 (write_predicate): Likewise.
21620 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
21622 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
21623 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
21624 (rs6000_option_override_internal): Likewise.
21625 (rs6000_expand_vector_set): Likewise.
21626 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
21627 (TARGET_UPPER_REGS_SF): Likewise.
21628 (TARGET_UPPER_REGS_DI): Likewise.
21629 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
21630 (TARGET_DIRECT_MOVE_64BIT): Likewise.
21631 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
21632 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
21633 (Splitters for DI constants in Altivec registers): Likewise.
21634 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
21635 (vsx_set_v4sf_p9): Likewise.
21636 (vsx_set_v4sf_p9_zero): Likewise.
21637 (vsx_insert_extract_v4sf_p9): Likewise.
21638 (vsx_insert_extract_v4sf_p9_2): Likewise.
21640 2017-07-25 Carl Love <cel@us.ibm.com>
21642 * doc/extend.texi: Update the built-in documentation file for the
21643 existing built-in functions
21644 vector signed char vec_cnttz (vector signed char);
21645 vector unsigned char vec_cnttz (vector unsigned char);
21646 vector signed short vec_cnttz (vector signed short);
21647 vector unsigned short vec_cnttz (vector unsigned short);
21648 vector signed int vec_cnttz (vector signed int);
21649 vector unsigned int vec_cnttz (vector unsigned int);
21650 vector signed long long vec_cnttz (vector signed long long);
21651 vector unsigned long long vec_cnttz (vector unsigned long long);
21653 2017-07-25 Andrew Pinski <apinski@cavium.com>
21655 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
21656 accesses where the use is for the first operand of a BIT_INSERT.
21658 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
21661 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
21662 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
21664 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
21666 * config/i386/gstabs.h: Delete.
21667 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
21669 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
21671 * config/i386/i386.c (ix86_decompose_address): Do not check for
21672 register RTX when looking at index_reg or base_reg.
21673 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
21675 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
21677 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
21678 to update EH info here.
21680 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
21682 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
21684 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
21686 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
21688 2017-07-25 Torsten Duwe <duwe@suse.de>
21690 * common.opt: Introduce -fpatchable-function-entry
21691 command line option, and its variables function_entry_patch_area_size
21692 and function_entry_patch_area_start.
21693 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
21694 including a two-value parser.
21695 * target.def (print_patchable_function_entry): New target hook.
21696 * targhooks.h (default_print_patchable_function_entry): New function.
21697 * targhooks.c (default_print_patchable_function_entry): Likewise.
21698 * toplev.c (process_options): Switch off IPA-RA if
21699 patchable function entries are being generated.
21700 * varasm.c (assemble_start_function): Look at the
21701 patchable-function-entry command line switch and current
21702 function attributes and maybe generate NOP instructions by
21703 calling the print_patchable_function_entry hook.
21704 * doc/extend.texi: Document patchable_function_entry attribute.
21705 * doc/invoke.texi: Document -fpatchable_function_entry
21706 command line option.
21707 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
21709 * doc/tm.texi: Re-generate.
21711 2017-07-25 Jakub Jelinek <jakub@redhat.com>
21714 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
21715 TARGET_AVX512DQ rather than TARGET_AVX512BW.
21717 2017-07-25 Tamar Christina <tamar.christina@arm.com>
21719 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
21721 2017-07-25 Richard Biener <rguenther@suse.de>
21723 PR tree-optimization/81455
21724 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
21725 not walk in cycles when looking for guards.
21727 2017-07-25 Richard Biener <rguenther@suse.de>
21729 PR tree-optimization/81529
21730 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
21731 when optimizing backedge uses.
21733 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
21735 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
21737 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
21738 to dl_section_ref. On AIX, append an expression to subtract
21739 the size of the section length to dl_section_ref.
21741 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
21743 * configure.ac: If any of the config.* scripts fail, exit 1.
21744 * configure: Regenerate.
21746 2017-07-25 Richard Biener <rguenther@suse.de>
21748 PR middle-end/81546
21749 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
21750 of immediate uses, be more verbose on errors.
21752 2017-07-25 Richard Biener <rguenther@suse.de>
21754 PR tree-optimization/81510
21755 * tree-vect-loop.c (vect_is_simple_reduction): When the
21756 reduction stmt is not inside the loop bail out.
21758 2017-07-25 Richard Biener <rguenther@suse.de>
21760 PR tree-optimization/81303
21761 * tree-vect-loop-manip.c (vect_loop_versioning): Build
21762 profitability check against LOOP_VINFO_NITERSM1.
21764 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
21766 * domwalk.c (cmp_bb_postorder): Simplify.
21767 (sort_bbs_postorder): New function. Use it...
21768 (dom_walker::walk): ...here to optimize common cases.
21770 2017-07-25 Martin Liska <mliska@suse.cz>
21773 * ipa-visibility.c (function_and_variable_visibility): Make the
21774 redirection just on target that supports aliasing.
21775 Fix GNU coding style.
21777 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
21780 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
21782 * config/arm/rtems.h: Likewise.
21783 * config/bfin/rtems.h: Likewise.
21784 * config/i386/rtemself.h: Likewise.
21785 * config/lm32/rtems.h: Likewise.
21786 * config/m32c/rtems.h: Likewise.
21787 * config/m68k/rtemself.h: Likewise.
21788 * config/microblaze/rtems.h: Likewise.
21789 * config/mips/rtems.h: Likewise.
21790 * config/moxie/rtems.h: Likewise.
21791 * config/nios2/rtems.h: Likewise.
21792 * config/powerpcspe/rtems.h: Likewise.
21793 * config/rs6000/rtems.h: Likewise.
21794 * config/rtems.h: Likewise.
21795 * config/sh/rtems.h: Likewise.
21796 * config/sh/rtemself.h: Likewise.
21797 * config/sparc/rtemself.h: Likewise.
21799 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
21802 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
21803 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
21804 * tree-ssa-structalias.c (alias_get_name): Same.
21806 2017-07-25 Bin Cheng <bin.cheng@arm.com>
21809 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
21810 instructions if no du chain is found.
21812 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
21814 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
21816 2017-07-25 Richard Biener <rguenther@suse.de>
21818 PR middle-end/81505
21819 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
21822 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
21824 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
21825 upper-regs options.
21826 (ISA_2_7_MASKS_SERVER): Likewise.
21827 (ISA_3_0_MASKS_IEEE): Likewise.
21828 (OTHER_P8_VECTOR_MASKS): Likewise.
21829 (OTHER_VSX_VECTOR_MASKS): Likewise.
21830 (POWERPC_MASKS): Likewise.
21831 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
21832 duplicate list of options.
21833 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
21834 explicit -mupper-regs options.
21835 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
21836 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
21837 alias for -mupper-regs-df.
21838 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
21839 (rs6000_init_hard_regno_mode_ok): Likewise.
21840 (rs6000_option_override_internal): Likewise.
21841 (rs6000_opt_masks): Likewise.
21842 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
21843 options in terms of whether -mvsx or -mpower8-vector was used.
21844 (TARGET_UPPER_REGS_DI): Likewise.
21845 (TARGET_UPPER_REGS_SF): Likewise.
21846 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
21847 -mupper-regs-* options.
21849 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
21851 * passes.c (emergency_dump_function): Print some empty lines and a
21852 header before the RTL dump.
21854 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
21856 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
21858 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
21861 * config/aarch64/aarch64.c (aarch64_classify_symbol):
21862 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
21864 2017-07-24 Carl Love <cel@us.ibm.com>
21866 * config/rs6000/rs6000-c.c: Add support for built-in functions
21867 vector float vec_extract_fp32_from_shorth (vector unsigned short);
21868 vector float vec_extract_fp32_from_shortl (vector unsigned short);
21869 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
21870 vec_extract_fp_from_shortl): Add defines for the two builtins.
21871 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
21872 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
21874 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
21875 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
21876 * doc/extend.texi: Update the built-in documentation file for the
21877 new built-in function.
21879 2017-07-24 Jakub Jelinek <jakub@redhat.com>
21882 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
21884 * doc/generic.texi: Likewise.
21885 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
21886 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
21888 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
21890 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
21891 (aarch64_mls_elt_merge<mode>): Likewise.
21893 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
21895 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
21896 having __cxa_atexit.
21898 2017-07-23 Michael Collison <michael.collison@arm.com>
21900 * config/arm/arm.c (arm_option_override): Deprecate
21901 use of -mstructure-size-boundary.
21902 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
21903 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
21905 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21908 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
21909 Reduce cost estimate for direct moves.
21911 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
21914 * config/i386/i386.c (ix86_option_override_internal): Disable
21915 BMI, BMI2 and TBM instructions for -m16.
21917 2017-07-21 Carl Love <cel@us.ibm.com>
21919 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21920 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
21921 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
21922 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
21923 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
21924 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
21925 VMULOSW): New enum "unspec" values.
21926 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
21927 altivec_vmulosw): New patterns.
21928 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
21929 VMULOSW): Add definitions.
21931 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
21933 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
21934 (qdf24xx): Likewise.
21935 * config/aarch64/aarch64-options-extensions.def (rdma); New.
21936 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
21937 (AARCH64_FL_V8_1): Renumber.
21938 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
21939 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
21940 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
21941 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
21942 rdma to feature modifiers list.
21944 2017-07-21 Yury Gribov <tetra2005@gmail.com>
21946 PR middle-end/56727
21947 * ipa-visibility (function_and_variable_visibility): Convert
21948 recursive PLT call to direct call if appropriate.
21950 2017-07-21 Andrew Pinski <apinski@cavium.com>
21952 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
21953 operand 1 to see if the types precision matches.
21954 * fold-const.c (operand_equal_p): Likewise.
21956 2017-07-21 Richard Biener <rguenther@suse.de>
21958 PR tree-optimization/81303
21959 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
21960 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
21961 (vect_peeling_hash_get_lowest_cost): Adjust.
21962 (vect_enhance_data_refs_alignment): Likewise. Use
21963 vect_get_peeling_costs_all_drs to compute the penalty for no
21964 peeling to match up costs.
21966 2017-07-21 Richard Biener <rguenther@suse.de>
21968 PR tree-optimization/81500
21969 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
21970 we didn't identify a reduction path.
21972 2017-07-21 Tom de Vries <tom@codesourcery.com>
21973 Cesar Philippidis <cesar@codesourcery.com>
21975 PR gcov-profile/81442
21976 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
21979 2017-07-21 Tom de Vries <tom@codesourcery.com>
21982 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
21984 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
21985 nvptx_override_options_after_change.
21987 2017-07-21 Ulrich Drepper <drepper@redhat.com>
21989 * dwarf2out.c (output_file_names): Avoid double testing for
21990 dwarf_version >= 5.
21992 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
21994 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
21996 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
21998 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
22000 (try_crossjump_to_edge): Do not punt on partitioned functions.
22002 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
22004 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
22005 Put all BBs reachable only via paths crossing cold region to cold
22007 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
22009 2016-07-21 Richard Biener <rguenther@suse.de>
22011 PR tree-optimization/81303
22012 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
22013 into account prologue and epilogue iterations when raising
22014 min_profitable_iters to sth at least covering one vector iteration.
22016 2017-07-21 Tamar Christina <tamar.christina@arm.com>
22018 * config/arm/arm.c (arm_test_cpu_arch_dat):
22021 2017-07-20 Nathan Sidwell <nathan@acm.org>
22023 Remove TYPE_METHODS.
22024 * tree.h (TYPE_METHODS): Delete.
22025 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
22026 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
22027 (dbxout_type_methods): Scan TYPE_FIELDS.
22028 (dbxout_type): Don't check TYPE_METHODS here.
22029 * function.c (use_register_for_decl): Always ignore register for
22030 class types when not optimizing.
22031 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
22032 * tree.c (free_lang_data_in_type): Stitch out member functions and
22033 templates from TYPE_FIELDS.
22034 (build_distinct_type_copy, verify_type_variant,
22035 verify_type): Member fns are on TYPE_FIELDS.
22036 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
22037 * tree-pretty-print.c (dump_generic_node): Likewise.
22039 2017-07-20 Jakub Jelinek <jakub@redhat.com>
22042 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
22043 V2TImode and V4TImode.
22044 (ix86_expand_vector_extract): Likewise.
22045 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
22046 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
22047 (ssescalarmode): Handle V4TImode and V2TImode.
22048 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
22049 (*vec_extractv2ti, *vec_extractv4ti): New insns.
22050 (VEXTRACTI128_MODE): New mode iterator.
22051 (splitter for *vec_extractv?ti first element): New.
22052 (VEC_INIT_MODE): New mode iterator.
22053 (vec_init<mode>): Consolidate 3 expanders into one using
22054 VEC_INIT_MODE mode iterator.
22056 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
22058 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
22059 non_spilled_static_chain_regno_p.
22061 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
22063 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
22065 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
22067 * bb-reorder.c (connect_traces): Allow copying of blocks within
22070 2017-07-20 Richard Biener <rguenther@suse.de>
22072 * gimple.h (gimple_phi_result): Add gphi * overload.
22073 (gimple_phi_result_ptr): Likewise.
22074 (gimple_phi_arg): Likewise. Adjust index assert to only
22075 allow actual argument accesses rather than all slots available
22077 (gimple_phi_arg_def): Add gphi * overload.
22078 * tree-phinodes.c (make_phi_node): Initialize only actual
22080 (resize_phi_node): Clear memory not covered by old node,
22081 do not initialize excess argument slots.
22082 (reserve_phi_args_for_new_edge): Initialize new argument slot
22085 2017-07-20 Bin Cheng <bin.cheng@arm.com>
22087 PR tree-optimization/81388
22089 2016-07-21 Bin Cheng <bin.cheng@arm.com>
22091 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
22092 by removing computation of may_be_zero.
22094 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
22095 Tom de Vries <tom@codesourcery.com>
22097 PR middle-end/81030
22098 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
22099 when gimple level profile disagrees with what RTL expander did.
22101 2017-07-20 Richard Biener <rguenther@suse.de>
22103 PR tree-optimization/61171
22104 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
22105 (vect_analyze_stmt): Add slp instance parameter.
22106 (vectorizable_reduction): Likewise.
22107 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
22108 (vect_is_simple_reduction): Deal with chains not detected
22109 as SLP reduction chain, specifically not properly associated
22110 chains containing a mix of plus/minus.
22111 (get_reduction_op): Remove.
22112 (get_initial_defs_for_reduction): Simplify, pass in whether
22113 this is a reduction chain, pass in the SLP node for the PHIs.
22114 (vect_create_epilog_for_reduction): Get the SLP instance as
22116 (vectorizable_reduction): Get the SLP instance as arg.
22117 During analysis remember the SLP node with the PHIs in the
22118 instance. Simplify getting at the vectorized reduction PHIs.
22119 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
22120 through SLP instance.
22121 (vect_slp_analyze_operations): Likewise.
22122 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
22123 (vect_transform_stmt): Likewise.
22125 2017-07-20 Tom de Vries <tom@codesourcery.com>
22127 PR tree-optimization/81489
22128 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
22129 read of phi arg location to before loop that modifies phi.
22131 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22133 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
22136 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
22138 PR middle-end/81331
22139 * except.c (execute): Fix ordering issue.
22141 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
22143 PR rtl-optimization/81423
22144 * combine.c (make_compound_operation_int): Don't try to optimize
22145 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
22147 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
22149 PR rtl-optimization/81423
22150 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
22151 with a constant that is -1 in the truncated to mode.
22153 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
22155 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
22156 (determine_unlikely_bbs): ... here.
22157 * predict.h (propagate_unlikely_bbs_forward): Declare.
22158 * cfgexpand.c (pass_expand::execute): Use it.
22159 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
22161 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
22162 propagate_unlikely_bbs_forward.
22164 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
22166 PR middle-end/81331
22167 * except.c (maybe_add_nop_after_section_switch): New function.
22170 2017-07-19 Tom de Vries <tom@codesourcery.com>
22172 * gimple.h (gimple_phi_set_arg): Make assert more strict.
22174 2017-07-19 Tom de Vries <tom@codesourcery.com>
22176 * gimple.h (gimple_phi_arg): Make assert more strict.
22178 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
22180 * config.gcc (powerpc*-*-*): Add mmintrin.h.
22181 * config/rs6000/mmintrin.h: New file.
22182 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
22184 2017-07-19 Jakub Jelinek <jakub@redhat.com>
22186 PR tree-optimization/81346
22187 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
22189 2017-07-19 Tom de Vries <tom@codesourcery.com>
22191 * config/nvptx/nvptx.md (VECIM): Add V2DI.
22193 2017-07-19 Tom de Vries <tom@codesourcery.com>
22195 * config/nvptx/nvptx-modes.def: Add V2DImode.
22196 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
22197 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
22198 (nvptx_output_mov_insn): Handle lack of mov.b128.
22199 (nvptx_print_operand): Handle 'H' and 'L' codes.
22200 (nvptx_vector_mode_supported): Allow V2DImode.
22201 (nvptx_preferred_simd_mode): New function.
22202 (nvptx_data_alignment): New function.
22203 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
22204 nvptx_preferred_simd_mode.
22205 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
22207 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
22209 2017-07-19 Tom de Vries <tom@codesourcery.com>
22211 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
22212 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
22213 (nvptx_vector_mode_supported): New function. Allow V2SImode.
22214 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
22215 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
22216 (mov<VECIM>_insn): New define_insn.
22217 (define_expand "mov<VECIM>): New define_expand.
22219 2017-07-19 Tom de Vries <tom@codesourcery.com>
22221 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
22223 2017-07-19 Jakub Jelinek <jakub@redhat.com>
22225 PR tree-optimization/81346
22226 * fold-const.h (fold_div_compare, range_check_type): Declare.
22227 * fold-const.c (range_check_type): New function.
22228 (build_range_check): Use range_check_type.
22229 (fold_div_compare): No longer static, rewritten into
22230 a match.pd helper function.
22231 (fold_comparison): Don't call fold_div_compare here.
22232 * match.pd (X / C1 op C2): New optimization using fold_div_compare
22233 as helper function.
22235 2017-07-19 Nathan Sidwell <nathan@acm.org>
22237 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
22238 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
22239 * tree.c (find_decls_types_r, verify_type): Use
22240 TYPE_{MIN,MAX}_VALUE_RAW.
22241 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
22242 (hash_tree): Likewise.
22243 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
22245 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
22248 2017-07-18 Tom de Vries <tom@codesourcery.com>
22250 PR middle-end/81464
22251 * omp-expand.c (expand_omp_for_static_chunk): Handle
22252 equal-argument loop exit phi.
22254 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
22257 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
22258 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
22259 operand 2 predicate.
22260 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
22261 operand 2 predicate.
22262 (ror,rol -> rorx splitters): Use const_int_operand as
22263 operand 2 predicate.
22265 2017-06-18 Richard Biener <rguenther@suse.de>
22267 PR tree-optimization/81410
22268 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
22269 the gap in the ! slp_perm SLP case after each group.
22271 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
22273 PR middle-end/81463
22274 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
22277 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
22279 PR middle-end/81462
22280 * predict.c (set_even_probabilities): Cleanup; do not affect
22281 probabilities that are already known.
22282 (combine_predictions_for_bb): Call even when count is set.
22284 2017-07-18 Nathan Sidwell <nathan@acm.org>
22286 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
22289 2017-07-18 Bin Cheng <bin.cheng@arm.com>
22292 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
22293 optimization for loop niter analysis.
22295 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
22298 * config/avr/avr.c (avr_optimize_casesi): Don't use
22299 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
22301 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
22303 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
22304 body_cost_vec from _vect_peel_extended_info.
22305 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
22306 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
22309 2017-07-18 Bin Cheng <bin.cheng@arm.com>
22311 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
22313 2017-07-18 Richard Biener <rguenther@suse.de>
22315 PR tree-optimization/80620
22316 PR tree-optimization/81403
22317 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
22318 info when re-using a VN table entry.
22320 2017-07-18 Richard Biener <rguenther@suse.de>
22322 PR tree-optimization/81418
22323 * tree-vect-loop.c (vectorizable_reduction): Properly compute
22324 vectype_in. Verify that with lane-reducing reduction operations
22325 we have a single def-use cycle.
22327 2017-07-17 Carl Love <cel@us.ibm.com>
22329 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
22331 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22332 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
22333 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
22334 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
22335 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
22336 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
22337 VMULOSW): New enum "unspec" values.
22338 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
22339 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
22340 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
22341 altivec_vmulosw): New patterns.
22342 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
22343 VMULOSW): Add definitions.
22345 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
22347 * config/alpha/alpha.c: Include predict.h.
22349 2017-07-17 Yury Gribov <tetra2005@gmail.com>
22351 * tree-vrp.c (compare_assert_loc): Fix comparison function
22352 to return predictable results.
22354 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
22356 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
22358 (subdi3): Likewise.
22359 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
22360 * doc/invoke.texi (mexpand-adddi): Update text.
22362 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
22364 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
22365 that also clobbers the CC register. The old expand code is moved
22367 (*arc_clzsi2): ... here.
22368 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
22369 the CC register. The old expand code is moved to ...
22370 (arc_ctzsi2): ... here.
22372 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
22374 * config/arc/arc.opt (mindexed-loads): Use initial value
22375 TARGET_INDEXED_LOADS_DEFAULT.
22376 (mauto-modify-reg): Use initial value
22377 TARGET_AUTO_MODIFY_REG_DEFAULT.
22378 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
22379 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
22380 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
22381 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
22383 2017-07-17 Martin Liska <mliska@suse.cz>
22386 * opts.c (finish_options): Do not allow -fgnu-tm
22387 w/ -fsanitize={kernel-,}address. Say sorry.
22389 2017-07-17 Bin Cheng <bin.cheng@arm.com>
22392 * tree-loop-distribution.c (classify_partition): Only assert on
22393 numer of iterations.
22394 (merge_dep_scc_partitions): Delete prameter. Update function call.
22395 (distribute_loop): Remove code handling loop with unknown niters.
22396 (pass_loop_distribution::execute): Skip loop with unknown niters.
22398 2017-07-17 Bin Cheng <bin.cheng@arm.com>
22401 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
22402 function sort_partitions_by_post_order.
22404 2017-07-17 Bin Cheng <bin.cheng@arm.com>
22406 PR tree-optimization/81374
22407 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
22408 the max index of basic blocks, rather than number of basic blocks.
22410 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
22412 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
22414 (arc_legitimate_pic_operand_p): Likewise.
22415 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
22417 (arc_needs_pcl_p): Likewise.
22418 (arc_legitimate_pc_offset_p): Likewise.
22419 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
22420 function is also used in constrains.md.
22421 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
22422 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
22423 PLUS. Only return true/false in known cases, otherwise assert.
22424 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
22425 is already called in arc_legitimate_constant_p.
22426 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
22428 (LEGITIMATE_PIC_OPERAND_P): Use
22429 arc_raw_symbolic_reference_mentioned_p function.
22430 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
22435 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
22436 Andrew Burgess <andrew.burgess@embecosm.com>
22438 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
22439 (arc_return_address_register): New function.
22440 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
22441 (arc_handle_fndecl_attribute): Add naked attribute.
22442 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
22443 (TARGET_WARN_FUNC_RETURN): Likewise.
22444 (arc_allocate_stack_slots_for_args): New function.
22445 (arc_warn_func_return): Likewise.
22446 (machine_function): Change type fn_type.
22447 (arc_compute_function_type): Consider new naked function type,
22448 change function return type.
22449 (arc_must_save_register): Adapt to handle new
22450 arc_compute_function_type's return type.
22451 (arc_expand_prologue): Likewise.
22452 (arc_expand_epilogue): Likewise.
22453 (arc_return_address_regs): Delete.
22454 (arc_return_address_register): New function.
22455 (arc_epilogue_uses): Use above function.
22456 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
22457 (arc_function_type): Change encoding, add naked type.
22458 (ARC_INTERRUPT_P): Change to handle the new encoding.
22459 (ARC_FAST_INTERRUPT_P): Likewise.
22460 (ARC_NORMAL_P): Define.
22461 (ARC_NAKED_P): Likewise.
22462 (arc_compute_function_type): Delete prototype.
22463 * config/arc/arc.md (in_ret_delay_slot): Use
22464 arc_return_address_register function.
22465 (simple_return): Likewise.
22466 (p_return_i): Likewise.
22468 2017-07-17 Jakub Jelinek <jakub@redhat.com>
22470 PR tree-optimization/81428
22471 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
22472 can't be built for those types.
22474 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
22476 Remove stuff dead since r239246.
22478 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
22479 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
22480 (avr_inform_devices): Remove dead stuff.
22482 2017-07-17 Tamar Christina <tamar.christina@arm.com>
22484 * config/arm/arm_neon.h: Fix softp typo.
22486 2017-07-17 Jakub Jelinek <jakub@redhat.com>
22488 PR tree-optimization/81365
22489 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
22490 aggregate moves onto bb predecessor edges, make sure there are no
22491 loads that could alias the lhs in between the start of bb and the
22494 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
22497 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
22498 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
22499 [LSHIFTRT, outer_code = TRUNCATE]: Same.
22501 2017-07-17 Jakub Jelinek <jakub@redhat.com>
22503 PR tree-optimization/81396
22504 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
22505 (init_symbolic_number): Initialize it to 1.
22506 (perform_symbolic_merge): Add n_ops from both operands into the new
22508 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
22509 without base_addr as useless if they need more than one operation.
22510 (bswap_replace): Handle !bswap case for NULL base_addr.
22512 2017-07-17 Tom de Vries <tom@codesourcery.com>
22515 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
22518 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
22520 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
22521 conditional builtin define __FIX_LEON3FT_B2BST.
22523 2017-07-17 Daniel Cederman <cederman@gaisler.com>
22525 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
22526 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
22529 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
22531 PR rtl-optimization/81424
22532 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
22533 to remove potential trapping from operands if -fnon-call-exceptions.
22535 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
22537 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
22538 profile_proability for scalling.
22539 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
22541 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
22543 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
22545 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
22547 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
22548 fixpoint arithmetics.
22550 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
22552 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
22553 fixpoint arithmetics.
22555 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
22557 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
22558 fixpoint arithmetics.
22560 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
22562 * profile-count.h (profile_probability::from_reg_br_prob_note,
22563 profile_probability::to_reg_br_prob_note): New functions.
22564 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
22565 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
22566 * predict.c (probability_reliable_p): Update.
22567 (edge_probability_reliable_p): Update.
22568 (br_prob_note_reliable_p): Update.
22569 (invert_br_probabilities): Update.
22570 (add_reg_br_prob_note): New function.
22571 (combine_predictions_for_insn): Update.
22572 * asan.c (asan_clear_shadow): Update.
22573 * cfgbuild.c (compute_outgoing_frequencies): Update.
22574 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
22575 (update_br_prob_note): Update.
22576 (rtl_verify_edges): Update.
22577 (purge_dead_edges): Update.
22578 (fixup_reorder_chain): Update.
22579 * emit-rtl.c (try_split): Update.
22580 * ifcvt.c (cond_exec_process_insns): Update.
22581 (cond_exec_process_if_block): Update.
22582 (dead_or_predicable): Update.
22583 * internal-fn.c (expand_addsub_overflow): Update.
22584 (expand_neg_overflow): Update.
22585 (expand_mul_overflow): Update.
22586 * loop-doloop.c (doloop_modify): Update.
22587 * loop-unroll.c (compare_and_jump_seq): Update.
22588 * optabs.c (emit_cmp_and_jump_insn_1): Update.
22589 * predict.h: Update.
22590 * reorg.c (mostly_true_jump): Update.
22592 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
22593 * config/alpha/alpha.c (emit_unlikely_jump): Update.
22594 * config/arc/arc.c: (emit_unlikely_jump): Update.
22595 * config/arm/arm.c: (emit_unlikely_jump): Update.
22596 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
22597 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
22598 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
22599 (ix86_print_operand): Update.
22600 (ix86_split_fp_branch): Update.
22601 (predict_jump): Update.
22602 * config/ia64/ia64.c (ia64_print_operand): Update.
22603 * config/mmix/mmix.c (mmix_print_operand): Update.
22604 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
22605 (rs6000_expand_split_stack_prologue): Update.
22606 * config/rs6000/rs6000.c: Update.
22607 * config/s390/s390.c (s390_expand_vec_strlen): Update.
22608 (s390_expand_vec_movstr): Update.
22609 (s390_expand_cs_tdsi): Update.
22610 (s390_expand_split_stack_prologue): Update.
22611 * config/sh/sh.c (sh_print_operand): Update.
22612 (expand_cbranchsi4): Update.
22613 (expand_cbranchdi4): Update.
22614 * config/sparc/sparc.c (output_v9branch): Update.
22615 * config/spu/spu.c (get_branch_target): Update.
22616 (ea_load_store_inline): Update.
22617 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
22618 * config/tilepro/tilepro.c: Update.
22620 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
22622 * gimplify.c (mostly_copy_tree_r): Revert latest change.
22623 (gimplify_save_expr): Likewise.
22625 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
22627 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
22629 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
22631 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
22633 * timevar.def (TV_IPA_FNSUMMARY): Define.
22635 2017-07-16 Daniel Cederman <cederman@gaisler.com>
22637 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
22638 to back store errata sensitive sequence from being generated.
22639 (sqrtdf2_fix): Likewise.
22641 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
22643 * tree-ssa-threadupdate.c (compute_path_counts,
22644 update_joiner_offpath_counts): Use profile_probability.
22646 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
22649 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
22651 * config/arm/arm-c.c (arm_cpu_builtins): Define
22652 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
22654 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
22656 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22657 array entries to represent __ieee128 versions of the
22658 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
22659 scalar_extract_sig, and scalar_insert_exp built-in functions.
22660 (altivec_resolve_overloaded_builtin): Add special case handling
22661 for the __builtin_scalar_insert_exp function, as represented by
22662 the P9V_BUILTIN_VEC_VSIEDP constant.
22663 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
22664 exponent support for __ieee128 argument.
22665 (VSESQP): Add scalar extract signature support for __ieee128
22667 (VSTDCNQP): Add scalar test negative support for __ieee128
22669 (VSIEQP): Add scalar insert exponent support for __int128 argument
22670 with __ieee128 result.
22671 (VSIEQPF): Add scalar insert exponent support for __ieee128
22672 argument with __ieee128 result.
22673 (VSTDCQP): Add scalar test data class support for __ieee128
22675 (VSTDCNQP): Add overload support for scalar test negative with
22676 __ieee128 argument.
22677 (VSTDCQP): Add overload support for scalar test data class
22678 __ieee128 argument.
22679 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
22680 UNSPEC_VSX_SXSIGDP.
22681 (UNSPEC_VSX_SIEXPQP): New constant.
22682 (xsxexpqp): New insn for VSX scalar extract exponent quad
22684 (xsxsigqp): New insn for VSX scalar extract significand quad
22686 (xsiexpqpf): New insn for VSX scalar insert exponent quad
22687 precision with floating point argument.
22688 (xststdcqp): New expand for VSX scalar test data class quad
22690 (xststdcnegqp): New expand for VSX scalar test negative quad
22692 (xststdcqp): New insn to match expansions for VSX scalar test data
22693 class quad precision and VSX scalar test negative quad precision.
22694 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
22695 special case operand checking to enforce that second operand of
22696 VSX scalar test data class with quad precision argument is a 7-bit
22698 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
22699 prototypes and descriptions of __ieee128 versions of
22700 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
22701 scalar_test_data_class, and scalar_test_neg built-in functions.
22703 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22705 PR tree-optimization/81162
22706 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
22707 replace a negate with an add.
22709 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
22711 * doc/invoke.texi (arm/-mcpu): Document +crypto.
22713 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
22715 * config/arm/arm-c.c (arm_cpu_builtins): Define
22716 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
22718 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
22720 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
22721 (armv8-r): Set ARM Cortex-R52 as default CPU.
22722 * config/arm/arm-tables.opt: Regenerate.
22723 * config/arm/arm-tune.md: Regenerate.
22724 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
22726 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
22727 extension for -mcpu=cortex-r52.
22729 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
22731 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
22732 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
22733 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
22734 (fp-armv8): Define it as FP_ARMv8 only.
22735 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
22736 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
22738 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
22739 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
22740 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
22741 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
22742 than TARGET_FPU_ARMV8.
22743 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
22744 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
22745 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
22747 * config/arm/neon.md (neon_vrint): Likewise.
22748 (neon_vcvt): Likewise.
22749 (neon_<fmaxmin_op><mode>): Likewise.
22750 (<fmaxmin><mode>3): Likewise.
22751 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
22752 * config/arm/predicates.md (arm_cond_move_operator): Check against
22753 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
22755 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
22757 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
22758 to top of function.
22760 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22762 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
22763 loop in comment with memset.
22765 2017-07-14 Martin Liska <mliska@suse.cz>
22767 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
22768 * dwarf2out.c (is_java): Remove the function.
22769 (output_pubname): Remove usage of the function.
22770 (lower_bound_default): Remove usage of DW_LANG_Java.
22771 (gen_compile_unit_die): Likewise.
22772 * gcc.c: Remove compiler defaults for .java and .zip files.
22773 * gimple-expr.c (remove_suffix): Change as there's no longer
22774 extension than 4-letter one.
22775 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
22776 (gimplify_save_expr): Likewise.
22777 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
22778 as it's possible even for other languages than Java.
22779 * langhooks.h (struct lang_hooks): Remove Java from a comment.
22780 * lto-opts.c (lto_write_options): Remove reference to Java.
22781 * opts.c (strip_off_ending): Update file extension handling.
22782 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
22783 * tree-eh.c (lower_resx): Likewise.
22784 * tree.c (free_lang_data_in_type): Remove dead code.
22785 (find_decls_types_r): Likewise.
22786 (build_common_builtin_nodes): Remove Java from a comment.
22787 (verify_type): Remove dead code.
22788 * varasm.c (assemble_external): Remove Java from a comment.
22790 2017-07-14 Martin Liska <mliska@suse.cz>
22792 * opts.c (finish_options): Add quotes.
22793 (common_handle_option): Likewise.
22795 2017-07-14 Martin Liska <mliska@suse.cz>
22797 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
22798 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
22799 Remove N_SO_PASCAL.
22800 * dwarf2out.c (lower_bound_default): Do not handle
22802 (gen_compile_unit_die): Likewise.
22803 * gcc.c: Remove default extension binding for GNU Pascal.
22804 * stmt.c: Remove Pascal language from a comment.
22805 * xcoffout.c: Likewise.
22807 2017-07-13 David Malcolm <dmalcolm@redhat.com>
22810 * diagnostic-show-locus.c (fixit_cmp): New function.
22811 (layout::layout): Sort m_fixit_hints.
22812 (column_range::column_range): Assert that the values are valid.
22813 (struct char_span): New struct.
22814 (correction::overwrite): New method.
22815 (struct source_line): New struct.
22816 (line_corrections::add_hint): Add assertions. Reimplement memcpy
22817 calls in terms of classes source_line and char_span, and
22818 correction::overwrite.
22819 (selftest::test_overlapped_fixit_printing_2): New function.
22820 (selftest::diagnostic_show_locus_c_tests): Call it.
22822 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
22824 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
22825 early if there is no lhs.
22827 2017-07-13 Martin Liska <mliska@suse.cz>
22829 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
22830 (gen_reference_type_die): Likewise.
22831 * stor-layout.c: Remove Pascal-related comment.
22833 2017-07-13 Martin Liska <mliska@suse.cz>
22835 * opts.c (finish_options): Add quotes to error messages.
22836 (parse_sanitizer_options): Likewise.
22838 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
22840 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
22842 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
22844 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
22846 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
22848 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
22850 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
22852 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
22855 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
22856 provides the hardware capability bits, define the macro
22857 __BUILTIN_CPU_SUPPORTS__.
22858 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
22859 if GLIBC does not provide the hardware capability bits. Add a
22860 gcc_unreachable call if the built-in cpu function is neither
22861 __builtin_cpu_is nor __builtin_cpu_supports.
22862 (rs6000_get_function_versions_dispatcher): Change the warning
22863 that an old GLIBC is used which does not export the capability
22864 bits to be an error.
22865 * doc/extend.texi (target_clones attribute): Document the
22866 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
22867 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
22868 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
22869 the macros defined by GCC if the newer GLIBC is available.
22871 2017-07-12 Jeff Law <law@redhat.com>
22873 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
22874 remaining includes slightly.
22875 * config/riscv/riscv-builtins.c: Include profile-count.h.
22877 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
22880 * config/avr/avr.c (avr_set_current_function): In diagnostic
22881 messages: Quote keywords and (parts of) identifiers.
22882 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
22885 2017-07-12 Carl Love <cel@us.ibm.com>
22887 * config/rs6000/rs6000-c.c: Add support for built-in functions
22888 vector bool char vec_revb (vector bool char);
22889 vector bool short vec_revb (vector short char);
22890 vector bool int vec_revb (vector bool int);
22891 vector bool long long vec_revb (vector bool long long);
22892 * doc/extend.texi: Update the built-in documentation file for the
22893 new built-in functions.
22895 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22897 * config/s390/s390.md: Remove movcc splitter.
22899 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22901 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
22902 load/store on condition.
22904 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
22907 * config/avr/avr.c (avr_encode_section_info)
22908 [progmem && !TREE_READONLY]: Error if progmem object needs
22911 2017-07-11 Michael Collison <michael.collison@arm.com>
22913 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
22916 2017-07-11 Carl Love <cel@us.ibm.com>
22918 * config/rs6000/rs6000-c.c: Add support for builtins
22919 vector unsigned int vec_parity_lsbb (vector signed int);
22920 vector unsigned int vec_parity_lsbb (vector unsigned int);
22921 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
22922 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
22923 vector unsigned long long vec_parity_lsbb (vector signed long long);
22924 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
22925 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
22926 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
22927 * doc/extend.texi: Update the built-in documentation file for the
22928 new built-in functions.
22930 2017-07-11 David Malcolm <dmalcolm@redhat.com>
22932 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
22933 (layout::m_primary_loc): New field.
22934 (layout::layout): Initialize new field. Move location filtering
22935 logic from here to...
22936 (layout::maybe_add_location_range): ...this new method. Add
22937 support for filtering to just the lines already specified by other
22939 (layout::will_show_line_p): New method.
22940 (gcc_rich_location::add_location_if_nearby): New method.
22941 (selftest::test_add_location_if_nearby): New test function.
22942 (selftest::diagnostic_show_locus_c_tests): Call it.
22943 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
22946 2017-07-11 Tom de Vries <tom@codesourcery.com>
22948 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
22949 (bb_first_real_insn): New function.
22950 (nvptx_single): Add extra initialization of broadcasted condition
22953 2017-07-11 Nathan Sidwell <nathan@acm.org>
22955 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
22957 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
22959 * doc/extend.texi (AVR Function Attributes): Remove weblink to
22960 Binutils doc as TEXI will mess them up.
22961 * doc/invoke.texi (AVR Options): Same here.
22963 2017-07-11 Daniel Cederman <cederman@gaisler.com>
22965 * config/sparc/sparc.opt (mfix-ut700): New option.
22966 (mfix-gr712rc): Likewise.
22967 (sparc_fix_b2bst): New variable.
22968 * doc/invoke.texi (SPARC options): Document them.
22969 (ARM options): Fix warnings.
22970 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
22971 instructions to prevent sequences that can trigger the store-store
22972 errata for certain LEON3FT processors.
22973 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
22974 (sparc_option_override): Set sparc_fix_b2bst appropriately.
22975 * config/sparc/sparc.md (fix_b2bst): New attribute.
22976 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
22978 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
22981 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
22983 (*rsqrtsf2_sse): Ditto.
22985 (div<mode>3): Macroize insn from divdf3 and divsf3
22986 using MODEF mode iterator.
22988 2017-07-10 Martin Sebor <msebor@redhat.com>
22990 PR tree-optimization/80397
22991 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
22992 instead of testing for equality to INTEGER_TYPE.
22994 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
22996 * config.gcc: Remove uclibc from arc target spec.
22998 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
23000 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
23002 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23005 * lto-wrapper.c (remove_option): New function.
23006 (merge_and_complain): Merge PIC/PIE options more realistically.
23008 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
23010 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
23014 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
23015 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
23016 * config.in: Regenerate.
23017 * configure: Regenerate.
23018 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
23019 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
23020 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
23021 (TARGET_GASISR_PROLOGUES): ...target mask.
23022 * common/config/avr/avr-common.c
23023 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
23024 Set -mgas-isr-prologues.
23025 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
23026 INSERT_PASS_BEFORE for it.
23027 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
23028 * config/avr/avr.c (avr_option_override)
23029 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
23030 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
23031 (avr_attribute_table) <no_gccisr>: Add new function attribute.
23032 (avr_set_current_function) <is_no_gccisr>: Init machine field.
23033 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
23035 (make_avr_pass_pre_proep): New function.
23036 (emit_push_sfr) <treg>: Add argument to function and use it
23037 instead of TMP_REG.
23038 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
23039 and set machine->gasisr.yes.
23040 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
23041 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
23042 __gcc_isr.n_pushed to .L__stack_usage.
23043 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
23044 (avr_asm_final_postscan_insn): ...this new static function.
23045 * config/avr/avr.h (machine_function)
23046 <is_no_gccisr, use_L__stack_usage>: New fields.
23047 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
23048 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
23049 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
23050 (gasisr, *gasisr): New expander and insn.
23051 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
23052 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
23053 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
23055 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
23057 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
23060 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
23062 Move jump-tables out of .text again.
23065 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
23066 (ASM_OUTPUT_ADDR_VEC): New function.
23067 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
23068 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
23069 INSN_ADDRESSes as asm comment.
23070 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
23071 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
23072 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
23073 * config/avr/avr.md (*tablejump): Adjust comment.
23074 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
23075 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
23077 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
23078 (avr_output_addr_vec): New proto.
23079 (avr_log_t) <insn_addresses>: New field.
23081 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
23084 * config/i386/i386.c (ix86_function_arg_advance): Set
23085 outgoing_args_on_stack to true if there are outgoing arguments
23087 (ix86_function_arg): Likewise.
23088 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
23089 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
23090 * config/i386/i386.h (machine_function): Add
23091 outgoing_args_on_stack.
23093 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
23095 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
23096 supporting pthreds.
23097 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
23099 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
23101 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
23102 (REAL_H): Remove $(MACHMODE_H).
23103 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
23105 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
23106 $(MACHMODE_H) and double-int.h.
23107 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
23109 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
23112 2017-07-07 Andrew Pinski <apinski@cavium.com>
23114 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
23115 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
23117 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
23119 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
23120 Add warning if GCC was not configured to link against a GLIBC that
23121 exports the hardware capability bits.
23122 (make_resolver_func): Make resolver function private and not a
23123 COMDAT function. Create the name with clone_function_name instead
23124 of make_unique_name.
23127 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
23128 correct operand in doing the split.
23130 2017-07-07 Carl Love <cel@us.ibm.com>
23132 * config/rs6000/rs6000-c: Add support for built-in function
23133 vector unsigned short vec_pack_to_short_fp32 (vector float,
23135 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
23136 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
23137 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
23138 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
23139 (convert_4f32_8i16): Add define_expand.
23140 * doc/extend.texi: Update the built-in documentation file for the
23141 new built-in function.
23143 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23145 * config/sparc/m8.md: New file.
23146 * config/sparc/sparc.md: Include m8.md.
23148 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23150 * config/sparc/sparc.opt: New option -mvis4b.
23151 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
23152 (sparc_option_override): Handle VIS4B.
23153 (enum sparc_builtins): Define
23154 SPARC_BUILTIN_DICTUNPACK{8,16,32},
23155 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
23156 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
23157 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
23158 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
23159 (check_constant_argument): New function.
23160 (sparc_vis_init_builtins): Define builtins
23161 __builtin_vis_dictunpack{8,16,32},
23162 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
23163 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
23164 __builtin_vis_fpcmpde{8,16,32}shl and
23165 __builtin_vis_fpcmpur{8,16,32}shl.
23166 (sparc_expand_builtin): Check that the constant operands to
23167 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
23168 constant and in range.
23169 * config/sparc/sparc-c.c (sparc_target_macros): Handle
23171 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
23172 (SPARC_IMM5_P): Likewise.
23173 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
23174 (enabled): Handle vis4b.
23175 (UNSPEC_DICTUNPACK): New unspec.
23176 (UNSPEC_FPCMPSHL): Likewise.
23177 (UNSPEC_FPUCMPSHL): Likewise.
23178 (UNSPEC_FPCMPDESHL): Likewise.
23179 (UNSPEC_FPCMPURSHL): Likewise.
23180 (cpu_feature): New CPU feature `vis4b'.
23181 (dictunpack{8,16,32}): New insns.
23182 (FPCSMODE): New mode iterator.
23183 (fpcscond): New code iterator.
23184 (fpcsucond): Likewise.
23185 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
23186 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
23187 (fpcmpde{8,16,32}{si,di}shl): Likewise.
23188 (fpcmpur{8,16,32}{si,di}shl): Likewise.
23189 * config/sparc/constraints.md: Define constraints `q' for unsigned
23190 2-bit integer constants and `t' for unsigned 5-bit integer
23192 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
23194 (imm5_operand_dictunpack16): Likewise.
23195 (imm5_operand_dictunpack32): Likewise.
23196 (imm2_operand): Likewise.
23197 * doc/invoke.texi (SPARC Options): Document -mvis4b.
23198 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
23199 ditunpack* and fpcmp*shl builtins.
23201 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23203 * config.gcc: Handle m8 in --with-{cpu,tune} options.
23204 * config.in: Add HAVE_AS_SPARC6 define.
23205 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
23207 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
23209 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
23210 (CPP_CPU_SPEC): Handle m8.
23211 (ASM_CPU_SPEC): Likewise.
23212 * config/sparc/sparc-opts.h (enum processor_type): Add
23214 * config/sparc/sparc.c (m8_costs): New struct.
23215 (sparc_option_override): Handle TARGET_CPU_m8.
23216 (sparc32_initialize_trampoline): Likewise.
23217 (sparc64_initialize_trampoline): Likewise.
23218 (sparc_issue_rate): Likewise.
23219 (sparc_register_move_cost): Likewise.
23220 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
23221 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
23222 (ASM_CPU64_DEFAULT_SPEC): Likewise.
23223 (CPP_CPU_SPEC): Handle M8.
23224 (ASM_CPU_SPEC): Likewise.
23225 (AS_M8_FLAG): Define.
23226 * config/sparc/sparc.md: Add m8 to the cpu attribute.
23227 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
23228 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
23230 * configure: Regenerate.
23231 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
23234 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23236 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
23238 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
23239 ("*movdi_insn_sp32"): Do not set v3pipe.
23240 ("*movsi_insn"): Likewise.
23241 ("*movdi_insn_sp64"): Likewise.
23242 ("*movsf_insn"): Likewise.
23243 ("*movdf_insn_sp32"): Likewise.
23244 ("*movdf_insn_sp64"): Likewise.
23245 ("*zero_extendsidi2_insn_sp64"): Likewise.
23246 ("*sign_extendsidi2_insn"): Likewise.
23247 ("*mov<VM32:mode>_insn"): Likewise.
23248 ("*mov<VM64:mode>_insn_sp64"): Likewise.
23249 ("*mov<VM64:mode>_insn_sp32"): Likewise.
23250 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
23251 ("<vlop:code><VL:mode>3"): Likewise.
23252 ("*not_<vlop:code><VL:mode>3"): Likewise.
23253 ("*nand<VL:mode>_vis"): Likewise.
23254 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
23255 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
23256 ("one_cmpl<VL:mode>2"): Likewise.
23257 ("faligndata<VM64:mode>_vis"): Likewise.
23258 ("alignaddrsi_vis"): Likewise.
23259 ("alignaddrdi_vis"): Likweise.
23260 ("alignaddrlsi_vis"): Likewise.
23261 ("alignaddrldi_vis"): Likewise.
23262 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
23263 ("bmaskdi_vis"): Likewise.
23264 ("bmasksi_vis"): Likewise.
23265 ("bshuffle<VM64:mode>_vis"): Likewise.
23266 ("cmask8<P:mode>_vis"): Likewise.
23267 ("cmask16<P:mode>_vis"): Likewise.
23268 ("cmask32<P:mode>_vis"): Likewise.
23269 ("pdistn<P:mode>_vis"): Likewise.
23270 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
23272 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23274 * config/sparc/sparc.md ("subtype"): New insn attribute.
23275 ("*wrgsr_sp64"): Set insn subtype.
23276 ("*rdgsr_sp64"): Likewise.
23277 ("alignaddrsi_vis"): Likewise.
23278 ("alignaddrdi_vis"): Likewise.
23279 ("alignaddrlsi_vis"): Likewise.
23280 ("alignaddrldi_vis"): Likewise.
23281 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
23282 ("fexpand_vis"): Likewise.
23283 ("fpmerge_vis"): Likewise.
23284 ("faligndata<VM64:mode>_vis"): Likewise.
23285 ("bshuffle<VM64:mode>_vis"): Likewise.
23286 ("cmask8<P:mode>_vis"): Likewise.
23287 ("cmask16<P:mode>_vis"): Likewise.
23288 ("cmask32<P:mode>_vis"): Likewise.
23289 ("fchksm16_vis"): Likewise.
23290 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
23291 ("fmean16_vis"): Likewise.
23292 ("fp<plusminus_insn>64_vis"): Likewise.
23293 ("<plusminus_insn>v8qi3"): Likewise.
23294 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
23295 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
23296 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
23297 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
23298 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
23299 ("*movqi_insn"): Likewise.
23300 ("*movhi_insn"): Likewise.
23301 ("*movsi_insn"): Likewise.
23302 ("movsi_pic_gotdata_op"): Likewise.
23303 ("*movdi_insn_sp32"): Likewise.
23304 ("*movdi_insn_sp64"): Likewise.
23305 ("movdi_pic_gotdata_op"): Likewise.
23306 ("*movsf_insn"): Likewise.
23307 ("*movdf_insn_sp32"): Likewise.
23308 ("*movdf_insn_sp64"): Likewise.
23309 ("*zero_extendhisi2_insn"): Likewise.
23310 ("*zero_extendqihi2_insn"): Likewise.
23311 ("*zero_extendqisi2_insn"): Likewise.
23312 ("*zero_extendqidi2_insn"): Likewise.
23313 ("*zero_extendhidi2_insn"): Likewise.
23314 ("*zero_extendsidi2_insn_sp64"): Likewise.
23315 ("ldfsr"): Likewise.
23316 ("prefetch_64"): Likewise.
23317 ("prefetch_32"): Likewise.
23318 ("tie_ld32"): Likewise.
23319 ("tie_ld64"): Likewise.
23320 ("*tldo_ldub_sp32"): Likewise.
23321 ("*tldo_ldub1_sp32"): Likewise.
23322 ("*tldo_ldub2_sp32"): Likewise.
23323 ("*tldo_ldub_sp64"): Likewise.
23324 ("*tldo_ldub1_sp64"): Likewise.
23325 ("*tldo_ldub2_sp64"): Likewise.
23326 ("*tldo_ldub3_sp64"): Likewise.
23327 ("*tldo_lduh_sp32"): Likewise.
23328 ("*tldo_lduh1_sp32"): Likewise.
23329 ("*tldo_lduh_sp64"): Likewise.
23330 ("*tldo_lduh1_sp64"): Likewise.
23331 ("*tldo_lduh2_sp64"): Likewise.
23332 ("*tldo_lduw_sp32"): Likewise.
23333 ("*tldo_lduw_sp64"): Likewise.
23334 ("*tldo_lduw1_sp64"): Likewise.
23335 ("*tldo_ldx_sp64"): Likewise.
23336 ("*mov<VM32:mode>_insn"): Likewise.
23337 ("*mov<VM64:mode>_insn_sp64"): Likewise.
23338 ("*mov<VM64:mode>_insn_sp32"): Likewise.
23340 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23342 * config/sparc/sparc.md ("type"): New insn type viscmp.
23343 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
23345 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
23346 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
23347 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
23348 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
23350 ("n7_vis_logical_11cycle"): Likewise.
23351 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
23352 * config/sparc/niagara2.md ("niag3_vis": Likewise.
23353 * config/sparc/niagara.md ("niag_vis"): Likewise.
23354 * config/sparc/ultra3.md ("us3_fga"): Likewise.
23355 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
23357 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23359 * config/sparc/sparc.md: New instruction type `bmask'.
23360 (bmaskdi_vis): Use the `bmask' type.
23361 (bmasksi_vis): Likewise.
23362 * config/sparc/ultra3.md (us3_array): Likewise.
23363 * config/sparc/niagara7.md (n7_array): Likewise.
23364 * config/sparc/niagara4.md (n4_array): Likewise.
23365 * config/sparc/niagara2.md (niag2_vis): Likewise.
23366 (niag3_vis): Likewise.
23367 * config/sparc/niagara.md (niag_vis): Likewise.
23369 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
23371 * ipa-comdats.c: Remove optimize check from gate.
23372 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
23373 for functions not optimized.
23374 (ipa_fn_summary_read): Skip optimize check.
23375 (ipa_fn_summary_write): Likewise.
23376 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
23378 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
23380 (can_inline_edge_p): Check flag_pcc_struct_return for match.
23381 (check_callers): Give up on caller which is not optimized.
23382 (inline_small_functions): Likewise.
23383 (ipa_inline): Do not give up when not optimizing.
23384 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
23385 away unoptimizes cdtors.
23386 (whole_program_function_and_variable_visibility): Do
23387 ipa_discover_readonly_nonaddressable_vars in LTO mode.
23388 * ipa.c (process_references): Do not check optimize.
23389 (symbol_table::remove_unreachable_nodes): Update optimize check.
23390 (set_writeonly_bit): Update optimize check.
23391 (pass_ipa_cdtor_merge::gate): Do not check optimize.
23392 (pass_ipa_single_use::gate): Remove.
23394 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23396 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
23397 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
23398 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
23399 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
23400 permute_load, permute_store, adjust_extract, adjust_splat,
23401 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
23402 replace_swap_with_copy, dump_swap_insn_table,
23403 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
23404 recombine_lvx_pattern, recombine_stvx_pattern,
23405 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
23406 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
23407 to file rs6000-p8swap.c.
23408 * config/rs6000/rs6000-p8swap.c: New file.
23409 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
23410 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
23411 and rs6000*-*-* targets.
23413 2017-07-06 David Malcolm <dmalcolm@redhat.com>
23415 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
23417 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
23419 * lto-wrapper.c (merge_and_complain): Do not merge
23420 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
23421 fsigned_zeros, ftrapping_math, fwrapv.
23422 (append_compiler_options): Do not track these options.
23423 (append_linker_options): Likewie
23425 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
23427 * cgraphunit.c (cgraph_node::finalize_function): When
23428 !flag_toplevel_reorde set no_reorder flag.
23429 (varpool_node::finalize_decl): Likewise.
23430 (symbol_table::compile): Drop no toplevel reorder path.
23432 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
23434 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
23435 edges; zero probability is not better than uninitialized.
23437 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
23439 * asan.h (asan_sanitize_allocas_p): Declare.
23440 * asan.c (asan_sanitize_allocas_p): New function.
23441 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
23442 (handle_builtin_alloca): Likewise.
23443 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
23444 if !asan_sanitize_allocas_p.
23445 * params.def (asan-instrument-allocas): Add new option.
23446 * params.h (ASAN_PROTECT_ALLOCAS): Define.
23447 * opts.c (common_handle_option): Disable allocas sanitization for
23450 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
23452 * asan.c: Include gimple-fold.h.
23453 (get_last_alloca_addr): New function.
23454 (handle_builtin_stackrestore): Likewise.
23455 (handle_builtin_alloca): Likewise.
23456 (asan_emit_allocas_unpoison): Likewise.
23457 (get_mem_refs_of_builtin_call): Add new parameter, remove const
23458 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
23459 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
23460 (instrument_builtin_call): Pass gimple iterator to
23461 get_mem_refs_of_builtin_call.
23462 (last_alloca_addr): New global.
23463 * asan.h (asan_emit_allocas_unpoison): Declare.
23464 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
23465 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
23466 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
23467 if function calls alloca.
23468 * gimple-fold.c (replace_call_with_value): Remove static keyword.
23469 * gimple-fold.h (replace_call_with_value): Declare.
23470 * internal-fn.c: Include asan.h.
23471 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
23472 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
23474 2017-07-06 David Malcolm <dmalcolm@redhat.com>
23476 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
23477 (C_SELFTEST_FLAGS): New.
23478 (CPP_SELFTEST_FLAGS): New.
23479 (SELFTEST_DEPS): New, from deps of s-selftest.
23480 (C_SELFTEST_DEPS): New, from deps of s-selftest.
23481 (CPP_SELFTEST_DEPS): New.
23482 (selftest): Add dependency on s-selftest-c++.
23483 (s-selftest): Rename to...
23484 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
23485 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
23486 than SELFTEST_FLAGS.
23487 (selftest-gdb): Rename to...
23488 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
23490 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
23491 (selftest-valgrind): Rename to...
23492 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
23494 (selftest-valgrind): Reintroduce as an alias for
23495 selftest-c-valgrind.
23496 (s-selftest-c++): New.
23497 (selftest-c++-gdb): New.
23498 (selftest-c++-valgrind): New.
23500 2017-07-06 Olivier Hainque <hainque@adacore.com>
23502 * gcc.c (process_command): When deciding if undefined variables
23503 should be ignored when processing specs, accept "gcc -v" as well.
23505 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
23507 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
23508 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
23510 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
23512 * config/arm/arm-cpus.in (armv8-r): Add new entry.
23513 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
23514 * config/arm/arm-tables.opt: Regenerate.
23515 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
23517 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
23519 2017-07-06 Carl Love <cel@us.ibm.com>
23521 * ChangeLog: Clean up from mid air collision
23523 2017-07-06 Carl Love <cel@us.ibm.com>
23525 * config/rs6000/rs6000-c.c: Add support for built-in functions
23526 vector signed int vec_subc (vector signed int, vector signed int);
23527 vector signed __int128 vec_subc (vector signed __int128,
23528 vector signed __int128);
23529 vector unsigned __int128 vec_subc (vector unsigned __int128,
23530 vector unsigned __int128);
23531 vector signed int vec_sube (vector signed int, vector signed int,
23532 vector signed int);
23533 vector unsigned int vec_sube (vector unsigned int,
23534 vector unsigned int,
23535 vector unsigned int);
23536 vector signed __int128 vec_sube (vector signed __int128,
23537 vector signed __int128,
23538 vector signed__int128);
23539 vector unsigned __int128 vec_sube (vector unsigned __int128,
23540 vector unsigned __int128,
23541 vector unsigned __int128);
23542 vector signed int vec_subec (vector signed int, vector signed int,
23543 vector signed int);
23544 vector unsigned int vec_subec (vector unsigned int,
23545 vector unsigned int,
23546 vector unsigned int);
23547 vector signed __int128 vec_subec (vector signed __int128,
23548 vector signed __int128,
23549 vector signed__int128);
23550 vector unsigned __int128 vec_subec (vector unsigned __int128,
23551 vector unsigned __int128,
23552 vector unsigned __int128);
23553 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
23554 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
23555 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
23556 BU_ALTIVEC_OVERLOAD_X definitions.
23557 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
23558 * doc/extend.texi: Update the built-in documentation file for the new
23559 built-in functions.
23561 2017-07-06 David Malcolm <dmalcolm@redhat.com>
23564 * diagnostic-show-locus.c (layout::layout): Use start and finish
23565 spelling location for the start and finish of each range.
23566 * genmatch.c (linemap_client_expand_location_to_spelling_point):
23567 Add unused aspect param.
23568 * input.c (expand_location_1): Add "aspect" param, and use it
23569 to access the correct part of the location.
23570 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
23572 (expand_location_to_spelling_point): Likewise.
23573 (linemap_client_expand_location_to_spelling_point): Add "aspect"
23574 param, and pass it to expand_location_1.
23576 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
23578 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
23579 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
23580 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
23581 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
23582 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
23583 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
23584 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
23585 _mm_maskz_getmant_ss): New intrinsics.
23586 (__builtin_ia32_getexpss128_mask): Changed to ...
23587 __builtin_ia32_getexpss128_round ... this.
23588 (__builtin_ia32_getexpsd128_mask): Changed to ...
23589 __builtin_ia32_getexpsd128_round ... this.
23590 * config/i386/i386-builtin-types.def
23591 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
23592 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
23593 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
23594 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
23595 __builtin_ia32_getmantss_mask_round): New builtins.
23596 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
23597 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
23598 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
23599 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
23600 * config/i386/sse.md
23601 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
23602 avx512f_sgetexp<mode><mask_scalar_name>
23603 <round_saeonly_scalar_name> ... this.
23604 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
23605 %0, %1, %2<round_saeonly_op3>}): Changed to ...
23606 vgetexp<ssescalarmodesuffix>
23607 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
23608 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
23609 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
23610 avx512f_vgetmant<mode><mask_scalar_name>
23611 <round_saeonly_scalar_name> ... this.
23612 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
23613 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
23614 vgetmant<ssescalarmodesuffix>
23615 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
23616 %0<mask_scalar_operand4>, %1, %2
23617 <round_saeonly_scalar_mask_op4>, %3} ... this.
23618 * config/i386/subst.md (mask_scalar_operand4,
23619 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
23620 round_saeonly_scalar_nimm_predicate): New subst attributes.
23622 2017-07-06 Julia Koval <julia.koval@intel.com>
23624 * config/i386/i386.c (ix86_erase_embedded_rounding):
23625 Remove code for old rounding pattern.
23627 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
23629 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
23631 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
23633 * doc/sourcebuild.texi (Test Directives, Variants of
23634 dg-require-support): Add documentation for dg-require-stack-check.
23636 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
23638 * config/i386/subst.md (mask_scalar, round_scalar,
23639 round_saeonly_scalar): New meta-templates.
23640 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
23641 round_scalar_mask_operand3, round_scalar_mask_op3,
23642 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
23643 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
23644 round_saeonly_scalar_constraint,
23645 round_saeonly_scalar_prefix): New subst attribute.
23646 * config/i386/sse.md
23647 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
23648 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
23649 <round_scalar_name> ... this.
23650 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
23651 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
23652 <round_scalar_name> ... this.
23653 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
23654 <sse>_vm<code><mode>3<mask_scalar_name>
23655 <round_saeonly_scalar_name> ... this.
23656 (v<plusminus_mnemonic><ssescalarmodesuffix>
23657 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
23658 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
23659 v<plusminus_mnemonic><ssescalarmodesuffix>
23660 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
23661 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
23662 (v<multdiv_mnemonic><ssescalarmodesuffix>
23663 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
23664 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
23665 v<multdiv_mnemonic><ssescalarmodesuffix>
23666 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
23667 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
23668 (v<maxmin_float><ssescalarmodesuffix>
23669 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
23670 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
23671 v<maxmin_float><ssescalarmodesuffix>
23672 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
23673 %0<mask_scalar_operand3>, %1, %<iptr>2
23674 <round_saeonly_scalar_mask_op3>} ... this.
23676 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
23678 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
23679 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
23681 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
23682 Alan Hayward <alan.hayward@arm.com>
23683 David Sherwood <david.sherwood@arm.com>
23685 * combine.c (simplify_if_then_else): Remove "enum" before
23687 * compare-elim.c (can_eliminate_compare): Likewise.
23688 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
23690 (aarch64_lookup_simd_builtin_type): Likewise.
23691 (aarch64_simd_builtin_type): Likewise.
23692 (aarch64_init_simd_builtin_types): Likewise.
23693 (aarch64_simd_expand_args): Likewise.
23694 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
23696 (aarch64_reverse_mask): Likewise.
23697 (aarch64_simd_emit_reg_reg_move): Likewise.
23698 (aarch64_gen_adjusted_ldpstp): Likewise.
23699 (aarch64_ccmp_mode_to_code): Likewise.
23700 (aarch64_operands_ok_for_ldpstp): Likewise.
23701 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
23702 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
23704 (aarch64_min_divisions_for_recip_mul): Likewise.
23705 (aarch64_reassociation_width): Likewise.
23706 (aarch64_get_condition_code_1): Likewise.
23707 (aarch64_simd_emit_reg_reg_move): Likewise.
23708 (aarch64_simd_attr_length_rglist): Likewise.
23709 (aarch64_reverse_mask): Likewise.
23710 (aarch64_operands_ok_for_ldpstp): Likewise.
23711 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
23712 (aarch64_gen_adjusted_ldpstp): Likewise.
23713 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
23715 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
23716 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
23717 (arm_lookup_simd_builtin_type): Likewise.
23718 (arm_simd_builtin_type): Likewise.
23719 (arm_init_simd_builtin_types): Likewise.
23720 (arm_expand_builtin_args): Likewise.
23721 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
23722 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
23723 (ft32_setup_incoming_varargs): Likewise.
23724 (ft32_function_arg): Likewise.
23725 (ft32_function_arg_advance): Likewise.
23726 (ft32_pass_by_reference): Likewise.
23727 (ft32_arg_partial_bytes): Likewise.
23728 (ft32_valid_pointer_mode): Likewise.
23729 (ft32_addr_space_pointer_mode): Likewise.
23730 (ft32_addr_space_legitimate_address_p): Likewise.
23731 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
23733 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
23734 (ix86_emit_outlined_ms2sysv_restore): Likewise.
23735 (iamcu_alignment): Likewise.
23736 (canonicalize_vector_int_perm): Likewise.
23737 (ix86_noce_conversion_profitable_p): Likewise.
23738 (ix86_mpx_bound_mode): Likewise.
23739 (ix86_operands_ok_for_move_multiple): Likewise.
23740 * config/microblaze/microblaze-protos.h
23741 (microblaze_expand_conditional_branch_reg): Likewise.
23742 * config/microblaze/microblaze.c
23743 (microblaze_expand_conditional_branch_reg): Likewise.
23744 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
23746 (rs6000_reassociation_width): Likewise.
23747 (rs6000_invalid_binary_op): Likewise.
23748 (fusion_p9_p): Likewise.
23749 (emit_fusion_p9_load): Likewise.
23750 (emit_fusion_p9_store): Likewise.
23751 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
23753 (riscv_hard_regno_mode_ok_p): Likewise.
23754 (riscv_address_insns): Likewise.
23755 (riscv_split_symbol): Likewise.
23756 (riscv_legitimize_move): Likewise.
23757 (riscv_function_value): Likewise.
23758 (riscv_hard_regno_nregs): Likewise.
23759 (riscv_expand_builtin): Likewise.
23760 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
23761 (riscv_build_integer): Likewise.
23762 (riscv_split_integer): Likewise.
23763 (riscv_legitimate_constant_p): Likewise.
23764 (riscv_cannot_force_const_mem): Likewise.
23765 (riscv_regno_mode_ok_for_base_p): Likewise.
23766 (riscv_valid_base_register_p): Likewise.
23767 (riscv_valid_offset_p): Likewise.
23768 (riscv_valid_lo_sum_p): Likewise.
23769 (riscv_classify_address): Likewise.
23770 (riscv_legitimate_address_p): Likewise.
23771 (riscv_address_insns): Likewise.
23772 (riscv_load_store_insns): Likewise.
23773 (riscv_force_binary): Likewise.
23774 (riscv_split_symbol): Likewise.
23775 (riscv_force_address): Likewise.
23776 (riscv_legitimize_address): Likewise.
23777 (riscv_move_integer): Likewise.
23778 (riscv_legitimize_const_move): Likewise.
23779 (riscv_legitimize_move): Likewise.
23780 (riscv_address_cost): Likewise.
23781 (riscv_subword): Likewise.
23782 (riscv_output_move): Likewise.
23783 (riscv_canonicalize_int_order_test): Likewise.
23784 (riscv_emit_int_order_test): Likewise.
23785 (riscv_function_arg_boundary): Likewise.
23786 (riscv_pass_mode_in_fpr_p): Likewise.
23787 (riscv_pass_fpr_single): Likewise.
23788 (riscv_pass_fpr_pair): Likewise.
23789 (riscv_get_arg_info): Likewise.
23790 (riscv_function_arg): Likewise.
23791 (riscv_function_arg_advance): Likewise.
23792 (riscv_arg_partial_bytes): Likewise.
23793 (riscv_function_value): Likewise.
23794 (riscv_pass_by_reference): Likewise.
23795 (riscv_setup_incoming_varargs): Likewise.
23796 (riscv_print_operand): Likewise.
23797 (riscv_elf_select_rtx_section): Likewise.
23798 (riscv_save_restore_reg): Likewise.
23799 (riscv_for_each_saved_reg): Likewise.
23800 (riscv_register_move_cost): Likewise.
23801 (riscv_hard_regno_mode_ok_p): Likewise.
23802 (riscv_hard_regno_nregs): Likewise.
23803 (riscv_class_max_nregs): Likewise.
23804 (riscv_memory_move_cost): Likewise.
23805 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
23806 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
23807 (rl78_addr_space_address_mode): Likewise.
23808 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
23810 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
23811 (rs6000_reassociation_width): Likewise.
23812 (rs6000_invalid_binary_op): Likewise.
23813 (fusion_p9_p): Likewise.
23814 (emit_fusion_p9_load): Likewise.
23815 (emit_fusion_p9_store): Likewise.
23816 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
23817 (ok_for_simple_move_operands): Likewise.
23818 (ok_for_simple_move_strict_operands): Likewise.
23819 (ok_for_simple_arith_logic_operands): Likewise.
23820 (visium_legitimize_reload_address): Likewise.
23821 (visium_select_cc_mode): Likewise.
23822 (output_cbranch): Likewise.
23823 (visium_split_double_move): Likewise.
23824 (visium_expand_copysign): Likewise.
23825 (visium_expand_int_cstore): Likewise.
23826 (visium_expand_fp_cstore): Likewise.
23827 * config/visium/visium.c (visium_pass_by_reference): Likewise.
23828 (visium_function_arg): Likewise.
23829 (visium_function_arg_advance): Likewise.
23830 (visium_libcall_value): Likewise.
23831 (visium_setup_incoming_varargs): Likewise.
23832 (visium_legitimate_constant_p): Likewise.
23833 (visium_legitimate_address_p): Likewise.
23834 (visium_legitimize_address): Likewise.
23835 (visium_secondary_reload): Likewise.
23836 (visium_register_move_cost): Likewise.
23837 (visium_memory_move_cost): Likewise.
23838 (prepare_move_operands): Likewise.
23839 (ok_for_simple_move_operands): Likewise.
23840 (ok_for_simple_move_strict_operands): Likewise.
23841 (ok_for_simple_arith_logic_operands): Likewise.
23842 (visium_function_value_1): Likewise.
23843 (rtx_ok_for_offset_p): Likewise.
23844 (visium_legitimize_reload_address): Likewise.
23845 (visium_split_double_move): Likewise.
23846 (visium_expand_copysign): Likewise.
23847 (visium_expand_int_cstore): Likewise.
23848 (visium_expand_fp_cstore): Likewise.
23849 (visium_split_cstore): Likewise.
23850 (visium_select_cc_mode): Likewise.
23851 (visium_split_cbranch): Likewise.
23852 (output_cbranch): Likewise.
23853 (visium_print_operand_address): Likewise.
23854 * expmed.c (flip_storage_order): Likewise.
23855 * expmed.h (emit_cstore): Likewise.
23856 (flip_storage_order): Likewise.
23857 * genrecog.c (validate_pattern): Likewise.
23858 * hsa-gen.c (gen_hsa_addr): Likewise.
23859 * internal-fn.c (expand_arith_overflow): Likewise.
23860 * ira-color.c (allocno_copy_cost_saving): Likewise.
23861 * lra-assigns.c (find_hard_regno_for_1): Likewise.
23862 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
23863 (process_invariant_for_inheritance): Likewise.
23864 * lra-eliminations.c (move_plus_up): Likewise.
23865 * omp-low.c (lower_oacc_reductions): Likewise.
23866 * simplify-rtx.c (simplify_subreg): Likewise.
23867 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
23868 (TARGET_CHKP_BOUND_MODE): Likewise..
23869 * targhooks.c (default_chkp_bound_mode): Likewise.
23870 (default_setup_incoming_vararg_bounds): Likewise.
23871 * targhooks.h (default_chkp_bound_mode): Likewise.
23872 (default_setup_incoming_vararg_bounds): Likewise.
23873 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
23874 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
23875 (have_whole_vector_shift): Likewise.
23876 * tree-vect-stmts.c (vectorizable_load): Likewise.
23877 * doc/tm.texi: Regenerate.
23879 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
23881 Graceful degrade if Binutils PR21472 is not available.
23884 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
23885 .rodata in flash test fails.
23886 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
23887 * confgure: Regenerate.
23888 * config.in: Regenerate.
23889 * config/avr/avr.c (avr_asm_named_section)
23890 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
23891 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
23892 (avr_asm_init_sections): Same.
23894 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
23896 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
23897 (fma<VH:mode>4_intrinsic): Likewise.
23898 (*fmsub<VCVTF:mode>4): Likewise.
23899 (*fmsub<VH:mode>4_intrinsic): Likewise.
23901 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
23904 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
23905 Don't depend on "optimize > 0".
23906 (out_movhi_r_mr, out_movqi_mr_r): Same.
23907 (out_movhi_mr_r, out_movqi_r_mr): Same.
23908 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
23909 io_address_operand on "optimize > 0".
23911 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23913 * tree-loop-distribution.c: Add general explanantion on the pass.
23914 (generate_loops_for_partition): Mark distributed loop.
23915 (pg_add_dependence_edges): New parameter. Handle alias data
23916 dependence specially and record it in the parameter if asked.
23917 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
23918 (init_partition_graph_vertices, add_partition_graph_edge): New.
23919 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
23920 (free_partition_graph_vdata, build_partition_graph): New.
23921 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
23922 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
23923 (data_ref_segment_size, latch_dominated_by_data_ref): New.
23924 (compute_alias_check_pairs, version_loop_by_alias_check): New.
23925 (version_for_distribution_p, finalize_partitions): New.
23926 (distribute_loop): Handle alias data dependence specially. Factor
23927 out loop fusion code as functions and call these functions.
23929 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23931 * tree-loop-distribution.c (classify_partition): New parameter and
23932 better handle reduction statement.
23933 (rdg_build_partitions): Revise comment.
23934 (distribute_loop): Compute statements in all partitions and pass it
23935 to classify_partition.
23937 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23939 * tree-loop-distribution.c (enum partition_type): New.
23940 (struct partition): New field type.
23941 (partition_merge_into): Add parameter. Update partition type.
23942 (data_dep_in_cycle_p, update_type_for_merge): New functions.
23943 (build_rdg_partition_for_vertex): Compute partition type.
23944 (rdg_build_partitions): Dump partition type.
23945 (distribute_loop): Update calls to partition_merge_into.
23947 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23949 * tree-loop-distribution.c (struct ddr_hasher): New.
23950 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
23952 (classify_partition): Call get_data_dependence.
23953 (pg_add_dependence_edges): Ditto.
23954 (distribute_loop): Release data dependence hash table.
23956 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23958 * tree-loop-distribution.c (ref_base_address): Delete.
23959 (similar_memory_accesses): Rename ...
23960 (share_memory_accesses): ... to this. Check if partitions access
23961 the same memory reference.
23962 (distribute_loop): Call share_memory_accesses.
23964 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23966 * tree-loop-distribution.c (struct partition): New field recording
23967 its data reference.
23968 (partition_alloc, partition_free): Init and release data refs.
23969 (partition_merge_into): Merge data refs.
23970 (build_rdg_partition_for_vertex): Collect data refs for partition.
23971 (pg_add_dependence_edges): Change parameters from vector to bitmap.
23973 (distribute_loop): Remve data refs from vertice data of partition
23976 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23978 * tree-loop-distribution.c (params.h): Include header file.
23979 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
23980 (datarefs_vec): New global var.
23981 (create_rdg_vertices): Use datarefs_vec directly.
23982 (free_rdg): Don't free data references.
23983 (build_rdg): Update use. Don't free data references.
23984 (distribute_loop): Compute global variable for data references.
23985 Bail out if there are too many data references.
23987 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23989 * tree-loop-distribution.c (loop_nest): New global var.
23990 (build_rdg): Use loop directly, rather than loop nest.
23991 (pg_add_dependence_edges): Remove loop nest parameter. Use global
23993 (distribute_loop): Compute global variable loop nest. Update use.
23995 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23997 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
23998 (partition_merge_into): New parameter. Dump reason for fusion.
23999 (distribute_loop): Update use of partition_merge_into.
24001 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24003 * tree-loop-distribution.c (bb_top_order_index): New.
24004 (bb_top_order_index_size, bb_top_order_cmp): New.
24005 (stmts_from_loop): Use topological order.
24006 (pass_loop_distribution::execute): Compute and release topological
24007 order for basic blocks.
24009 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24011 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
24014 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24016 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
24017 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
24018 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
24019 * internal-fn.def (LOOP_DIST_ALIAS): New.
24020 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
24021 (fold_loop_internal_call): ... this.
24022 (vect_loop_dist_alias_call): New function.
24023 (set_uid_loop_bbs): Call fold_loop_internal_call.
24024 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
24027 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
24030 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
24031 Require dead FLAGS_REG at the beginning of a peephole.
24033 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
24036 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
24037 arguments in the call to __builtin_ia32_sbb_u32.
24038 (_subborrow_u64): Swap _X and _Y arguments in the call to
24039 __builtin_ia32_sbb_u64.
24041 2017-07-04 Jakub Jelinek <jakub@redhat.com>
24044 * tree-vrp.c (compare_assert_loc): Turn into a function template
24045 with stable template parameter. Only test if a->e is NULL,
24046 !a->e == !b->e has been verified already. Use e == NULL or
24047 e != NULL instead of e or ! e tests. If stable is true, don't use
24048 iterative_hash_expr, on the other side allow a or b or both NULL
24049 and sort the NULLs last.
24050 (process_assert_insertions): Sort using compare_assert_loc<false>
24051 instead of compare_assert_loc, later sort using
24052 compare_assert_loc<true> before calling process_assert_insertions_for
24053 in a loop. Use break instead of continue once seen NULL pointer.
24055 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
24057 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
24058 Cortex-R7 and Cortex-R8 processors.
24060 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
24062 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
24063 uninitialized while src is not.
24065 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
24067 * common/config/arm/arm-common.c: Adjust include path for
24069 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
24070 (arm-cpu.h): Create in build directory. Adjust dependency rules.
24071 (arm-cpu-data.h): Likewise.
24072 (arm-cpu-cdata.h): Likewise.
24073 * config/arm/arm-cpu.h: Delete.
24074 * config/arm/arm-cpu-cdata.h: Delete.
24075 * config/arm/arm-cpu-data.h: Delete.
24077 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
24079 * config/arm/arm-cpus.in (cortex-a55): New.
24080 (cortex-a75): Likewise.
24081 (cortex-a75.cortex-a55): Likewise.
24082 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
24084 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
24085 * config/arm/arm-cpu-cdata.h: Regenerate.
24086 * config/arm/arm-cpu-data.h: Regenerate.
24087 * config/arm/arm-cpu.h: Regenerate.
24088 * config/arm/arm-tables.opt: Regenerate.
24089 * config/arm/arm-tune.md: Regenerate.
24091 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
24093 * haifa-sched.c (sched_create_recovery_edges): Update profile.
24095 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
24097 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
24100 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
24102 PR tree-optimization/81292
24103 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
24104 full_string_p, also call adjust_related_strinfos if the adjustment
24105 is simple, otherwise invalidate related strinfos.
24107 2017-07-04 Martin Liska <mliska@suse.cz>
24110 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
24111 newly created variable as DECL_IGNORED_P.
24113 2017-07-04 Martin Liska <mliska@suse.cz>
24116 * ipa-inline.c (inline_small_functions):
24117 Use xstrdup_for_dump.
24119 2017-07-04 Tom de Vries <tom@codesourcery.com>
24121 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
24123 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
24126 * config/darwin.c (darwin_function_switched_text_sections):
24129 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
24131 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
24133 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
24135 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
24137 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24139 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
24140 min_profitable_iters, and th as inclusive lower bounds.
24141 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
24142 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
24143 for min_profitable_iters and min_profitable_estimate.
24144 (vect_transform_loop): Treat th as an inclusive lower bound.
24145 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
24147 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
24150 * config/darwin.c (darwin_function_switched_text_sections):
24151 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
24152 in two pieces, and suppress the use of buf.
24154 2017-07-03 Nathan Sidwell <nathan@acm.org>
24156 * hash-table.h (hash_table_mod1): Fix indentation.
24158 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24160 PR middle-end/81290
24161 * predict.c (force_edge_cold): Be more careful about propagation
24163 * profile-count.h (profile_probability::guessed,
24164 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
24166 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
24168 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
24170 * doc/invoke.texi (rcpc architecture extension): Document it.
24172 2017-07-03 Richard Biener <rguenther@suse.de>
24174 PR tree-optimization/60510
24175 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
24176 the scalar reduction PHI and use it.
24177 (vectorizable_reduction): Properly guard the single_defuse_cycle
24178 path for non-SLP reduction chains where we cannot use it.
24179 Rework reduc_def/index and vector type deduction. Rework
24180 vector operand gathering during reduction op code-gen.
24181 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
24182 chains dissolve the chain and leave it to non-SLP reduction
24185 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24187 * tree-data-ref.h (dr_alignment): Declare.
24188 * tree-data-ref.c (dr_alignment): New function.
24189 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
24190 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
24192 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
24194 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24196 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
24197 and base_misalignment fields.
24198 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
24199 * tree-data-ref.c: Include builtins.h.
24200 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
24201 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
24202 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
24203 * tree-vect-data-refs.c: Include tree-cfg.h.
24204 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
24205 fields instead of calculating an alignment here.
24206 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
24207 innermost_loop_behavior fields.
24209 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24211 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
24213 (DR_STEP_ALIGNMENT): New macro.
24214 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
24215 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
24216 (create_data_ref): Print it.
24217 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
24218 to tell whether the step preserves vector (mis)alignment.
24219 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
24220 Move the check for an integer step and generalise to all INTEGER_CST.
24221 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
24222 Print the outer step alignment.
24224 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24226 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
24227 with offset_alignment.
24228 (DR_ALIGNED_TO): Delete.
24229 (DR_OFFSET_ALIGNMENT): New macro.
24230 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
24231 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
24232 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
24233 (create_data_ref): Likewise.
24234 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
24235 (vect_analyze_data_refs): Likewise.
24236 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
24237 creating dummy innermost behavior.
24239 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24241 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
24242 with a "innermost_loop_behavior *" and refeence tree.
24243 * tree-data-ref.c (dr_analyze_innermost): Likewise.
24244 (create_data_ref): Update call accordingly.
24245 * tree-predcom.c (find_looparound_phi): Likewise.
24247 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24249 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
24250 fields with dr_wrt_vec_loop.
24251 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
24252 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
24253 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
24254 (vect_dr_behavior): New function.
24255 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
24256 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
24257 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
24258 track whether the step preserves the misalignment.
24259 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
24260 Use vect_dr_behavior.
24261 (vect_setup_realignment): Update call accordingly.
24262 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
24263 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
24264 call to vect_create_addr_base_for_vector_ref.
24265 (vect_create_cond_for_align_checks): Likewise.
24266 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
24267 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
24268 (vect_recog_mask_conversion_pattern): Likewise.
24269 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
24270 (new_stmt_vec_info): Remove redundant zeroing.
24272 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
24274 * common/config/arm/arm-common.c (arm_be8_option): New function.
24275 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
24276 (ISA_ARMv6): Add isa_bit_be8.
24277 * config/arm/arm.h (arm_be8_option): Add prototype.
24278 (BE8_SPEC_FUNCTION): New define.
24279 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
24280 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
24281 (mlittle-endian): Similarly.
24282 (mbe8, mbe32): New options.
24283 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
24284 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
24286 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24288 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
24290 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24292 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
24293 (cleanup_tree_cfg_bb): Use it.
24294 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
24296 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
24298 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24301 * loop-doloop.c (add_test): Update profile.
24303 2017-07-03 Martin Liska <mliska@suse.cz>
24306 * sanopt.c (rewrite_usage_of_param): New function.
24307 (sanitize_rewrite_addressable_params): Likewise.
24308 (pass_sanopt::execute): Call rewrite_usage_of_param.
24310 2017-07-03 Richard Biener <rguenther@suse.de>
24312 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
24313 back to using VIEW_CONVERT_EXPR.
24315 2017-07-03 Martin Liska <mliska@suse.cz>
24318 * doc/extend.texi: Document when a resolver function is
24319 generated for target_clones.
24321 2017-07-03 Martin Liska <mliska@suse.cz>
24323 * asan.c (asan_emit_stack_protection): Unpoison just red zones
24324 and shadow memory of auto variables which are subject of
24325 use-after-scope sanitization.
24326 (asan_expand_mark_ifn): Add do set only when is_poison.
24328 2016-07-03 Richard Biener <rguenther@suse.de>
24330 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
24332 (vect_force_simple_reduction): Record reduction def -> phi mapping.
24333 (vectorizable_reduction): Perform reduction PHI creation when
24334 visiting a reduction PHI and adjust and simplify code generation
24335 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
24336 (vect_transform_loop): Visit reduction PHIs.
24337 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
24338 defs into the SLP tree.
24339 (vect_build_slp_tree): Reduction defs terminate the recursion.
24340 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
24342 (vect_get_vec_defs_for_stmt_copy): Export.
24343 (vect_get_vec_defs): Likewise.
24344 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
24346 (vect_get_vec_defs_for_stmt_copy): Declare.
24347 (vect_get_vec_defs): Likewise.
24349 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24351 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
24352 parameter with a "loop" parameter and use it instead of the
24353 loop containing DR_STMT. Don't check simple_iv when doing
24354 BB analysis. Describe the two analysis modes in the comment.
24356 2017-07-03 Tom de Vries <tom@codesourcery.com>
24358 PR tree-optimization/69468
24359 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
24360 (find_same_succ_bb): Handle ignore_edge_flags.
24362 2017-07-03 Tom de Vries <tom@codesourcery.com>
24364 PR tree-optimization/81192
24365 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
24367 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
24369 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
24371 2017-07-03 Tom de Vries <tom@codesourcery.com>
24373 PR tree-optimization/81192
24374 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
24375 BB_SAME_SUCC (bb) == NULL.
24377 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24379 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
24382 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24384 * dumpfile.c: Include profile-count.h
24385 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
24387 (insert_cond_bb): Update profile.
24388 * tree-cfg.h (insert_cond_bb): Update prototype.
24389 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
24390 * tree-dump.c: Do not include tree-cfg.
24392 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24394 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
24396 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24398 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
24401 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24403 * tree-complex.c (expand_complex_div_wide): update profile.
24405 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
24406 Alan Hayward <alan.hayward@arm.com>
24407 David Sherwood <david.sherwood@arm.com>
24409 * Makefile.in (MACHMODE_H): Remove insn-modes.h
24410 (CORETYPES_H): New define.
24411 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
24412 (insn-modes-inline.h, s-modes-inline-h): New rules.
24413 (generated_files): Add insn-modes-inline.h.
24414 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
24415 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
24416 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
24417 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
24418 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
24419 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
24420 (build/gencodes.o, build/genconditions.o): Likewise.
24421 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
24422 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
24423 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
24424 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
24425 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
24426 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
24427 * coretypes.h: Include everything up to real.h for generators.
24428 Include insn-modes.h first. Include wide-int-print.h after
24429 wide-int.h. Include insn-modes-inline.h and then machmode.h.
24430 * machmode.h: Don't include insn-modes.h here.
24431 * function-tests.c: Remove includes of signop.h, machmode.h,
24432 double-int.h and wide-int.h.
24434 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
24436 * optc-save-gen.awk: Likewise.
24437 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
24438 * godump.c: Remove include of wide-int-print.h.
24439 * pretty-print.h: Likewise.
24440 * wide-int-print.cc: Likewise.
24441 * wide-int.cc: Likewise.
24442 * hash-map-tests.c: Remove include of signop.h.
24443 * hash-set-tests.c: Likewise.
24444 * rtl-tests.c: Likewise.
24445 * mkconfig.sh: Remove include of machmode.h.
24446 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
24448 (emit_insn_modes_inline_h): ...this new function. Emit the code
24449 into an insn-modes-inline.h header file, adding appropriate
24450 include guards and end comments.
24451 (emit_insn_modes_c_header): Remove include of machmode.h.
24452 (emit_min_insn_modes_c_header): Include coretypes.h rather than
24454 (main): Handle -i flag and call emit_insn_modes_inline_h when
24457 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
24459 * tree-ssa-strlen.c (strinfo): Rename the length field to
24460 nonzero_chars. Add a full_string_p field.
24461 (compare_nonzero_chars, zero_length_string_p): New functions.
24462 (get_addr_stridx): Add an offset_out parameter.
24463 Use compare_nonzero_chars.
24464 (get_stridx): Update accordingly. Use compare_nonzero_chars.
24465 (new_strinfo): Update after above changes to strinfo.
24466 (set_endptr_and_length): Set full_string_p.
24467 (get_string_length): Update after above changes to strinfo.
24468 (unshare_strinfo): Update call to new_strinfo.
24469 (maybe_invalidate): Likewise.
24470 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
24471 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
24472 as a uhwi instead of an shwi. Update after above changes to
24473 strinfo and new_strinfo.
24474 (zero_length_string): Assert that chainsi contains full strings.
24475 Use zero_length_string_p. Update call to new_strinfo.
24476 (adjust_related_strinfos): Update after above changes to strinfo.
24477 Copy full_string_p from origsi.
24478 (adjust_last_stmt): Use zero_length_string_p.
24479 (handle_builtin_strlen): Update after above changes to strinfo and
24480 new_strinfo. Install the lhs as the string length if the previous
24481 entry didn't describe a full string.
24482 (handle_builtin_strchr): Update after above changes to strinfo
24484 (handle_builtin_strcpy): Likewise.
24485 (handle_builtin_strcat): Likewise.
24486 (handle_builtin_malloc): Likewise.
24487 (handle_pointer_plus): Likewise.
24488 (handle_builtin_memcpy): Likewise. Track nonzero characters
24489 that aren't necessarily followed by a nul terminator.
24490 (handle_char_store): Likewise.
24492 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
24494 PR tree-optimization/80769
24495 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
24496 for malloc and calloc. Document the new invariant that all related
24497 strinfos have delayed lengths or none do.
24498 (verify_related_strinfos): Move earlier in file.
24499 (set_endptr_and_length): New function, split out from...
24500 (get_string_length): ...here. Also set the lengths of related
24502 (zero_length_string): Assert that chainsi has known (rather than
24504 (adjust_related_strinfos): Likewise.
24506 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
24508 PR tree-optimization/81136
24509 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
24510 assert that two references with the same misalignment have the same
24511 compile-time misalignment if those compile-time misalignments
24514 2017-07-01 Andi Kleen <ak@linux.intel.com>
24516 * print-tree.c (print_node): Print all attributes.
24518 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
24520 * cfg.c (scale_bbs_frequencies): New function.
24521 * cfg.h (scale_bbs_frequencies): Declare it.
24522 * cfgloopanal.c (single_likely_exit): Cleanup.
24523 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
24525 (scale_loop_profile): Likewise.
24526 (loop_version): Likewise.
24527 (create_empty_loop_on_edge): Update.
24528 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
24529 scale_loop_frequencies, scale_loop_profile, loopify,
24530 loop_version): Update prototypes.
24531 * modulo-sched.c (sms_schedule): Update.
24532 * predict.c (unlikely_executed_edge_p): Also check probability.
24533 (probably_never_executed_edge_p): Fix typo.
24534 * tree-if-conv.c (version_loop_for_if_conversion): Update.
24535 * tree-parloops.c (gen_parallel_loop): Update.
24536 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
24537 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
24538 * tree-ssa-loop-split.c (split_loop): Update.
24539 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
24540 * tree-vect-loop-manip.c (vect_do_peeling): Update.
24541 (vect_loop_versioning): Update.
24542 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
24544 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
24546 * trans-mem.c (split_bb_make_tm_edge): Update profile.
24548 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
24550 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
24551 to keep profile consistent.
24553 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
24555 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
24556 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
24557 * profile-count.h (max_safe_multiplier): Make unsigned.
24558 (profile_count::guessed_zero): New.
24560 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
24562 * bb-reorder.c (fix_up_crossing_landing_pad,
24563 fix_crossing_conditional_branches): Use make_single_succ_edge
24564 to keep profile consistent.
24566 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
24568 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
24571 2017-07-01 Jakub Jelinek <jakub@redhat.com>
24574 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
24575 the right scopes, make sure cond_jump isn't preserved between multiple
24576 iterations. Search for fallthru edge whenever there are 3+ edges and
24577 use find_fallthru_edge for it.
24579 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
24581 Patch by Alexander Monakov <amonakov@ispras.ru>
24582 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
24583 probabilities consistently.
24585 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
24587 * pa.c (pa_expand_compare_and_swap_loop): Update call of
24588 emit_cmp_and_jump_insns.
24590 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
24593 * tree-inline.c (expand_call_inline): Combine profile statuses.
24595 2017-06-30 Andrew Pinski <apinski@cavium.com>
24597 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
24598 fold_stmt returned true.
24600 2017-06-30 Nathan Sidwell <nathan@acm.org>
24602 * ggc.h (empty_string): Delete.
24603 * cfgexpand.c (expand_asm_stmt): Use plain "".
24604 * optabs.c (expand_asm_memory_barrier): Likewise.
24605 * stringpool.c (empty_string): Delete.
24606 (digit_vector, digit_string): Delete.
24607 (ggc_alloc_string): Use plain "", don't optimize single digit
24608 strings. Use ggc_alloc_atomic.
24610 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
24612 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
24613 comparison set and one other set, use the cost of the non-comparison
24616 2017-06-30 Nathan Sidwell <nathan@acm.org>
24618 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
24621 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
24623 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
24624 loops. Remove now unneeded calls to gimple_switch_set_label() that
24625 just set removed labels to NULL_TREE.
24627 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
24629 * tree-ssanames.c (set_range_info_raw): Abstract from ...
24630 (set_range_info): ...here. Only call set_range_info_raw if domain
24632 (set_nonzero_bits): Call set_range_info_raw.
24633 * tree-ssanames.h (set_range_info_raw): New.
24635 2017-06-30 Jakub Jelinek <jakub@redhat.com>
24638 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
24639 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
24640 of nonimmediate_operand and <store_mask_constraint> instead of m
24641 for the input operand. For V8FI iterator, always split if input
24642 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
24643 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
24644 <store_mask_predicate> instead of register_operand and
24645 <store_mask_constraint> instead of v for the input operand. Make
24646 sure both operands aren't MEMs for if not <mask_applied>.
24648 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
24650 * lto-wrapper.c (copy_file) Close both file descriptors before
24653 2017-06-30 Martin Liska <mliska@suse.cz>
24656 * multiple_target.c (create_dispatcher_calls): Make ifunc
24657 also for function that don't have calls or are not referenced.
24659 2017-06-30 Richard Biener <rguenther@suse.de>
24661 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
24662 analyze the first scalar stmt. Move vector type computation
24663 for the BB case here from ...
24664 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
24665 live operation processing in the SLP case properly.
24667 2017-06-30 Richard Biener <rguenther@suse.de>
24669 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
24671 2017-06-30 Martin Liska <mliska@suse.cz>
24674 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
24675 before BUILT_IN_UNWIND_RESUME when ASAN is used.
24677 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
24679 * doc/invoke.texi (AArch64): Add missing options and remove redundant
24682 2017-06-30 Richard Biener <rguenther@suse.de>
24684 PR tree-optimization/81249
24685 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
24686 condition reduction result to original scalar type.
24688 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24690 * profile-count.h (enum profile_quality): Fix typos and whitespace
24693 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24695 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
24696 type for branch probabilities.
24698 2017-06-29 Julian Brown <julian@codesourcery.com>
24699 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24701 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
24702 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
24703 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
24704 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
24706 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24708 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
24709 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
24710 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
24711 CC usage from generic code to here.
24712 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
24713 CC usage into the target macros.
24715 2017-06-29 Maya Rashish <coypu@sdf.org>
24717 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
24720 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
24722 * arm/arm-builtins.c: Include profile-count.h
24723 * except.c (sjlj_emit_function_enter): Use
24724 profile_probability::unlikely.
24726 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24728 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
24729 and tocrel_offset be pointer args rather than implicitly using
24731 (legitimate_constant_pool_address_p, rs6000_emit_move,
24732 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
24733 tocrel_offset and use in toc_relative_expr_p call.
24734 (print_operand, print_operand_address): Use static tocrel_base_oac
24735 and tocrel_offset_oac.
24736 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
24739 2017-06-29 Maya Rashish <coypu@sdf.org>
24741 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
24743 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
24745 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
24746 objects, take into account only the alignment of 'op0' and 'mode1' if
24749 2017-06-29 Steve Ellcey <sellcey@cavium.com>
24751 * ccmp.c (ccmp_tree_comparison_p): New function.
24752 (ccmp_candidate_p): Update to use above function.
24753 (get_compare_parts): New function.
24754 (expand_ccmp_next): Update to use new functions.
24755 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
24757 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
24758 take mode as argument.
24759 * ccmp.h (expand_ccmp_expr): Add mode as argument.
24760 * expr.c (expand_expr_real_1): Pass mode as argument.
24762 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
24764 * combine.c (combine_instructions): Print insns to dump_file, together
24767 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
24769 * asan.c (asan_emit_stack_protection): Update.
24770 (create_cond_insert_point): Update.
24771 * auto-profile.c (afdo_propagate_circuit): Update.
24772 * basic-block.h (struct edge_def): Turn probability to
24773 profile_probability.
24774 (EDGE_FREQUENCY): Update.
24775 * bb-reorder.c (find_traces_1_round): Update.
24776 (better_edge_p): Update.
24777 (sanitize_hot_paths): Update.
24778 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
24779 (make_single_succ_edge): Update.
24780 (check_bb_profile): Update.
24781 (dump_edge_info): Update.
24782 (update_bb_profile_for_threading): Update.
24783 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
24784 probabilitycount to 0.
24785 * cfgbuild.c (compute_outgoing_frequencies): Update.
24786 * cfgcleanup.c (try_forward_edges): Update.
24787 (outgoing_edges_match): Update.
24788 (try_crossjump_to_edge): Update.
24789 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
24790 (expand_gimple_tailcall): Update.
24791 (construct_init_block): Use make_single_succ_edge.
24792 (construct_exit_block): Use make_single_succ_edge.
24793 * cfghooks.c (verify_flow_info): Update.
24794 (redirect_edge_succ_nodup): Update.
24795 (split_edge): Update.
24796 (account_profile_record): Update.
24797 * cfgloopanal.c (single_likely_exit): Update.
24798 * cfgloopmanip.c (scale_loop_profile): Update.
24799 (set_zero_probability): Remove.
24800 (duplicate_loop_to_header_edge): Update.
24801 * cfgloopmanip.h (loop_version): Update prototype.
24802 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
24803 (force_nonfallthru_and_redirect): Update.
24804 (update_br_prob_note): Update.
24805 (rtl_verify_edges): Update.
24806 (purge_dead_edges): Update.
24807 (rtl_lv_add_condition_to_bb): Update.
24808 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
24809 * cgraphunit.c (init_lowered_empty_function): Update.
24810 (cgraph_node::expand_thunk): Update.
24811 * cilk-common.c: Include profile-count.h
24812 * dojump.c (inv): Remove.
24813 (jumpifnot): Update.
24814 (jumpifnot_1): Update.
24815 (do_jump_1): Update.
24817 (do_jump_by_parts_greater_rtx): Update.
24818 (do_compare_rtx_and_jump): Update.
24819 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
24820 do_jump_1. do_compare_rtx_and_jump): Update prototype.
24821 * dwarf2cfi.c: Include profile-count.h
24822 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
24823 (sjlj_emit_dispatch_table): Likewise.
24824 * explow.c: Include profile-count.h
24825 * expmed.c (emit_store_flag_force): Update.
24826 (do_cmp_and_jump): Update.
24827 * expr.c (compare_by_pieces_d::generate): Update.
24828 (compare_by_pieces_d::finish_mode): Update.
24829 (emit_block_move_via_loop): Update.
24830 (store_expr_with_bounds): Update.
24831 (store_constructor): Update.
24832 (expand_expr_real_2): Update.
24833 (expand_expr_real_1): Update.
24834 * expr.h (try_casesi, try_tablejump): Update prototypes.
24835 * gimple-pretty-print.c (dump_probability): Update.
24836 (dump_profile): New.
24837 (dump_gimple_label): Update.
24838 (dump_gimple_bb_header): Update.
24839 * graph.c (draw_cfg_node_succ_edges): Update.
24840 * hsa-gen.c (convert_switch_statements): Update.
24841 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
24842 (find_if_case_1): Update.
24843 (find_if_case_2): Update.
24844 * internal-fn.c (expand_arith_overflow_result_store): Update.
24845 (expand_addsub_overflow): Update.
24846 (expand_neg_overflow): Update.
24847 (expand_mul_overflow): Update.
24848 (expand_vector_ubsan_overflow): Update.
24849 * ipa-cp.c (good_cloning_opportunity_p): Update.
24850 * ipa-split.c (split_function): Use make_single_succ_edge.
24851 * ipa-utils.c (ipa_merge_profiles): Update.
24852 * loop-doloop.c (add_test): Update.
24853 (doloop_modify): Update.
24854 * loop-unroll.c (compare_and_jump_seq): Update.
24855 (unroll_loop_runtime_iterations): Update.
24856 * lra-constraints.c (lra_inheritance): Update.
24857 * lto-streamer-in.c (input_cfg): Update.
24858 * lto-streamer-out.c (output_cfg): Update.
24859 * mcf.c (adjust_cfg_counts): Update.
24860 * modulo-sched.c (sms_schedule): Update.
24861 * omp-expand.c (expand_omp_for_init_counts): Update.
24862 (extract_omp_for_update_vars): Update.
24863 (expand_omp_ordered_sink): Update.
24864 (expand_omp_for_ordered_loops): Update.
24865 (expand_omp_for_generic): Update.
24866 (expand_omp_for_static_nochunk): Update.
24867 (expand_omp_for_static_chunk): Update.
24868 (expand_cilk_for): Update.
24869 (expand_omp_simd): Update.
24870 (expand_omp_taskloop_for_outer): Update.
24871 (expand_omp_taskloop_for_inner): Update.
24872 * omp-simd-clone.c (simd_clone_adjust): Update.
24873 * optabs.c (expand_doubleword_shift): Update.
24874 (expand_abs): Update.
24875 (emit_cmp_and_jump_insn_1): Update.
24876 (expand_compare_and_swap_loop): Update.
24877 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
24878 * predict.c (predictable_edge_p): Update.
24879 (edge_probability_reliable_p): Update.
24880 (set_even_probabilities): Update.
24881 (combine_predictions_for_insn): Update.
24882 (combine_predictions_for_bb): Update.
24883 (propagate_freq): Update.
24884 (estimate_bb_frequencies): Update.
24885 (force_edge_cold): Update.
24886 * profile-count.c (profile_count::dump): Add missing space into dump.
24887 (profile_count::debug): Add newline.
24888 (profile_count::differs_from_p): Explicitly convert to unsigned.
24889 (profile_count::stream_in): Update.
24890 (profile_probability::dump): New member function.
24891 (profile_probability::debug): New member function.
24892 (profile_probability::differs_from_p): New member function.
24893 (profile_probability::differs_lot_from_p): New member function.
24894 (profile_probability::stream_in): New member function.
24895 (profile_probability::stream_out): New member function.
24896 * profile-count.h (profile_count_quality): Rename to ...
24897 (profile_quality): ... this one.
24898 (profile_probability): New.
24899 (profile_count): Update.
24900 * profile.c (compute_branch_probabilities): Update.
24901 * recog.c (peep2_attempt): Update.
24902 * sched-ebb.c (schedule_ebbs): Update.
24903 * sched-rgn.c (find_single_block_region): Update.
24904 (compute_dom_prob_ps): Update.
24905 (schedule_region): Update.
24906 * sel-sched-ir.c (compute_succs_info): Update.
24907 * stmt.c (struct case_node): Update.
24908 (do_jump_if_equal): Update.
24909 (get_outgoing_edge_probs): Update.
24910 (conditional_probability): Update.
24911 (emit_case_dispatch_table): Update.
24912 (expand_case): Update.
24913 (expand_sjlj_dispatch_table): Update.
24914 (emit_case_nodes): Update.
24915 * targhooks.c: Update.
24916 * tracer.c (better_p): Update.
24917 (find_best_successor): Update.
24918 * trans-mem.c (expand_transaction): Update.
24919 * tree-call-cdce.c: Update.
24920 * tree-cfg.c (gimple_split_edge): Upate.
24921 (move_sese_region_to_fn): Upate.
24922 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
24923 * tree-eh.c (lower_resx): Upate.
24924 (cleanup_empty_eh_move_lp): Upate.
24925 * tree-if-conv.c (version_loop_for_if_conversion): Update.
24926 * tree-inline.c (copy_edges_for_bb): Update.
24927 (copy_cfg_body): Update.
24928 * tree-parloops.c (gen_parallel_loop): Update.
24929 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
24930 (gimple_gen_time_profiler): Update.
24931 * tree-ssa-dce.c (remove_dead_stmt): Update.
24932 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
24933 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
24934 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
24935 (unloop_loops): Update.
24936 (try_peel_loop): Update.
24937 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
24938 * tree-ssa-loop-split.c (connect_loops): Update.
24939 (split_loop): Update.
24940 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
24941 (hoist_guard): Update.
24942 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
24943 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
24944 (value_replacement): Update.
24945 * tree-ssa-reassoc.c (branch_fixup): Update.
24946 * tree-ssa-tail-merge.c (replace_block_by): Update.
24947 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
24948 (create_edge_and_update_destination_phis): Update.
24949 (compute_path_counts): Update.
24950 (recompute_probabilities): Update.
24951 (update_joiner_offpath_counts): Update.
24952 (freqs_to_counts_path): Update.
24953 (duplicate_thread_path): Update.
24954 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
24955 (struct switch_conv_info): Update.
24956 (gen_inbound_check): Update.
24957 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
24958 (vect_do_peeling): Update.
24959 (vect_loop_versioning): Update.
24960 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
24961 (optimize_mask_stores): Update.
24962 * ubsan.c (ubsan_expand_null_ifn): Update.
24963 * value-prof.c (gimple_divmod_fixed_value): Update.
24964 (gimple_divmod_fixed_value_transform): Update.
24965 (gimple_mod_pow2): Update.
24966 (gimple_mod_pow2_value_transform): Update.
24967 (gimple_mod_subtract): Update.
24968 (gimple_mod_subtract_transform): Update.
24969 (gimple_ic): Update.
24970 (gimple_stringop_fixed_value): Update.
24971 (gimple_stringops_transform): Update.
24972 * value-prof.h: Update.
24974 2017-06-29 Carl Love <cel@us.ibm.com>
24976 * config/rs6000/rs6000-c.c: Add support for built-in functions
24977 vector signed int vec_signed (vector float);
24978 vector signed long long vec_signed (vector double);
24979 vector signed int vec_signed2 (vector double, vector double);
24980 vector signed int vec_signede (vector double);
24981 vector signed int vec_signedo (vector double);
24982 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
24983 instruction generator.
24984 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
24985 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
24986 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
24988 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
24989 vunsignede_v2df): Add define_expands.
24990 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
24991 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
24992 VEC_UNSIGNEDO): Add definitions.
24993 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
24994 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
24995 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
24996 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
24997 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
24998 * config/rs6000/altivec.h (vec_signed, vec_signed2,
24999 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
25000 vec_unsignede, vec_unsignedo): Add builtin defines.
25001 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
25003 * doc/extend.texi: Update the built-in documentation file for the
25004 new built-in functions.
25006 2017-06-29 Richard Biener <rguenther@suse.de>
25008 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
25009 reduction chains to LOOP_VINFO_REDUCTIONS.
25010 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
25011 SLP reductions after processing reduction chains.
25013 2017-06-29 Nathan Sidwell <nathan@acm.org>
25015 * builtins.c (fold_builtin_FUNCTION): Use
25016 lang_hooks.decl_printable_name.
25018 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
25020 PR middle-end/81194
25021 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
25022 with only one label.
25023 * stmt.c (expand_case): Assert NCASES is greater than one.
25025 2017-06-29 Richard Biener <rguenther@suse.de>
25027 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
25029 (group_case_labels): Likewise.
25030 (find_taken_edge): Push sanity checking on val to workers...
25031 (find_taken_edge_cond_expr): ... here
25032 (find_taken_edge_switch_expr): ... and here, handle cases
25033 with just a default label.
25034 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
25035 (group_case_labels): Likewise.
25036 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
25037 group_case_labels does anything cleanup the CFG again.
25039 2017-06-29 Bin Cheng <bin.cheng@arm.com>
25041 PR tree-optimization/81196
25042 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
25043 exit condition comparing two IVs.
25045 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
25047 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
25048 profile to the dummy entry at the end of the list of architectures.
25049 * config/arm/arm-cpu-cdata.h: Regenerated.
25051 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25052 Michael Collison <michael.collison@arm.com>
25055 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
25057 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
25058 (*aarch64_reg_<mode>3_minus_mask): New pattern.
25059 (*aarch64_<optab>_reg_di3_mask2): New pattern.
25060 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
25061 of shift when the shift amount is masked with constant equal to
25062 the size of the mode.
25063 * config/aarch64/predicates.md (subreg_lowpart_operator): New
25066 2017-06-29 Martin Liska <mliska@suse.cz>
25068 * config/i386/i386.opt: Change range from [1,5] to [0,5].
25070 2017-06-29 Yury Gribov <tetra2005@gmail.com>
25073 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
25076 (edge_growth_cache_entry::edge_growth_cache_entry): New
25078 (reset_edge_growth_cache): Update to use constructor.
25080 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25082 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
25083 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
25084 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
25086 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
25088 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
25089 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
25091 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
25093 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
25094 (*-linux-uclibc*): Add t-uclibc tmake_file.
25095 * config/t-musl: New.
25096 * config/t-uclibc: New.
25098 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
25100 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
25102 (gen_comm_data): Emit architectural setting of arch_prof.
25103 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
25105 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
25106 (armv8-m.base, armv8-m.main): Likewise.
25107 * arm-protos.h (arm_build_target): Add profile field.
25108 (arch_option): Likewise.
25109 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
25111 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
25112 arm_active_target.profile.
25114 2017-06-28 Richard Biener <rguenther@suse.de>
25116 PR middle-end/81227
25117 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
25118 TYPE_OVERFLOW_WRAPS.
25119 * match.pd (negate_expr_p): Likewise.
25120 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
25121 fold_build2, not fold_binary.
25123 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
25125 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
25126 Convert memory address to Pmode.
25127 (aarch64_print_operand): Assert MEM operands are always Pmode.
25129 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
25132 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
25133 Remove redundant if.
25134 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
25135 * config/arm/aarch-common-protos.h
25136 (aarch_forward_to_shift_is_not_shifted_re): Remove.
25137 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
25139 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
25142 * multiple_target.c (create_dispatcher_calls): Set the default
25143 clone to be static, not public.
25145 2017-06-28 Richard Biener <rguenther@suse.de>
25147 * tree-vect-loop.c (vectorizable_reduction): Move special
25148 cond reduction IV var creation ...
25149 (vect_create_epilog_for_reduction): ... here. Remove induction_index
25150 parameter. Use STMT_VINFO_VECTYPE.
25151 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
25154 2017-06-28 Martin Liska <mliska@suse.cz>
25157 * ipa-visibility.c (non_local_p): Handle visibility.
25159 2017-06-28 Martin Liska <mliska@suse.cz>
25162 * common.opt: Add IntegerRange to various options.
25163 * opt-functions.awk (integer_range_info): New function.
25164 * optc-gen.awk: Add integer_range_info to cl_options struct.
25165 * opts-common.c (decode_cmdline_option): Handle
25166 CL_ERR_INT_RANGE_ARG.
25167 (cmdline_handle_error): Likewise.
25168 * opts.c (print_filtered_help): Show valid interval in
25169 when --help is provided.
25170 * opts.h (struct cl_option): Add range_min and range_max fields.
25171 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
25173 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
25175 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
25176 (x * C EQ/NE y * C): New transformation.
25178 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
25180 * genmultilib (combination_space): Accept '+' in option names.
25182 2017-06-28 Martin Liska <mliska@suse.cz>
25185 * asan.c (instrument_derefs): Bail out inner references
25186 that are hard register variables.
25188 2017-06-28 Jakub Jelinek <jakub@redhat.com>
25191 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
25192 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
25194 2017-06-28 Richard Biener <rguenther@suse.de>
25196 * tree-vectorizer.h (vect_get_vec_defs): Remove.
25197 (vect_get_slp_defs): Adjust.
25198 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
25200 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
25202 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
25203 get_initial_defs_for_reduction instead of vect_get_vec_defs.
25204 (vectorizable_reduction): Adjust.
25205 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
25207 (vect_get_slp_defs): Likewise.
25208 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
25209 (vectorizable_bswap): Adjust.
25210 (vectorizable_call): Likewise.
25211 (vectorizable_conversion): Likewise.
25212 (vectorizable_assignment): Likewise.
25213 (vectorizable_shift): Likewise.
25214 (vectorizable_operation): Likewise.
25215 (vectorizable_store): Likewise.
25216 (vectorizable_condition): Likewise.
25217 (vectorizable_comparison): Likewise.
25219 2017-06-28 Michael Collison <michael.collison@arm.com>
25222 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
25224 (arm_gen_movmemqi): Removed unused variable 'i'.
25225 Convert 'for' loop into 'while' loop.
25226 (arm_expand_prologue): Remove last unnecessary set of insn.
25227 (thumb_pop): Remove unused variable 'pushed_words'.
25228 (thumb_exit): Remove last unnecessary set of regs_to_pop.
25230 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25232 * config/s390/predicates.md: Use s390_rel_address_ok_p.
25233 * config/s390/s390-protos.h: Add prototype of
25234 s390_rel_address_ok_p.
25235 * config/s390/s390.c (s390_got_symbol): New function.
25236 (s390_rel_address_ok_p): New function.
25237 (legitimize_pic_address): Use s390_rel_address_ok_p.
25238 (s390_load_got): Use s390_got_symbol.
25239 (s390_option_override): Issue error if
25240 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
25241 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
25243 * config/s390/s390.opt: New option mpic-data-is-text-relative.
25245 2017-06-27 Andrew Pinski <apinski@cavium.com>
25247 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
25248 (X * copysign (1.0, X)): New pattern.
25249 (X * copysign (1.0, -X)): New pattern.
25250 (copysign (-1.0, CST)): New pattern.
25252 2017-06-27 Joseph Myers <joseph@codesourcery.com>
25254 * genmultilib (combination_space): Remove variable.
25255 Validate reuse rules against regular expression for any sequence
25256 of multilib options in any order.
25258 2017-06-27 Michael Collison <michael.collison@arm.com>
25260 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
25261 call aarch64_split_simd_combine.
25262 * (aarch64_combine_internal<mode>): Delete pattern.
25263 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
25264 Allow register and subreg operands.
25266 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25268 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
25269 specific need, just fallback on defaults.
25270 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
25272 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25273 Olivier Hainque <hainque@adacore.com>
25275 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
25277 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
25278 targets. Pick a default if no particular attempt applied.
25279 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
25282 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25284 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
25285 (x86_64-wrs-vxworks7): Likewise.
25287 2017-06-27 Marek Polacek <polacek@redhat.com>
25290 * ubsan.c (instrument_null): Check get_base_address's result for null.
25292 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
25294 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
25296 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
25298 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
25299 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
25300 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
25301 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
25302 New function types.
25303 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
25304 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
25305 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
25306 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
25307 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
25308 BUILT_IN_FEUPDATEENV): New builtins.
25309 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
25310 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
25311 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
25312 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
25314 (builtin_structptr_types): Adjust size.
25315 * tree.c (builtin_structptr_types): Add four entries.
25317 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25318 Olivier Hainque <hainque@adacore.com>
25320 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
25321 (TLS_SYM): New local macro, forcing reference to __tls__ on
25322 link command lines for VxWorks 7 RTPs, triggering initialization
25324 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
25325 OS features TLS support, true for RTPs on VxWorks 7.
25326 * config/vxworks.c (vxworks_override_options): Setup emutls
25329 2017-06-27 Jakub Jelinek <jakub@redhat.com>
25331 * predict.c (test_prediction_value_range): Use -1U instead of -1
25332 to avoid narrowing conversion warning.
25333 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
25334 to avoid narrowing conversion warning.
25335 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
25337 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
25339 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25341 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
25342 64bit configurations.
25343 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
25344 (SIZE_TYPE): Likewise.
25345 * config/vxworks.c (vxworks_emutls_var_fields): Use
25346 long_unsigned_type_node instead of unsigned_type_node as the offset
25347 field type, which is "pointer" mode in emutls.c.
25349 2017-06-27 Jakub Jelinek <jakub@redhat.com>
25352 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
25354 PR middle-end/81207
25355 * gimple-fold.c (replace_call_with_call_and_fold): Handle
25356 gimple_vuse copying separately from gimple_vdef copying.
25358 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25360 * value-prof.c (free_hist): Remove call to memset and the enclosing if
25363 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
25364 Olivier Hainque <hainque@adacore.com>
25366 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
25367 for all vxworks7 targets.
25368 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
25369 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
25370 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
25371 variations for VX6/VX7 and 32/64bits later on in ...
25372 (VXWORKS_LIB_SPEC): Leverage new macros.
25373 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
25374 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
25376 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
25378 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
25379 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
25381 2017-06-26 Carl Love <cel@us.ibm.com>
25383 * config/rs6000/rs6000-c.c: Add support for built-in functions
25384 vector bool char vec_reve (vector bool char);
25385 vector signed char vec_reve (vector signed char);
25386 vector unsigned char vec_reve (vector unsigned char);
25387 vector bool int vec_reve (vector bool int);
25388 vector signed int vec_reve (vector signed int);
25389 vector unsigned int vec_reve (vector unsigned int);
25390 vector bool long long vec_reve (vector bool long long);
25391 vector signed long long vec_reve (vector signed long long);
25392 vector unsigned long long vec_reve (vector unsigned long long);
25393 vector bool short vec_reve (vector bool short);
25394 vector signed short vec_reve (vector signed short);
25395 vector double vec_reve (vector double);
25396 vector float vec_reve (vector float);
25397 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
25398 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
25399 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
25400 (altivec_vreve): New pattern.
25401 * config/rs6000/altivec.h (vec_reve): New define.
25402 * doc/extend.texi (vec_rev): Update the built-in documentation file
25403 for the new built-in functions.
25405 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25407 PR tree-optimization/71815
25408 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
25410 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
25412 (slsr_process_phi): Likewise.
25413 (replace_uncond_cands_and_profitable_phis): Don't replace a
25414 multiply candidate with a stride of 1 (copy or cast).
25415 (phi_incr_cost): Call uses_consumed_by_stmt rather than
25417 (lowest_cost_path): Likewise.
25418 (total_savings): Likewise.
25420 2017-06-26 Richard Biener <rguenther@suse.de>
25423 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
25424 Use def_builtin_pure for all gather builtins.
25426 2017-06-26 Richard Biener <rguenther@suse.de>
25428 PR tree-optimization/81203
25429 * tree-tailcall.c (find_tail_calls): Do not move stmts into
25430 non-dominating BBs.
25432 2017-06-26 Marek Polacek <polacek@redhat.com>
25435 * doc/invoke.texi: Document -Wmultistatement-macros.
25437 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
25439 * doc/sourcebuild.texi (ARM-specific attributes): Document new
25440 arm_neon_ok_no_float_abi effective target.
25442 2017-06-26 Richard Biener <rguenther@suse.de>
25444 PR tree-optimization/80928
25445 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
25446 (copy_bbs): Set BB_DUPLICATED flag early.
25447 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
25449 (execute_on_shrinking_pred): Likewise.
25450 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
25451 BB_DUPLICATED blocks.
25452 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
25453 iterate over all PHIs considering removal of *gsi.
25455 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
25457 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
25460 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25462 * config/rs6000/rs6000-string.c: (expand_block_clear,
25463 do_load_for_compare, select_block_compare_mode,
25464 compute_current_alignment, expand_block_compare,
25465 expand_strncmp_align_check, expand_strn_compare,
25466 expand_block_move, rs6000_output_load_multiple)
25467 Move functions related to string/block move/compare
25468 to a separate file.
25469 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
25470 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
25471 for this function which is now used in two files.
25472 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
25473 * config.gcc: Add rs6000-string.o to extra_objs for
25474 targets powerpc*-*-* and rs6000*-*-*.
25476 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
25479 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
25480 32-bit, since indexed is not valid for DImode.
25481 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
25482 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
25483 (define_peephole2 for Altivec d-form load): Add 32-bit support.
25484 (define_peephole2 for Altivec d-form store): Likewise.
25487 * multiple_target.c (create_dispatcher_calls): Only create the
25488 dispatcher call if the function is the default clone of a
25489 versioned function.
25491 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
25493 PR middle-end/80902
25494 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
25495 a call, force the call to not be a tail call.
25497 2017-06-23 Jeff Law <law@redhat.com>
25499 * doc/contrib.texi: Add entry for Steven Pemberton's work on
25502 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
25504 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
25505 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
25506 handling for early expansion of vector shifts (sl,sr,sra,rl).
25507 (builtin_function_type): Add vector shift right instructions
25508 to the unsigned argument list.
25510 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
25512 rtl-optimizatoin/79286
25513 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
25514 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
25515 trap. PIC register plus a const unspec without offset can never trap.
25517 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
25519 * tree.h (builtin_structptr_type): New type.
25520 (builtin_structptr_types): Declare new array.
25521 * tree.c (builtin_structptr_types): New array.
25522 (free_lang_data, build_common_tree_nodes): Use it.
25524 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
25527 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
25530 2017-06-22 Matt Turner <mattst88@gmail.com>
25532 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
25533 Lake models to skylake case. Assume skylake for unknown
25534 models with clflushopt.
25536 2017-06-22 Jeff Law <law@redhat.com>
25538 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
25539 frame sizes that do not satisfy aarch64_uimm12_shift.
25541 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
25543 * profile-count.h (apply_probability,
25544 apply_scale, probability_in): Fix checks for zero.
25546 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
25548 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
25549 * doc/cppdiropts.texi (-I @var{dir}): Document it.
25551 2016-06-22 Richard Biener <rguenther@suse.de>
25553 * tree-vect-loop.c (vect_model_reduction_cost): Handle
25554 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
25555 REDUC_MAX_EXPR support.
25556 (vectorizable_reduction): Likewise.
25557 (vect_create_epilog_for_reduction): Likewise.
25559 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
25561 * match.pd (A / (1 << B) -> A >> B): New.
25562 * generic-match-head.c: Include optabs-tree.h.
25563 * gimple-match-head.c: Likewise.
25564 * optabs-tree.h (target_supports_op_p): New.
25565 * optabs-tree.c (target_supports_op_p): New.
25567 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
25569 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
25570 $gcc_cv_ld --help output.
25571 (gcc_cv_ld_demangle): Likewise.
25572 (gcc_cv_ld_eh_frame_hdr): Likewise.
25573 (gcc_cv_ld_pie): Likewise.
25574 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
25575 (gcc_cv_ld_buildid): Likewise.
25576 (gcc_cv_ld_sysroot): Likewise.
25577 (ld_bndplt_support): Likewise.
25578 (ld_pushpopstate_support): Likewise.
25579 * configure: Regenerate.
25580 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
25582 2017-06-21 Jakub Jelinek <jakub@redhat.com>
25585 * config/i386/sse.md (round<mode>2): Renumber match_dup and
25586 operands indexes to avoid gap between operands and match_dups.
25588 2017-06-21 Andrew Pinski <apinski@cavium.com>
25590 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
25591 Increment Arith_shift and Arith_shift_reg by 1.
25592 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
25594 * config/aarch64/aarch64.c (thunderx_tunings): Enable
25595 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
25596 (aarch64_strip_extend): Add new argument and test for it.
25597 (aarch64_cheap_mult_shift_p): New function.
25598 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
25599 add a cost if it is true.
25600 Update calls to aarch64_strip_extend.
25601 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
25603 2017-06-21 Andrew Pinski <apinski@cavium.com>
25605 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
25607 (thunderxt88): Likewise.
25608 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
25609 (thunderx_prefetch_tune): New variable.
25610 (thunderx2t99_prefetch_tune): Update for the correct values.
25611 (thunderxt88_tunings): New variable.
25612 (thunderx_tunings): Use thunderx_prefetch_tune instead of
25613 generic_prefetch_tune.
25614 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
25616 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25618 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
25619 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
25620 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
25621 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
25622 (aarch64_atomic_cas<mode>, GPI): Likewise.
25624 2017-06-21 Martin Liska <mliska@suse.cz>
25626 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
25627 statements on cold and hot labels.
25628 * predict.c (tree_estimate_probability_bb): Remove the
25629 prediction from this place.
25631 2017-06-21 Martin Liska <mliska@suse.cz>
25633 PR tree-optimization/79489
25634 * gimplify.c (maybe_add_early_return_predict_stmt): New
25636 (gimplify_return_expr): Call the function.
25637 * predict.c (tree_estimate_probability_bb): Remove handling
25639 * predict.def: Update comment about early return predictor.
25640 * gimple-predict.h (is_gimple_predict): New function.
25641 * predict.def: Change default value of early return to 66.
25642 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
25644 * passes.def: Put pass_strip_predict_hints to the beginning of
25647 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
25649 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
25650 FUNCTION_DECL declarations.
25651 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
25653 (dwaf2out_decl): Likewise.
25654 * godump.c (go_early_global_decl): Skip call to the real debug hook
25655 for FUNCTION_DECL declarations.
25656 * passes.c (rest_of_decl_compilation): Skip call to the
25657 early_global_decl debug hook for FUNCTION_DECL declarations, unless
25658 -fdump-go-spec is passed.
25660 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
25662 * config/i386/i386.c (struct builtin_isa): New field pure_p.
25663 Reorder for compactness.
25664 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
25665 (def_builtin_pure, def_builtin_pure2): New functions.
25666 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
25668 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
25670 * match.pd (nop_convert): New predicate.
25671 ((A +- CST1) +- CST2): Allow some NOP conversions.
25673 2017-06-21 Jakub Jelinek <jakub@redhat.com>
25676 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
25677 with ctors/dtors if GOVD_SHARED is set.
25679 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
25681 * config/aarch64/aarch64.md (movti_aarch64):
25682 Emit mov rather than orr.
25683 (movtf_aarch64): Likewise.
25684 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
25685 Emit mov rather than orr.
25687 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
25689 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
25690 Swap alternatives, make integer dup more expensive.
25692 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
25694 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
25695 Return true for non-tls symbols.
25697 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
25699 * config/aarch64/aarch64-cores.def (cortex-a55): New.
25700 (cortex-a75): Likewise.
25701 (cortex-a75.cortex-a55): Likewise.
25702 * config/aarch64/aarch64-tune.md: Regenerate.
25703 * doc/invoke.texi (-mtune): Document new values for -mtune.
25705 2017-06-21 Tom de Vries <tom@codesourcery.com>
25707 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
25708 stack_size feature.
25709 (Effective-Target Keywords, Other attributes): Suggest using
25710 dg-add-options stack_size feature to get stack limit in stack_size
25711 effective target documentation.
25713 2017-06-21 Julian Brown <julian@codesourcery.com>
25714 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25716 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
25717 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
25718 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
25720 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
25721 attribute type list for neon_multiply.
25722 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
25723 attribute type list for neon_multiply.
25724 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
25725 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
25726 attribute type list for neon_multiply.
25727 * config/arm/types.md (crypto_pmull): Add.
25728 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
25729 attribute type list.
25731 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
25733 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
25736 2017-06-20 Jakub Jelinek <jakub@redhat.com>
25738 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
25739 to make sure not to dereference a NULL cost_classes_ptr pointer.
25741 2017-06-20 Carl Love <cel@us.ibm.com>
25743 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
25744 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
25745 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
25746 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
25747 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
25748 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
25749 VMULOSW): New enum "unspec" values.
25750 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
25751 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
25752 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
25753 altivec_vmulosw): New patterns.
25754 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
25755 VMULOSW): Add definitions.
25757 2017-06-20 Julia Koval <julia.koval@intel.com>
25759 * config/i386/i386.c: Fix rounding expand for new pattern.
25760 * config/i386/subst.md: Fix pattern (parallel -> unspec).
25762 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
25764 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
25765 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
25767 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
25769 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
25772 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
25774 * config/aarch64/aarch64-cores.def: Rearrange to sort by
25775 architecture, then by implementer ID.
25776 * config/aarch64/aarch64-tune.md: Regenerate.
25778 2017-06-20 Richard Biener <rguenther@suse.de>
25780 PR middle-end/81097
25781 * fold-const.c (split_tree): Fold to type before negating.
25783 2017-06-20 David Malcolm <dmalcolm@redhat.com>
25785 * diagnostic-show-locus.c
25786 (selftest::test_fixit_deletion_affecting_newline): New function.
25787 (selftest::diagnostic_show_locus_c_tests): Call it.
25789 2017-06-20 Andreas Schwab <schwab@suse.de>
25792 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
25795 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25797 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
25798 __ARM_FEATURE_COPROC according to support.
25800 2017-06-20 Jakub Jelinek <jakub@redhat.com>
25802 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
25803 Rewritten to avoid overflow for > 32-bit pointers.
25806 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
25807 by removing enum keyword.
25808 (ubsan_type_descriptor): Likewise. Formatting fix.
25811 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
25812 splitter): Require TARGET_SSE2 in the condition.
25814 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
25817 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
25818 for doing vector set of SFmode on ISA 3.0.
25819 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
25820 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
25822 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
25823 SFmode value into a V4SF variable that was extracted from another
25824 V4SF variable without converting the element to double precision
25825 and back to single precision vector format.
25826 (vsx_insert_extract_v4sf_p9_2): Likewise.
25828 2017-06-19 Jakub Jelinek <jakub@redhat.com>
25830 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
25831 in UWHI to avoid undefined overflow.
25834 * ubsan.h (enum ubsan_encode_value_phase): New.
25835 (ubsan_encode_value): Change second argument to
25836 enum ubsan_encode_value_phase with default value of
25837 UBSAN_ENCODE_VALUE_GENERIC.
25838 * ubsan.c (ubsan_encode_value): Change second argument to
25839 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
25840 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
25841 create_tmp_var_raw instead of create_tmp_var and use a
25843 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
25844 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
25845 ubsan_encode_value callers.
25848 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
25849 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
25850 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
25852 2017-06-19 Richard Biener <rguenther@suse.de>
25854 PR middle-end/81118
25855 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
25856 estimates if we changed anything.
25858 2017-06-19 Richard Biener <rguenther@suse.de>
25860 PR tree-optimization/80887
25861 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
25862 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
25863 simplified lookups, then reset mprts_hook.
25864 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
25866 (try_to_simplify): Likewise.
25868 2017-06-19 Martin Liska <mliska@suse.cz>
25871 * gimplify.c (gimplify_switch_expr):
25872 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
25874 2017-06-19 Martin Liska <mliska@suse.cz>
25876 * doc/install.texi: Document that PGO runs in 4 stages.
25878 2017-06-19 Martin Liska <mliska@suse.cz>
25881 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
25882 to dispatcher function name.
25883 * multiple_target.c (replace_function_decl): New function.
25884 (create_dispatcher_calls): Redirect both edges and references.
25886 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
25888 * profile-count.c (profile_count::dump): Dump quality.
25889 (profile_count::differs_from_p): Update for unsigned val.
25890 * profile-count.h (profile_count_quality): New enum.
25891 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
25893 2017-06-19 Richard Biener <rguenther@suse.de>
25895 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
25896 struct function as arg.
25897 (estimate_numbers_of_iterations): Export overload with loop arg.
25898 (free_numbers_of_iterations_estimates_loop): Use an overload of
25899 free_numbers_of_iterations_estimates instead.
25900 * tree-cfg.c (remove_bb): Adjust.
25901 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
25902 * tree-parloops.c (gen_parallel_loop): Likewise.
25903 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
25905 (tree_unroll_loops_completely): Likewise.
25906 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
25907 Use an overload instead and export.
25908 (estimated_loop_iterations): Adjust.
25909 (max_loop_iterations): Likewise.
25910 (likely_max_loop_iterations): Likewise.
25911 (estimate_numbers_of_iterations): Take struct function as arg
25913 (loop_exits_before_overflow): Adjust.
25914 (free_numbers_of_iterations_estimates_loop): Use an overload.
25915 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
25916 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
25918 2017-06-19 Richard Biener <rguenther@suse.de>
25921 * ipa-prop.c (find_constructor_constant_at_offset): Handle
25922 RANGE_EXPR conservatively.
25924 2017-06-16 Carl Love <cel@us.ibm.com>
25926 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
25927 definitions for vec_float, vec_float2, vec_floato,
25928 vec_floate built-ins.
25929 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
25930 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
25932 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
25933 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
25934 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
25935 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
25936 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
25937 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
25938 vec_floato): Add builtin defines.
25939 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
25940 Update the built-in documentation file for the new built-in
25943 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25945 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
25946 (mthumb): Mark as the negative of -marm.
25948 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25950 * doc/invoke.texi (ARM Options, -mcpu): Document supported
25952 (ARM Options, -mtune): Document that this accepts the same
25953 extension options as -mcpu.
25954 (ARM Options, -mfpu): Document addition of -mfpu=auto.
25956 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25958 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
25959 permitted extensions.
25961 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25963 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
25964 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
25965 (armv8-m.main): Add option +nodsp.
25966 * config/arm/arm-cpu-cdata.h: Regenerated.
25968 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25970 * config/arm/t-fuchsia: New file.
25971 * config.gcc (arm*-*-fuchsia*): Use it.
25973 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25975 * config/arm/t-symbian: Rewrite for new option infrastructure.
25977 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25979 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
25980 (MULTILIB_REQUIRED): Likewise.
25982 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25984 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
25985 (MULTILIB_RESUE): Likewise.
25986 (MULTILIB_MATCHES): Likewise.
25987 (MULTLIB_REQUIRED): Likewise.
25989 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25991 * config/arm/t-rtems: Rewrite for new option framework.
25993 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25995 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
25996 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
25997 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
25998 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
25999 * config/arm/t-multilib: ... here.
26000 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
26001 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
26002 armv7-a and armv8*-a when A-profile libraries have not been built.
26003 * config/arm/t-rmprofile: Rewrite.
26005 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26007 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
26008 with a backslash. Remove the backslash after substituting unescaped
26010 * doc/fragments.texi (MULTILIB_REUSE): Document it.
26012 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26014 * config.gcc: (arm*-*-*): When building a-profile libraries, force
26015 the driver to pass through the default setting of -mfloat-abi.
26016 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
26018 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
26019 (all_feat_combs): New rule.
26020 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
26022 * config/arm/t-aprofile: Rewrite.
26024 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26026 * config/arm/arm.h (FPUTYPE_AUTO): Define.
26027 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
26028 fpu is not specified by the user/command-line.
26029 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
26030 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
26031 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
26032 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
26033 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
26034 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
26036 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26038 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
26039 * config/arm/t-arm-elf: Rewritten.
26041 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26043 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
26044 have some floating-point instructions.
26045 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
26046 (TARGET_MAYBE_HARD_FLOAT): New macro.
26047 * config/arm/arm-builtins.c (arm_init_builtins): Use
26048 TARGET_MAYBE_HARD_FLOAT.
26049 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
26051 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26053 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
26054 (configargs.h): Include it.
26055 (arm_print_hint_for_fpu_option): New function.
26056 (arm_parse_fpu_option): New function.
26057 (candidate_extension): New class.
26058 (arm_canon_for_multilib): New function.
26059 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
26060 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
26061 (ARCH_CANONICAL_SPECS): New macro.
26062 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
26064 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26066 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
26067 are set after handling multilib fragments. Set target_cpu_default2
26070 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26072 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
26074 (arm*-*-*): Set target_cpu_default2 to a quoted string.
26075 * config/arm/parsecpu.awk (check_cpu): Validate any extension
26077 (check_arch): Likewise.
26078 * config/arm/arm.c (arm_configure_build_target): Handle
26079 TARGET_CPU_DEFAULT being a string constant. Scan any feature
26080 options in the default.
26082 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26084 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
26085 when an option is an alias of another.
26086 * config/arm/parsecpu.awk (optalias): New parser token.
26087 (gen_comm_data): Mark non-alias options as such. Emit entries
26088 for extension aliases.
26089 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
26090 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
26091 (armv6kz, armv6zk, armv6t2): Likewise.
26092 (armv7): Make vfpv3-d16 an alias.
26093 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
26095 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
26096 Sort in canonical order.
26097 (armv8-a): Sort in canonical order.
26098 (armv8.1-a, armv8.2-a): Likewise.
26099 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
26101 (cortex-a9): Sort in canonical order.
26102 * config/arm/arm.c (selftests.h): Include it.
26103 (arm_test_cpu_arch_data): New function.
26104 (arm_run_self_tests): New function.
26105 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
26106 (targetm): Move declaration to the end of the file.
26107 * arm-cpu-cdata.h: Regenerated.
26109 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26111 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
26112 call to target_mode_check describing the type of option passed.
26113 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
26114 (arm_target_thumb_only): Use arm_parse_arch_option_name or
26115 arm_parse_cpu_option_name to match parameters against list of
26117 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
26118 arm_arch_core_flags data structure.
26119 * config/arm/arm-cpu_cdata.h: Regenerated.
26121 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26123 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
26125 (arm_print_hint_for_cpu_option): Likewise.
26126 (arm_print_hint_for_arch_option): Likewise.
26127 (arm_parse_cpu_option_name): Likewise.
26128 (arm_parse_arch_option_name): Likewise.
26129 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
26130 of entries in the all_fpus list.
26131 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
26132 (arm_parse_cpu_option_name): Declare.
26133 (arm_parse_arch_option_name): Declare.
26134 (arm_parse_option_features): Declare.
26135 (arm_intialize_isa): Declare.
26136 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
26138 (gen_comm_data): ... here. Make definitions non-static.
26139 * config/arm/arm-cpu-data.h: Regenerated.
26140 * config/arm/arm-cpu-cdata.h: Regenerated.
26142 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26144 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
26145 (cpu_arch_extension): New structure.
26146 (cpu_arch_option, arch_option, cpu_option): New structures.
26147 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
26148 architecture types.
26149 (gen_data): Generate new format data tables.
26150 * config/arm/arm.c (cpu_tune): New structure.
26151 (cpu_option, processors): Delete.
26152 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
26153 (arm_print_hint_for_cpu_option): ... this and ...
26154 (arm_print_hint_for_arch_option): ... this.
26155 (arm_parse_arch_cpu_name): Delete. Replace with ...
26156 (arm_parse_cpu_option_name): ... this and ...
26157 (arm_parse_arch_option_name): ... this.
26158 (arm_unrecognized_feature): Change type of target parameter to
26160 (arm_parse_arch_cpu_features): Delete. Replace with ...
26161 (arm_parse_option_features): ... this.
26162 (arm_configure_build_target): Rework to use new configuration data
26164 (arm_print_tune_info): Rework for new configuration data tables.
26165 * config/arm/arm-cpu-data.h: Regenerated.
26166 * config/arm/arm-cpu.h: Regenerated.
26168 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26170 * Makefile.in (OBJS): Move sbitmap.o from here ...
26171 (OBJS-libcommon): ... to here.
26173 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26175 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
26176 (ISA_ALL_CRYPTO): New macro.
26177 (ISA_ALL_SIMD): New macro
26178 (ISA_ALL_FP): New macro.
26179 * config/arm/arm.c (fpu_bitlist): Update initializer.
26180 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
26182 (arm9e): Add fpu. Add option for nofp
26183 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
26184 (arm926ej-s, arm1026ej-s): Likewise.
26185 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
26186 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
26187 neon-fp16, neon-vfpv4, nofp and nosimd.
26188 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
26189 (cortex-a8): Add fpu. Add option for nofp.
26190 (cortex-a9): Add fpu. Add options for nosimd and nofp.
26191 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
26192 (cortex-r4f): Add fpu.
26193 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
26194 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
26196 (cortex-r8): Likewise.
26197 (cortex-m4): Add fpu. Add option for nofp.
26198 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
26199 (cortex-a17.cortex-a7): Likewise.
26200 (cortex-a32): Add fpu. Add options for crypto and nofp.
26201 (cortex-a35, cortex-a53): Likewise.
26202 (cortex-a57): Add fpu. Add option for crypto.
26203 (cortex-a72, cortex-a73): Likewise.
26204 (exynos-m1): Likewise.
26205 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
26206 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
26207 (cortex-m33): Add fpu. Add option for nofp.
26208 * config/arm/arm-cpu-cdata.h: Regenerated
26209 * config/arm/arm-cpu-data.h: Regenerated.
26211 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26213 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
26214 (armv5te, armv5tej): Likewise.
26215 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
26216 (armv7): Add options fp and vfpv3-d16.
26217 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
26218 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
26220 (armv7ve): Likewise.
26221 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
26222 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
26223 (armv8-a): Add nocrypto option.
26224 (armv8.1-a, armv8.2-a): Likewise.
26225 (armv8-m.main): add options fp, fp.dp and nofp.
26227 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26229 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
26231 (armv8-a+crc): Delete.
26232 (armv8.1-a): Add options simd, crypto and nofp.
26233 (armv8.2-a): Add options fp16, simd, crypto and nofp.
26234 (armv8.2-a+fp16): Delete.
26235 (armv8-m.main): Add option dsp.
26236 (armv8-m.main+dsp): Delete.
26237 (cortex-a8): Add fpu. Add nofp option.
26238 (cortex-a9): Add fpu. Add nofp and nosimd options.
26239 * config/arm/parsecpu.awk (gen_data): Generate option tables and
26240 link to main cpu and architecture data structures.
26241 (gen_comm_data): Only put isa attributes from the main architecture
26243 (option): New statement for architecture and CPU entries.
26244 * arm.c (struct cpu_option): New structure.
26245 (struct processors): Add entry for options.
26246 (arm_unrecognized_feature): New function.
26247 (arm_parse_arch_cpu_name): Ignore any characters after the first
26249 (arm_parse_arch_cpu_feature): New function.
26250 (arm_configure_build_target): Separate out any CPU and architecture
26251 features and parse separately. Don't error out if -mfpu=auto is
26252 used with only an architecture string.
26253 (arm_print_asm_arch_directives): New function.
26254 (arm_file_start): Call it.
26255 * config/arm/arm-cpu-cdata.h: Regenerated.
26256 * config/arm/arm-cpu-data.h: Likewise.
26257 * config/arm/arm-tables.opt: Likewise.
26259 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26261 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
26262 assembler when it is not -mfpu=auto.
26264 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26266 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
26267 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
26268 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
26269 (ASM_CPU_SPEC): Rewrite.
26270 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
26271 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
26272 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
26273 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
26274 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
26275 copied string is NUL-terminated. Also strip any characters prefixed
26277 (arm_rewrite_selected_arch): New function.
26278 (arm_rewrite_march): New function.
26280 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26282 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
26283 (x_arm_cpu_string, x_arm_tune_string): Likewise.
26284 (march, mcpu, mtune): Convert to string-based options.
26285 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
26286 (arm_parse_arch_cpu_name): New function.
26287 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
26288 identify selected architecture or CPU.
26289 (arm_option_save): New function.
26290 (TARGET_OPTION_SAVE): Redefine.
26291 (arm_option_restore): Restore string options.
26292 (arm_option_print): Print string options.
26294 2017-06-16 Martin Sebor <msebor@redhat.com>
26296 PR tree-optimization/80933
26297 PR tree-optimization/80934
26298 * builtins.c (fold_builtin_3): Do not handle bcmp here.
26299 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
26300 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
26301 (gimple_fold_builtin): Call them.
26303 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26305 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
26306 as unlikely; update profile.
26308 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26310 * predict.c (force_edge_cold): Handle declaring edges impossible
26313 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26315 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
26317 (try_unroll_loop_completely): Fix reporting.
26319 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26321 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
26323 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
26326 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
26327 if given a non-constant argument for an intrinsic which requires a
26330 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26332 * profile.c (compare_freqs): New function.
26333 (branch_prob): Sort edge list.
26334 (find_spanning_tree): Assume that the list is priority sorted.
26336 2017-06-16 Richard Biener <rguenther@suse.de>
26338 PR tree-optimization/81090
26339 * passes.def (pass_record_bounds): Remove.
26340 * tree-pass.h (make_pass_record_bounds): Likewise.
26341 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
26342 make_pass_record_bounds): Likewise.
26343 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
26344 not free niter estimates at the beginning but at the end.
26345 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
26347 2017-06-16 Richard Biener <rguenther@suse.de>
26349 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
26350 initializer to workaround ICE in host GCC 4.8.
26352 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26354 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
26356 (clone_inlined_nodes): Update.
26358 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
26360 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
26361 prefetch settings, and enable prefetching by default at -O3.
26363 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
26365 * config/aarch64/aarch64.c (aarch64_override_options_internal):
26366 Set flag_prefetch_loop_arrays according to tuning data.
26368 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
26370 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
26371 New tune structure.
26372 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
26373 [Unrelated to main purpose of the patch] Place the pointer field last
26374 to enable type checking errors when tune structure are wrongly merged.
26375 * config/aarch64/aarch64.c (generic_prefetch_tune,)
26376 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
26377 (thunderx2t99_prefetch_tune): New tune constants.
26378 (tune_params *_tunings): Update all tunings (no functional change).
26379 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
26380 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
26381 from tunings structures.
26383 2017-06-16 Jakub Jelinek <jakub@redhat.com>
26386 * ubsan.c (instrument_null): Add T argument, use it instead
26387 of computing it based on IS_LHS.
26388 (instrument_object_size): Likewise.
26389 (pass_ubsan::execute): Adjust instrument_null and
26390 instrument_object_size callers to pass gimple_get_lhs or
26391 gimple_assign_rhs1 result to it. Use instrument_null instead of
26392 calling get_base_address and instrument_mem_ref. Handle
26393 aggregate call arguments for object-size sanitization.
26395 2017-06-16 Yury Gribov <tetra2005@gmail.com>
26397 PR tree-optimization/81089
26398 * tree-vrp.c (is_masked_range_test): Validate operands of
26401 2017-06-15 Martin Sebor <msebor@redhat.com>
26404 * dumpfile.c (dump_register): Avoid calling memset to initialize
26405 a class with a default ctor.
26406 * gcc.c (struct compiler): Remove const qualification.
26407 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
26408 * hash-table.h: Ditto.
26409 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
26411 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
26412 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
26414 * params.h (struct param_info): Make struct members non-const.
26415 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
26416 with default initialization.
26417 * vec.h (vec_copy_construct, vec_default_construct): New helper
26419 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
26420 with vec_copy_construct.
26421 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
26422 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
26423 * doc/invoke.texi (-Wclass-memaccess): Document.
26425 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26427 * emit-rtl.h (is_leaf): Update comment about local
26428 register allocator.
26430 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
26433 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
26434 for a variable to have a section before checking if the section has a
26436 Set section to.persistent if persistent attribute is set.
26437 Warn if .persistent attribute is used on an automatic variable.
26439 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
26441 PR rtl-optimization/80474
26442 * reorg.c (update_block): Do not ignore instructions in a delay slot.
26444 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
26446 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
26449 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
26451 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
26452 (casesi): Emit bounds checking as RTL.
26453 (casesi_internal_mips16_<mode>): Remove bounds checking.
26455 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
26457 * config/xtensa/xtensa.c (xtensa_option_override): Append
26458 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
26459 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
26460 xtensa_doloop_hooks): Define unconditionally.
26461 (xtensa_reorg_loops): Only call reorg_loops in the presence of
26463 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
26464 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
26465 for it in xtensa_option_override.
26466 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
26467 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
26469 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
26471 * doc/cppopts.texi: Document '-' special value to -MF.
26473 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
26475 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
26476 (cortex_a53_fconst): Likewise.
26477 (cortex_a53_fpmul): Likewise.
26478 (cortex_a53_f_load_64): Likewise.
26479 (cortex_a53_f_load_many): Likewise.
26480 (cortex_a53_advsimd_alu): Likewise.
26481 (cortex_a53_advsimd_alu_q): Likewise.
26482 (cortex_a53_advsimd_mul): Likewise.
26483 (cortex_a53_advsimd_mul_q): Likewise.
26484 (fpmac bypass): Add new bypass for fpmac-fpmac case.
26485 Add missing fmul, r2f_cvt and fconst cases.
26487 2017-06-14 Richard Biener <rguenther@suse.de>
26489 PR middle-end/81088
26490 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
26492 (fold_binary_loc): When associating do not treat pre-existing
26493 TREE_OVERFLOW on literal constants as a reason to allow
26494 TREE_OVERFLOW on associated literal constants.
26496 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
26498 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
26499 (MASK_FEATURES): New macro.
26500 * config/sparc/sparc.c (sparc_option_override): Remove the special
26501 handling of -mfpu and generalize it to all MASK_FEATURES switches.
26503 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
26505 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
26506 a division of 0 if non-call exceptions are enabled.
26508 2017-06-14 Andrew Pinski <apinski@cavium.com>
26509 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26512 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
26513 Improve vector initialization code gen for only variable case.
26515 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
26517 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
26519 2017-06-14 Richard Biener <rguenther@suse.de>
26521 PR tree-optimization/81083
26522 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
26525 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26527 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
26528 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
26529 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
26530 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
26531 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
26532 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
26534 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26536 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
26537 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
26539 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26541 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
26543 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26545 * config/rs6000/t-rtems: Don't handle SPE.
26547 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26549 * config/rs6000/t-linux: Don't handle SPE.
26551 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26553 * config/rs6000/eabispe.h: Delete file.
26555 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26557 * config/rs6000/t-spe: Delete file.
26559 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26561 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
26562 (rs6000_legitimate_offset_address_p): Return false for anything in
26563 V2SImode or V2SFmode.
26565 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26567 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
26568 except V2SF and V2SI. Rearrange the vector modes, and add comments.
26569 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
26571 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
26572 (rs6000_legitimate_offset_address_p): Ditto.
26573 (rs6000_emit_move): Ditto.
26574 (rs6000_init_builtins): Remove V4HI_type_node.
26576 2017-06-13 Martin Liska <mliska@suse.cz>
26579 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
26580 (gate_asan): Likewise.
26581 * asan.h (asan_no_sanitize_address_p): Remove the function.
26582 (sanitize_flags_p): New function.
26583 * builtins.def: Fix coding style.
26584 * common.opt: Use renamed enum value.
26585 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
26586 * doc/extend.texi: Document no_sanitize attribute.
26587 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
26588 to SANITIZE_UNDEFINED_NONDEFAULT.
26589 * gcc.c (sanitize_spec_function): Use the renamed enum value.
26590 * gimple-fold.c (optimize_atomic_compare_exchange_p):
26591 Use sanitize_flags_p.
26592 * gimplify.c (gimplify_function_tree): Likewise.
26593 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
26594 * opts.c (parse_no_sanitize_attribute): New function.
26595 (common_handle_option): Use renamed enum value.
26596 * opts.h (parse_no_sanitize_attribute): Declare.
26597 * tree.c (sanitize_flags_p): New function.
26598 * tree.h: Declared here.
26599 * tsan.c: Use sanitize_flags_p.
26600 * ubsan.c (ubsan_expand_null_ifn): Likewise.
26601 (instrument_mem_ref): Likewise.
26602 (instrument_bool_enum_load): Likewise.
26603 (do_ubsan_in_current_function): Remove the function.
26604 (pass_ubsan::execute): Use sanitize_flags_p.
26605 * ubsan.h: Remove do_ubsan_in_current_function
26606 * tree-cfg.c (print_no_sanitize_attr_value): New function.
26607 (dump_function_to_file): Use it here.
26609 2017-06-13 Martin Jambor <mjambor@suse.cz>
26611 PR tree-optimization/80803
26612 PR tree-optimization/81063
26613 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
26614 (propagate_subaccesses_across_link): Enqueue subtree whenever
26615 necessary instead of relying on the caller.
26617 2017-06-13 Martin Jambor <mjambor@suse.cz>
26619 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
26620 that have a first_link.
26621 (sort_and_splice_var_accesses): Do not check first_link before
26623 (subtree_mark_written_and_enqueue): Likewise.
26624 (propagate_all_subaccesses): Likewise and do not stop at first
26625 parent with a first_link.
26627 2017-06-13 Martin Jambor <mjambor@suse.cz>
26629 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
26632 2017-06-13 Yury Gribov <tetra2005@gmail.com>
26634 * match.pd: New pattern.
26636 2017-06-13 Yury Gribov <tetra2005@gmail.com>
26638 * tree-vrp.c (is_masked_range_test): New function.
26639 (register_edge_assert_for): Determine ranges for
26642 2017-06-13 Yury Gribov <tetra2005@gmail.com>
26644 PR tree-optimization/67328
26645 * fold-const.c (maskable_range_p): New function.
26646 (build_range_check): Generate bittests if possible.
26648 2017-06-13 Martin Liska <mliska@suse.cz>
26650 * gimple-pretty-print.c (dump_probability): Add new argument.
26651 (dump_edge_probability): Dump both probability and count.
26652 (dump_gimple_label): Likewise.
26653 (dump_gimple_bb_header): Likewise.
26655 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
26658 * config/avr/avr-devices.c: Fix indentation.
26659 * config/avr/gen-avr-mmcu-specs.c: Dito.
26661 2017-06-13 Richard Biener <rguenther@suse.de>
26663 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
26664 instead get vector type from stmt_info.
26665 (vectorizable_reduction): Adjust. Remove dead code.
26667 2017-06-13 Richard Biener <rguenther@suse.de>
26669 PR middle-end/81065
26670 * fold-const.c (extract_muldiv_1): Remove bogus distribution
26671 case of C * (x * C2 + C3).
26672 (fold_addr_of_array_ref_difference): Properly fold index difference.
26674 2017-06-12 David S. Miller <davem@davemloft.net>
26677 * config/sparc/sparc.md (return expander): Emit frame blockage if
26678 function uses alloca.
26680 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
26682 * combine.c (make_field_assignment): Check len rather than the mode
26683 precision when calling force_to_mode.
26685 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
26687 Support multilibs and devices that see flash in RAM address range.
26690 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
26691 (avr_mcu_t) <flash_pm_offset>: New field.
26692 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
26693 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
26694 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
26695 (AVR_TINY_PM_OFFSET): Remove macro.
26696 * config/avr/avr.opt (-mshort-calls): New option.
26697 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
26698 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
26699 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
26700 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
26701 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
26702 instead of avr_arch->have_jmp_call.
26703 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
26704 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
26705 avr_arch->flash_pm_offset to define.
26706 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
26707 new field flash_pm_offset. Add entry for avrxmega3.
26708 (avr_texinfo): Add entry for avrxmega3.
26709 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
26710 attiny212, attiny214,
26711 attiny412, attiny414, attiny416, attiny417,
26712 attiny814, attiny816, attiny817,
26713 attiny1614, attiny1616, attiny1617,
26714 attiny3214, attiny3216, attiny3217.
26715 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
26716 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
26717 (avr_print_operand_address) [AVR_TINY]: Same.
26718 (avr_asm_init_sections) <readonly_data_section>: Only patch
26719 callback if avr_arch->flash_pm_offset = 0.
26720 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
26721 for rodata if avr_arch->flash_pm_offset != 0.
26722 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
26723 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
26724 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
26725 (m_options): Append opt_rcall.
26726 (m_dirnames): Append dir_rcall.
26727 * config/avr/t-multilib: Regenerate.
26729 * configure.ac [target=avr]: Check whether avrxmega3 default
26730 linker description file works as needed.
26731 * configure: Regenerate.
26732 * doc/avr-mmcu.texi: Regenerate.
26733 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
26734 <__AVR_ARCH__>: Document avrxmega3 and 103.
26735 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
26736 <__AVR_SHORT_CALLS__>: Document it.
26737 <__AVR_PM_BASE_ADDRESS__>: Document it.
26738 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
26739 (AVR Variable Attributes) <progmem>: Document this is
26740 not needed for avrxmega3.
26741 (AVR Named Address Spaces) <__flash>: Dito.
26743 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
26745 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
26747 2017-06-12 Doug Rupp <rupp@adacore.com>
26749 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
26750 Append vxworks-stdint.h to the tm_file list.
26751 * config/vxworks-stdint.h: New file.
26753 2017-06-12 Martin Liska <mliska@suse.cz>
26755 PR tree-optimization/81041
26756 * tree-profile.c (gimple_gen_ic_func_profiler):
26757 Create an extra BB in profile-generate
26758 (gimple_gen_time_profiler): Likewise.
26760 2017-06-12 Jakub Jelinek <jakub@redhat.com>
26762 PR tree-optimization/81003
26763 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
26764 (update_range_test): Use it instead of force_gimple_operand_gsi.
26766 2017-06-12 Richard Biener <rguenther@suse.de>
26768 PR tree-optimization/81053
26769 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
26770 with backedge value not defined in loop. Simplify def stmt
26773 2017-06-11 Tom de Vries <tom@codesourcery.com>
26776 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
26778 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
26779 nvptx_cannot_force_const_mem.
26781 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
26783 * opts.c (finish_options): Move test for flag_split_stack after
26784 it has been initialized.
26786 2017-06-11 Jason Merrill <jason@redhat.com>
26788 * tree.h (id_equal): New.
26789 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
26790 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
26791 instead of strcmp of IDENTIFIER_POINTER.
26793 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
26795 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
26796 (mark_all_inlined_calls_cdtor): Fix formating.
26797 (inline_transform): Rescale profile before inlining.
26799 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
26801 * cgraph.h (cgraph_edge::clone): Update prototype.
26802 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
26803 (cgraph_node::create_clone): Update.
26804 (cgraph_node::create_version_clone): Update.
26805 * tree-inline.c (copy_bb): Update.
26806 (expand_call_inline): Update.
26808 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
26810 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
26811 factored out from ...
26812 (rs6000_emit_prologue): ... here.
26814 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
26816 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
26817 factored out from ...
26818 (rs6000_emit_prologue): ... here.
26820 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
26822 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
26824 (handle_missing_profiles): Fix computation of tp_first_run.
26825 (counts_to_freqs): Do not touch freqs when count is 0.
26827 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
26829 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
26832 2017-06-10 Tom de Vries <tom@codesourcery.com>
26834 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
26835 attributes): Document signal effective target.
26837 2017-06-10 Tom de Vries <tom@codesourcery.com>
26839 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
26840 Document effective target stack_size.
26842 2017-06-09 David Malcolm <dmalcolm@redhat.com>
26844 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
26845 to the edit_context if they can be auto-applied.
26847 2017-06-9 Ian Lance Taylor <iant@golang.org>
26849 * opts.c (finish_options): If -fsplit-stack, disable implicit
26850 -forder-blocks-and-partition.
26851 * doc/invoke.texi (Optimize Options): Document that when using
26852 -fsplit-stack -forder-blocks-and-partition is not implicitly
26855 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
26857 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
26858 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
26859 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
26860 * builtins.def (abort, trap, unreachable): Declare cold.
26861 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
26862 * tree-core.h (ECF_COLD): New.
26863 * tree.c (set_call_expr_flags): Handle ECF_COLD.
26864 (build_common_builtin_nodes): Mark unreachable and abort as cold.
26866 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
26868 * predict.c (unlikely_executed_stmt_p): Cleanup.
26870 2017-06-09 Richard Biener <rguenther@suse.de>
26872 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
26873 model if the ref is always written to.
26875 2017-06-09 Tamar Christina <tamar.christina@arm.com>
26877 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
26879 2017-06-09 Tamar Christina <tamar.christina@arm.com>
26881 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
26884 2017-06-09 Tom de Vries <tom@codesourcery.com>
26887 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
26888 "target cannot support label values" when encountering LABEL_REF.
26890 2017-06-09 Martin Liska <mliska@suse.cz>
26892 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
26893 (gimple_gen_ic_func_profiler): Emit direct comparison
26894 of __gcov_indirect_call_callee with NULL.
26895 (gimple_gen_time_profiler): Change probability from
26896 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
26898 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
26900 * profile.c (edge_gcov_counts): Turn to pointer.
26901 (compute_branch_probabilities, compute_branch_probabilities): Update.
26902 (branch_prob): Do not clear edge_gcov_count.
26903 * profile.h (edge_gcov_counts): Turn to pointer.
26904 (edge_gcov_count): Update.
26906 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
26908 * gimple.h (gimple_check_failed): Mark cold.
26910 2017-06-09 Richard Biener <rguenther@suse.de>
26912 PR tree-optimization/66623
26913 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
26914 refactor check_reduction into two parts, properly computing
26915 whether we have to check reduction validity for outer loop
26918 2017-06-09 Richard Biener <rguenther@suse.de>
26920 PR tree-optimization/79483
26921 * graphite-scop-detection.c (order): New global.
26922 (get_order): Compute bb to order mapping that satisfies code
26923 generation constraints.
26924 (cmp_pbbs): New helper.
26925 (build_scops): Start domwalk at entry block, sort generated
26928 2017-06-09 Richard Biener <rguenther@suse.de>
26930 PR middle-end/81007
26931 * ipa-polymorphic-call.c
26932 (ipa_polymorphic_call_context::restrict_to_inner_class):
26933 Skip FIELD_DECLs with error_mark_node type.
26934 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
26937 2017-06-09 Martin Liska <mliska@suse.cz>
26939 * predict.c (struct branch_predictor): New struct.
26940 (test_prediction_value_range): New test.
26941 (predict_c_tests): New function.
26942 * selftest-run-tests.c (selftest::run_tests): Run the function.
26943 * selftest.h: Declare new tests.
26945 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
26948 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
26949 gen_add3_insn did not fail.
26950 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
26951 r0, construct that number in a temporary reg and add that reg to r0.
26952 If asked to put the result in r0 as well, fail.
26954 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
26956 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
26957 for early expansion of vec_eqv.
26959 2017-06-08 Jakub Jelinek <jakub@redhat.com>
26961 PR middle-end/81005
26962 * ubsan.c (instrument_null): Avoid pointless code temporary.
26963 (pass_ubsan::execute): Instrument aggregate arguments of calls.
26965 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
26969 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
26972 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
26975 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
26977 * predict.c (unlikely_executed_edge_p): Move ahead.
26978 (probably_never_executed_edge_p): Use it.
26980 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
26982 PR middle-end/79988
26983 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
26984 gimple_call_builtin_p call.
26986 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
26988 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
26989 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
26990 rtl_check_failed_type2, rtl_check_failed_code1,
26991 rtl_check_failed_code2, rtl_check_failed_code_mode,
26992 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
26993 rtvec_check_failed_bounds, rtl_check_failed_flag,
26994 _fatal_insn_not_found, _fatal_insn): Likewise.
26995 * tree.h (tree_contains_struct_check_failed,
26996 tree_check_failed, tree_not_check_failed,
26997 tree_class_check_failed, tree_range_check_failed,
26998 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
26999 tree_vec_elt_check_failed, phi_node_elt_check_failed,
27000 tree_operand_check_failed, omp_clause_check_failed,
27001 omp_clause_operand_check_failed, omp_clause_range_check_failed):
27004 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27006 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
27007 flag_branch_probabilities.
27008 * ipa-inline.c (edge_badness): Likewise.
27009 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
27010 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
27011 * predict.c (maybe_hot_frequency_p): Likewise.
27012 (probably_never_executed): Likewise.
27013 * sched-ebb.c (schedule_ebbs): Likewise.
27014 * sched-rgn.c (find_single_block_region): Likewise.
27015 * tracer.c (tail_duplicate): Likewise.
27017 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27019 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
27020 longer requires x_flag_profile_use.
27022 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27024 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
27025 instead of flag_reorder_blocks_and_partition.
27026 * dbxout.c (dbxout_function_end): Likewise.
27027 * dwarf2out.c (gen_subprogram_die): Likewise.
27028 * haifa-sched.c (sched_create_recovery_edges): Likewise.
27029 * hw-doloop.c (reorg_loops): Likewise.
27030 * varasm.c (assemble_start_function,
27031 assemble_end_function): Likewise.
27032 (decide_function_section): Do not check for
27033 flag_reorder_blocks_and_partition.
27035 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
27037 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
27039 (chkp_get_hard_register_fake_addr_expr): Ditto.
27040 (chkp_build_addr_expr): Add check for hard reg case.
27041 (chkp_parse_array_and_component_ref): Ditto.
27042 (chkp_find_bounds_1): Ditto.
27043 (chkp_process_stmt): Don't generate bounds store for
27046 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27048 * predict.c (maybe_hot_bb_p): Do not check profile status.
27049 (maybe_hot_edge_p): Likewise.
27050 (probably_never_executed): Check for zero counts even if profile
27052 (unlikely_executed_edge_p): New function.
27053 (unlikely_executed_stmt_p): New function.
27054 (unlikely_executed_bb_p): New function.
27055 (set_even_probabilities): Use unlikely predicates.
27056 (combine_predictions_for_bb): Likewise.
27057 (predict_paths_for_bb): Likewise.
27058 (predict_paths_leading_to_edge): Likewise.
27059 (determine_unlikely_bbs): New function.
27060 (estimate_bb_frequencies): Use it.
27061 (compute_function_frequency): Use zero counts even if profile is
27063 * profile-count.h: Fix typo.
27065 2017-08-08 Julia Koval <julia.koval@intel.com>
27067 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
27068 _mm512_mask_cvtsepi16_storeu_epi8,
27069 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
27070 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
27071 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
27072 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
27073 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
27074 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
27075 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
27076 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
27077 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
27078 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
27079 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
27080 __builtin_ia32_pmovuswb256mem_mask,
27081 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
27082 __builtin_ia32_pmovwb512mem_mask): New builtins.
27084 2017-08-08 Julia Koval <julia.koval@intel.com>
27086 PR target/73350,80862
27087 * config/i386/subst.md (round): Fix round pattern.
27088 * config/i386/i386.c (ix86_erase_embedded_rounding):
27089 Fix erasing rounding for the fixed pattern.
27091 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27093 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
27095 2017-06-08 Martin Liska <mliska@suse.cz>
27097 PR gcov-profile/80911
27098 * gcov.c (block_info::block_info): New constructor.
27100 2017-06-07 Carl Love <cel@us.ibm.com>
27102 * config/rs6000/rs6000-c: The return type of the following
27103 built-in functions was implemented as int not long long. Fix sign
27104 of return value for the unsigned version of vec_mulo and vec_mule.
27105 vector unsigned long long vec_bperm (vector unsigned long long,
27106 vector unsigned char)
27107 vector signed long long vec_mule (vector signed int,
27109 vector unsigned long long vec_mule (vector unsigned int,
27110 vector unsigned int)
27111 vector signed long long vec_mulo (vector signed int,
27113 vector unsigned long long vec_mulo (vector unsigned int,
27114 vector unsigned int)
27115 * doc/extend.texi: Fix the documentation for the built-in
27118 2017-06-07 Carl Love <cel@us.ibm.com>
27121 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
27124 2017-06-07 Carl Love <cel@us.ibm.com>
27126 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
27127 support, Generate doublehv for signed int/float for BE case only.
27129 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
27131 * doc/invoke.texi (mcx16): Rewrite.
27133 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27135 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
27136 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
27137 *mov<mode>_softfloat, and an anonymous splitter): Use
27138 nonimmediate_operand instead of rs6000_nonimmediate_operand.
27140 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27142 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
27144 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
27145 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
27146 (rs6000_debug_reg_global): Adjust.
27147 (rs6000_init_hard_regno_mode_ok): Adjust.
27148 (rs6000_dbx_register_number): Adjust.
27149 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
27150 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
27151 Remove SPE_ACC and SPEFSCR.
27152 (REG_ALLOC_ORDER): Ditto.
27153 (FRAME_POINTER_REGNUM): Change to 111.
27154 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
27155 (REG_CLASS_NAMES): Ditto.
27156 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
27157 (REGISTER_NAMES): Ditto.
27158 (ADDITIONAL_REG_NAMES): Ditto.
27159 (rs6000_reg_names): Ditto.
27160 * config/rs6000/rs6000.md: Renumber some register number
27163 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27165 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
27167 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
27168 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
27170 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
27171 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
27172 Delete the SPE high registers.
27173 (REG_ALLOC_ORDER): Ditto.
27174 (enum reg_class): Remove SPE_HIGH_REGS.
27175 (REG_CLASS_NAMES): Ditto.
27176 (REG_CLASS_CONTENTS): Delete the SPE high registers.
27177 (REGISTER_NAMES): Ditto.
27178 (rs6000_reg_names): Ditto.
27179 * doc/tm.texi.in: Remove SPE as example.
27180 * doc/tm.texi: Regenerate.
27182 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27184 * config/rs6000/8540.md (ppc8540_brinc): Delete.
27185 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
27186 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
27187 * config/rs6000/rs6000.md (type): Remove "brinc".
27189 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27191 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
27192 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
27193 * config/rs6000/linuxspe.h: Delete file.
27194 * config/rs6000/rs6000.md: Don't include spe.md.
27195 * config/rs6000/spe.h: Delete file.
27196 * config/rs6000/spe.md: Delete file.
27197 * config/rs6000/t-rs6000: Remove spe.md.
27199 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27201 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
27202 (reg_or_none500mem_operand): Delete.
27203 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
27204 instead of reg_or_none500mem_operand.
27206 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27208 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
27209 handling of SPE flags.
27210 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
27212 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27214 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
27216 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
27217 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
27218 paired_divv2sf3): Similar.
27219 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
27220 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
27221 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
27223 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
27224 Rename the paired_* instruction patterns.
27225 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
27227 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
27228 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
27229 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
27230 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
27231 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
27232 PAIRED_VECTOR_MODE.
27233 (struct machine_function): Delete field spe_insn_chain_scanned_p.
27234 (spe_func_has_64bit_regs_p): Delete.
27235 (spe_expand_predicate_builtin): Delete.
27236 (spe_expand_evsel_builtin): Delete.
27237 (TARGET_DWARF_REGISTER_SPAN): Do not define.
27238 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
27239 (invalid_e500_subreg): Delete.
27240 (rs6000_legitimize_address): Always force_reg op2 as well, for
27241 paired single memory accesses.
27242 (rs6000_member_type_forces_blk): Delete.
27243 (rs6000_spe_function_arg): Delete.
27244 (rs6000_expand_unop_builtin): Delete SPE handling.
27245 (rs6000_expand_binop_builtin): Ditto.
27246 (spe_expand_stv_builtin): Delete.
27247 (bdesc_2arg_spe): Delete.
27248 (spe_expand_builtin): Delete.
27249 (spe_expand_predicate_builtin): Delete.
27250 (spe_expand_evsel_builtin): Delete.
27251 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
27252 (spe_init_builtins): Delete.
27253 (spe_func_has_64bit_regs_p): Delete.
27254 (savres_routine_name): Delete "info" parameter. Adjust callers.
27255 (rs6000_emit_stack_reset): Ditto.
27256 (rs6000_dwarf_register_span): Delete.
27257 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
27258 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
27259 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
27261 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
27263 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
27264 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
27265 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
27266 mulv2sf3, divv2sf3): Delete expanders.
27268 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27270 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
27272 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27274 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
27275 * config/rs6000/rs6000.c: Ditto.
27277 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27279 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
27280 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
27281 comparison_operator.
27283 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27285 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
27286 * config/rs6000/rs6000.opt: Ditto.
27287 * config/rs6000/t-rtems: Ditto.
27289 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27291 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
27292 TARGET_E500_SINGLE by 0, simplify.
27293 * config/rs6000/rs6000.c: Ditto.
27294 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
27295 (spe_build_register_parallel): Delete.
27296 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
27297 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
27298 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
27299 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
27300 (E500_CONVERT): Delete.
27301 * config/rs6000/spe.md: Remove many patterns and all define_constants.
27303 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27305 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
27306 * config/rs6000/dfp.md: Ditto.
27307 (negdd2, *negdd2_fpr): Merge.
27308 (absdd2, *absdd2_fpr): Merge.
27309 (negtd2, *negtd2_fpr): Merge.
27310 (abstd2, *abstd2_fpr): Merge.
27311 * config/rs6000/e500.h: Delete file.
27312 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
27313 TARGET_FPRS by 1 and simplify.
27314 * config/rs6000/rs6000-c.c: Ditto.
27315 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
27316 TARGET_DF_SPE by 0.
27317 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
27319 * config/rs6000/rs6000.md: Ditto.
27320 (floatdidf2, *floatdidf2_fpr): Merge.
27321 (move_from_CR_gt_bit): Delete.
27322 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
27323 (E500_CR_IOR_COMPARE): Delete.
27324 (All patterns that require !TARGET_FPRS): Delete.
27325 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
27327 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27329 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
27331 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27333 * graphds.c (add_edge): Intitialize edge's attached data.
27334 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
27335 pointer parameter. Call pointed function on each edge during
27336 graph traversing. Skip traversing the edge when the function
27338 (graphds_dfs, graphds_scc): Ditto.
27339 (for_each_edge): New parameter. Pass the new parameter to callback
27341 * graphds.h (skip_edge_callback): New function pointer type.
27342 (graphds_dfs, graphds_scc): New function pointer parameter.
27343 (graphds_edge_callback, for_each_edge): New parameter.
27345 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27347 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
27348 out code checking if runtime alias check is possible to below ...
27349 Call the new function.
27350 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
27351 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
27353 2017-06-07 Marek Polacek <polacek@redhat.com>
27356 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
27357 TYPE_OVERFLOW_WRAPS check.
27359 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27361 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
27362 if versioning is required.
27363 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
27364 peeling with the check for versioning.
27366 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27368 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
27369 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
27370 Set true to new parameter if new ssa variable is defined.
27371 (vect_gen_vector_loop_niters): Refactor. Set range information
27372 for the new vector loop bound variable.
27373 (vect_do_peeling): Ditto.
27375 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27377 * tree-affine.c (ssa.h): Include header file.
27378 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
27379 has wrapping overflow behavior.
27381 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27383 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
27385 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27387 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
27388 (tree_to_aff_combination): ... here.
27390 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27392 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
27393 reg_pressure model function.
27394 (ivopts_global_cost_for_size): Delete.
27395 (determine_set_costs, iv_ca_recount_cost): Call new model function
27396 ivopts_estimate_reg_pressure.
27398 2017-06-07 Tamar Christina <tamar.christina@arm.com>
27400 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
27401 expensive than udiv. Remove floating point cases from mod.
27403 2017-06-07 Tamar Christina <tamar.christina@arm.com>
27405 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
27406 Increase idiv cost.
27408 2017-06-07 Tamar Christina <tamar.christina@arm.com>
27410 * config/aarch64/aarch64.md
27411 (copysignsf3): Fix mask generation.
27413 2017-06-07 Jakub Jelinek <jakub@redhat.com>
27415 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
27417 (class dump_manager): Add register_dumps method.
27418 * dumpfile.c: Include langhooks.h.
27419 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
27420 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
27421 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
27422 (dump_manager::dump_register): Start with 512 entries instead of 32.
27423 (dump_manager::register_dumps): New method.
27424 * toplev.c (general_init): Instead of invoking register_dumps
27425 langhook, invoke register_dumps method on the dump manager.
27426 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
27429 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
27431 * doc/md.texi: Clarify the restrictions on a define_insn condition.
27432 Say that # requires an associated define_split to exist, and that
27433 the define_split must be suitable for use after register allocation.
27435 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27437 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
27438 (compute_outgoing_frequencies): Also initialize zero counts.
27439 (find_many_sub_basic_blocks): Do not produce uninitialized profile
27440 around loops; preserve more of profile when nothing changes.
27442 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
27444 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
27446 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
27447 * config/arm/arm-cpu-cdata.h: Regenerate.
27448 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
27449 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
27450 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
27451 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
27452 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
27454 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
27456 * config/arm/t-rmprofile: Likewise.
27457 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
27459 2017-06-06 David S. Miller <davem@davemloft.net>
27462 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
27463 blockage if function uses alloca.
27465 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27467 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
27468 New "uid" fields to hold pretty-print IDs of group and ref.
27469 Memory references are now identified as <group_id>:<ref_id>
27470 instead of using [random] addresses.
27471 (dump_mem_details): Simplify, no functional change.
27472 (dump_mem_ref): Simplify and make output more concise.
27473 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
27474 (find_or_create_group): Initialize group uid.
27475 (record_ref): Initialize ref uid. Improve debug output.
27476 (prune_group_by_reuse, should_issue_prefetch_p,)
27477 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
27478 (mark_nontemporal_store, determine_loop_nest_reuse):
27479 Improve debug output.
27481 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27483 * dbgcnt.def (prefetch): New debug counter.
27484 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
27485 (schedule_prefetches): Stop issueing prefetches if debug counter
27488 2017-06-06 Tom de Vries <tom@codesourcery.com>
27490 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
27491 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
27493 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27495 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
27496 Use aarch64_reg_or_zero predicate for operand 4.
27497 (aarch64_compare_and_swap<mode> define_insn_and_split):
27498 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
27499 (aarch64_store_exclusive<mode>): Likewise for operand 2.
27501 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
27503 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
27504 (arm_compute_save_core_reg_mask): This.
27505 (thumb1_compute_save_reg_mask): Rename into ...
27506 (thumb1_compute_save_core_reg_mask): This.
27507 (arm_compute_save_reg0_reg12_mask): Adapt comment.
27508 (arm_compute_frame_layout): Likewise.
27510 2017-06-06 Richard Biener <rguenther@suse.de>
27512 PR tree-optimization/80974
27513 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
27514 keep or clear leaders SSA info.
27516 2017-06-06 Tom de Vries <tom@codesourcery.com>
27518 * config/nvptx/nvptx.c (split_mode_p): New function.
27519 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
27521 2017-06-06 Tom de Vries <tom@codesourcery.com>
27523 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
27525 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27528 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
27531 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27533 * shrink-wrap.c (handle_simple_exit): Update profile.
27534 (try_shrink_wrapping): Upate profile.
27536 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27538 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
27539 (tree_guess_outgoing_edge_probabilities): New.
27540 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
27541 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
27543 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27545 * ipa-split.c (split_function): Initialize return bb profile.
27547 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27549 * profile.c (compute_branch_probabilities): Also initialize
27550 EXIT_BLOCK profile.
27552 2017-06-06 Richard Biener <rguenther@suse.de>
27554 PR tree-optimization/80928
27555 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
27556 (vect_analyze_loop_operations): Properly guard analysis for
27558 (vect_transform_loop): Likewise.
27559 (vect_analyze_loop_2): Also reset SLP type on PHIs.
27560 (vect_model_induction_cost): Do not cost for pure SLP.
27561 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
27562 of induction in inner loop vectorization.
27563 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
27564 (vect_get_and_check_slp_defs): Handle vect_induction_def.
27565 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
27567 (vect_analyze_slp_cost_1): Cost induction.
27568 (vect_detect_hybrid_slp_stmts): Handle PHIs.
27569 (vect_get_slp_vect_defs): Likewise.
27570 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
27571 (vect_transform_stmt): Handle SLP reductions.
27572 * tree-vectorizer.h (vectorizable_induction): Adjust.
27574 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
27576 * config/rs6000/rs6000.c (make_resolver_func): Update
27577 init_lowered_empty_function call.
27579 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
27581 * doc/invoke.texi: Document the -fprofile-abs-path option.
27582 * common.opt (fprofile-abs-path): New option.
27583 * gcov-io.h (gcov_write_filename): Declare.
27584 * gcov-io.c (gcov_write_filename): New function.
27585 * coverage.c (coverage_begin_function): Use gcov_write_filename.
27586 * profile.c (output_location): Likewise.
27588 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
27590 * shring-wrap.c: Revert accidental commit.
27592 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
27594 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
27596 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
27598 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
27600 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
27601 profile in callgraph edge.
27602 * profile-count.h (apply_probability): If THIS is 0, then result is 0
27603 (apply_scale): Likewise.
27604 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
27605 Also scale profile when inlining function with zero profile.
27606 (initialize_cfun): Update exit block profile even when it is zero.
27607 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
27608 when profile is read.
27610 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
27612 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
27613 (CLONE_*): New constants to define the processors we can generate
27614 code for with the target_clone attribute.
27615 (rs6000_clone_map): New array to identify which clone processors
27616 the current program is running on.
27617 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
27618 target_clone attribute.
27619 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
27620 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
27621 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
27622 (cpu_expand_builtin): Add support for target_clone attribute.
27623 (rs6000_valid_attribute_p): Allow "default" attribute.
27624 (get_decl_name): New debug function to simplify printing the
27625 current function name in debugging statements.
27626 (rs6000_clone_priority): New functions to support the target_clone
27627 attribute, and be able to generate code to switch between ISA 2.05
27628 through ISA 3.0 (power6 through power9).
27629 (rs6000_compare_version_priority): Likewise.
27630 (rs6000_get_function_versions_dispatcher): Likewise.
27631 (make_resolver_func): Likewise.
27632 (add_condition_to_bb): Likewise.
27633 (dispatch_function_versions): Likewise.
27634 (rs6000_generate_version_dispatcher_body): Likewise.
27635 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
27636 (fusion_gpr_load_p): Fix a spacing issue.
27637 * doc/extend.texi (Common Function Attributes): Document that the
27638 PowerPC supports the target_clone attribute.
27640 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
27642 * config/arm/arm.h: explain F symbol found in description of ARM
27643 register allocation in its legend.
27645 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
27647 * config/mips/frame-header-opt.c: Include profile-count.h.
27648 * config/riscv/riscv.c: Include profile-count.h
27650 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
27652 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
27654 (sm_set_flag_if_changed): Add bbs field.
27655 (execute_sm_if_changed_flag_set): Pass BBS.
27656 (execute_sm): Update.
27658 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27660 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
27663 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27665 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
27666 (peephole2): New peephole2 to emit the above.
27667 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
27669 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27671 * config/aarch64/aarch64.c (define_peephole2 above
27672 *sub_<shift>_<mode>): New peephole.
27674 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
27676 * config/i386/i386.c (make_resolver_func): Update.
27677 * Makefile.in: Add profile-count.h and profile-count.o
27678 * auto-profile.c (afdo_indirect_call): Update to new API.
27679 (afdo_set_bb_count): Update.
27680 (afdo_propagate_edge): Update.
27681 (afdo_propagate_circuit): Update.
27682 (afdo_calculate_branch_prob): Update.
27683 (afdo_annotate_cfg): Update.
27684 * basic-block.h: Include profile-count.h
27685 (struct edge_def): Turn count to profile_count.
27686 (struct basic_block_def): Likewie.
27687 (REG_BR_PROB_BASE): Move to profile-count.h
27688 (RDIV): Move to profile-count.h
27689 * bb-reorder.c (max_entry_count): Turn to profile_count.
27690 (find_traces): Update.
27691 (rotate_loop):Update.
27692 (connect_traces):Update.
27693 (sanitize_hot_paths):Update.
27694 * bt-load.c (migrate_btr_defs): Update.
27695 * cfg.c (RDIV): Remove.
27696 (init_flow): Use alloc_block.
27697 (alloc_block): Uninitialize count.
27698 (unchecked_make_edge): Uninitialize count.
27699 (check_bb_profile): Update.
27700 (dump_edge_info): Update.
27701 (dump_bb_info): Update.
27702 (update_bb_profile_for_threading): Update.
27703 (scale_bbs_frequencies_int): Update.
27704 (scale_bbs_frequencies_gcov_type): Update.
27705 (scale_bbs_frequencies_profile_count): New.
27706 * cfg.h (update_bb_profile_for_threading): Update.
27707 (scale_bbs_frequencies_profile_count): Declare.
27708 * cfgbuild.c (compute_outgoing_frequencies): Update.
27709 (find_many_sub_basic_blocks): Update.
27710 * cfgcleanup.c (try_forward_edges): Update.
27711 (try_crossjump_to_edge): Update.
27712 * cfgexpand.c (expand_gimple_tailcall): Update.
27713 (construct_exit_block): Update.
27714 * cfghooks.c (verify_flow_info): Update.
27715 (dump_bb_for_graph): Update.
27716 (split_edge): Update.
27717 (make_forwarder_block): Update.
27718 (duplicate_block): Update.
27719 (account_profile_record): Update.
27720 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
27721 (get_estimated_loop_iterations): Update.
27722 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
27723 (single_likely_exit): Update.
27724 * cfgloopmanip.c (scale_loop_profile): Update.
27726 (set_zero_probability): Update.
27727 (lv_adjust_loop_entry_edge): Update.
27728 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
27729 (purge_dead_edges): Update.
27730 (rtl_account_profile_record): Update.
27731 * cgraph.c (cgraph_node::create): Uninitialize count.
27732 (symbol_table::create_edge): Uninitialize count.
27733 (cgraph_update_edges_for_call_stmt_node): Update.
27734 (cgraph_edge::dump_edge_flags): Update.
27735 (cgraph_node::dump): Update.
27736 (cgraph_edge::maybe_hot_p): Update.
27737 * cgraph.h: Include profile-count.h
27738 (create_clone), create_edge, create_indirect_edge): Update.
27739 (cgraph_node): Turn count to profile_count.
27740 (cgraph_edge0: Likewise.
27741 (make_speculative, clone): Update.
27742 (create_edge): Update.
27743 (init_lowered_empty_function): Update.
27744 * cgraphclones.c (cgraph_edge::clone): Update.
27745 (duplicate_thunk_for_node): Update.
27746 (cgraph_node::create_clone): Update.
27747 * cgraphunit.c (cgraph_node::analyze): Update.
27748 (cgraph_node::expand_thunk): Update.
27749 * final.c (dump_basic_block_info): Update.
27750 * gimple-streamer-in.c (input_bb): Update.
27751 * gimple-streamer-out.c (output_bb): Update.
27752 * graphite.c (print_global_statistics): Update.
27753 (print_graphite_scop_statistics): Update.
27754 * hsa-brig.c: Include basic-block.h.
27755 * hsa-dump.c: Include basic-block.h.
27756 * hsa-gen.c (T sum_slice): Update.
27757 (convert_switch_statements):Update.
27758 * hsa-regalloc.c: Include basic-block.h.
27759 * ipa-chkp.c (chkp_produce_thunks): Update.
27760 * ipa-cp.c (struct caller_statistics): Update.
27761 (init_caller_stats): Update.
27762 (gather_caller_stats): Update.
27763 (ipcp_cloning_candidate_p): Update.
27764 (good_cloning_opportunity_p): Update.
27765 (get_info_about_necessary_edges): Update.
27766 (dump_profile_updates): Update.
27767 (update_profiling_info): Update.
27768 (update_specialized_profile): Update.
27769 (perhaps_add_new_callers): Update.
27770 (decide_about_value): Update.
27771 (ipa_cp_c_finalize): Update.
27772 * ipa-devirt.c (struct odr_type_warn_count): Update.
27773 (struct decl_warn_count): Update.
27774 (struct final_warning_record): Update.
27775 (possible_polymorphic_call_targets): Update.
27776 (ipa_devirt): Update.
27777 * ipa-fnsummary.c (redirect_to_unreachable): Update.
27778 * ipa-icf.c (sem_function::merge): Update.
27779 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
27780 * ipa-inline.c (compute_uninlined_call_time): Update.
27781 (compute_inlined_call_time): Update.
27782 (want_inline_small_function_p): Update.
27783 (want_inline_self_recursive_call_p): Update.
27784 (edge_badness): Update.
27785 (lookup_recursive_calls): Update.
27786 (recursive_inlining): Update.
27787 (inline_small_functions): Update.
27788 (dump_overall_stats): Update.
27789 (dump_inline_stats): Update.
27790 * ipa-profile.c (ipa_profile_generate_summary): Update.
27791 (ipa_propagate_frequency): Update.
27792 (ipa_profile): Update.
27793 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
27794 * ipa-utils.c (ipa_merge_profiles): Update.
27795 * loop-doloop.c (doloop_modify): Update.
27796 * loop-unroll.c (report_unroll): Update.
27797 (unroll_loop_runtime_iterations): Update.
27798 * lto-cgraph.c (lto_output_edge): Update.
27799 (lto_output_node): Update.
27800 (input_node): Update.
27801 (input_edge): Update.
27802 (merge_profile_summaries): Update.
27803 * lto-streamer-in.c (input_cfg): Update.
27804 * lto-streamer-out.c (output_cfg): Update.
27805 * mcf.c (create_fixup_graph): Update.
27806 (adjust_cfg_counts): Update.
27807 (sum_edge_counts): Update.
27808 * modulo-sched.c (sms_schedule): Update.
27809 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
27810 * predict.c (maybe_hot_count_p): Update.
27811 (probably_never_executed): Update.
27812 (dump_prediction): Update.
27813 (combine_predictions_for_bb): Update.
27814 (propagate_freq): Update.
27815 (handle_missing_profiles): Update.
27816 (counts_to_freqs): Update.
27817 (rebuild_frequencies): Update.
27818 (force_edge_cold): Update.
27819 * predict.h: Include profile-count.h
27820 (maybe_hot_count_p, counts_to_freqs): UPdate.
27821 * print-rtl-function.c: Do not include cfg.h
27822 * print-rtl.c: Include basic-block.h
27823 * profile-count.c: New file.
27824 * profile-count.h: New file.
27825 * profile.c (is_edge_inconsistent): Update.
27826 (correct_negative_edge_counts): Update.
27827 (is_inconsistent): Update.
27828 (set_bb_counts): Update.
27829 (read_profile_edge_counts): Update.
27830 (compute_frequency_overlap): Update.
27831 (compute_branch_probabilities): Update; Initialize and deinitialize
27833 (branch_prob): Update.
27834 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
27835 (edge_gcov_count): New.
27836 (bb_gcov_count): New.
27837 * shrink-wrap.c (try_shrink_wrapping): Update.
27838 * tracer.c (better_p): Update.
27839 * trans-mem.c (expand_transaction): Update.
27840 (ipa_tm_insert_irr_call): Update.
27841 (ipa_tm_insert_gettmclone_call): Update.
27842 * tree-call-cdce.c: Update.
27843 * tree-cfg.c (gimple_duplicate_sese_region): Update.
27844 (gimple_duplicate_sese_tail): Update.
27845 (gimple_account_profile_record): Update.
27846 (execute_fixup_cfg): Update.
27847 * tree-inline.c (copy_bb): Update.
27848 (copy_edges_for_bb): Update.
27849 (initialize_cfun): Update.
27850 (freqs_to_counts): Update.
27851 (copy_cfg_body): Update.
27852 (expand_call_inline): Update.
27853 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
27854 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
27855 (try_unroll_loop_completely): Update.
27856 (try_peel_loop): Update.
27857 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
27858 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
27859 * tree-ssa-loop-split.c (connect_loops): Update.
27860 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
27861 * tree-ssa-reassoc.c (branch_fixup): Update.
27862 * tree-ssa-tail-merge.c (replace_block_by): Update.
27863 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
27864 (compute_path_counts): Update.
27865 (update_profile): Update.
27866 (recompute_probabilities): Update.
27867 (update_joiner_offpath_counts): Update.
27868 (estimated_freqs_path): Update.
27869 (freqs_to_counts_path): Update.
27870 (clear_counts_path): Update.
27871 (ssa_fix_duplicate_block_edges): Update.
27872 (duplicate_thread_path): Update.
27873 * tree-switch-conversion.c (case_bit_test_cmp): Update.
27874 (struct switch_conv_info): Update.
27875 * tree-tailcall.c (decrease_profile): Update.
27876 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
27877 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
27878 * value-prof.c (check_counter): Update.
27879 (gimple_divmod_fixed_value): Update.
27880 (gimple_mod_pow2): Update.
27881 (gimple_mod_subtract): Update.
27882 (gimple_ic_transform): Update.
27883 (gimple_stringop_fixed_value): Update.
27884 * value-prof.h (gimple_ic): Update.
27886 2017-06-02 Carl Love <cel@us.ibm.com>
27888 * config/rs6000/rs6000-c: Add support for built-in functions
27889 vector double vec_doublee (vector signed int);
27890 vector double vec_doublee (vector unsigned int);
27891 vector double vec_doublee (vector float);
27892 vector double vec_doubleh (vector signed int);
27893 vector double vec_doubleh (vector unsigned int);
27894 vector double vec_doubleh (vector float);
27895 vector double vec_doublel (vector signed int);
27896 vector double vec_doublel (vector unsigned int);
27897 vector double vec_doublel (vector float);
27898 vector double vec_doubleo (vector signed int);
27899 vector double vec_doubleo (vector unsigned int);
27900 vector double vec_doubleo (vector float);.
27901 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
27902 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
27904 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
27905 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
27906 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
27908 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
27909 vec_doublel, vec_doubleh.
27910 * doc/extend.texi: Update the built-in documentation file for the
27911 new built-in functions.
27913 2017-06-02 David Malcolm <dmalcolm@redhat.com>
27916 * ipa-inline-analysis.c (free_growth_caches): Set
27917 edge_removal_hook_holder to NULL after removing it.
27919 2017-06-02 Sudakshina Das <sudi.das@arm.com>
27921 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
27922 comparision with zero.
27924 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
27925 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
27926 for early expansion of vec_min and vec_max builtins.
27927 (builtin_function_type): Add min/max unsigned variants to those
27928 identified as having unsigned arguments.
27930 2017-06-02 Olivier Hainque <hainque@adacore.com>
27932 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
27934 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27936 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
27937 Use VALL_F16 iterator rather than VALL.
27939 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27941 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
27942 Emit CBNZ inside loop when doing a strong exchange and comparing
27943 against zero. Generate the CC flags after the loop.
27945 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
27947 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
27948 (dl_section_ref): New.
27949 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
27950 On AIX, append an expression to subtract the size of the
27951 section length to dl_section_ref.
27953 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
27955 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
27956 for early expansion of vector absolute builtins.
27958 2017-06-02 Richard Biener <rguenther@suse.de>
27960 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
27961 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
27963 2017-06-02 Richard Biener <rguenther@suse.de>
27965 PR tree-optimization/80948
27966 * tree-tailcall.c (find_tail_calls): Track stmts to move in
27967 stmt order as well.
27969 2017-06-02 Richard Biener <rguenther@suse.de>
27971 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
27973 * tree-vect-stmts.c (process_use): Do not mark backedge defs
27974 for inductions as relevant.
27976 2017-06-02 Richard Biener <rguenther@suse.de>
27978 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
27979 (vectorizable_induction): ... this. Remove dead code.
27981 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
27983 * builtins. (expand_builtin_alloca): Remove second parameter and
27984 infer its value from the first parameter instead.
27985 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
27987 2017-06-02 Jakub Jelinek <jakub@redhat.com>
27989 PR rtl-optimization/80903
27990 * loop-doloop.c (add_test): Unshare sequence.
27992 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
27994 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
27996 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
27998 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
28000 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
28001 xlogue_layout::get_instance, logue_layout::xlogue_layout,
28002 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
28003 (xlogue_layout::get_stub_rtx): Make static.
28004 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
28005 (xlogue_layout::compute_stub_managed_regs): Rename to...
28006 (xlogue_layout::count_stub_managed_regs): ...this.
28007 (xlogue_layout::is_stub_managed_reg): New function.
28008 (xlogue_layout::m_stub_names): Rename to...
28009 (xlogue_layout::s_stub_names): ...this, make static.
28010 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
28011 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
28012 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
28013 xlogue_layout::s_stub_names): Instantiate statics.
28014 (stub_managed_regs): Remove.
28015 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
28016 (disable_call_ms2sysv_xlogues): Rename to...
28017 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
28018 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
28020 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
28021 change after reload_completed.
28022 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
28024 (ix86_expand_prologue): Likewise.
28025 (ix86_expand_epilogue): Likewise.
28026 (ix86_expand_split_stack_prologue): Likewise.
28027 (ix86_compute_frame_layout): Remove frame parameter ...
28028 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
28029 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
28031 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
28032 (ix86_frame): Move from here ...
28033 * config/i386/i386.h (ix86_frame): ... to here.
28034 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
28035 complete ix86_frame data structure instead. Remove some_ld_name.
28037 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
28039 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
28040 symbols that hold a DECL_VALUE_EXPR.
28042 2017-06-01 Martin Jambor <mjambor@suse.cz>
28044 PR tree-optimization/80898
28045 * tree-sra.c (process_subtree_disqualification): Removed.
28046 (disqualify_candidate): Do not acll
28047 process_subtree_disqualification.
28048 (subtree_mark_written_and_enqueue): New function.
28049 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
28050 RHS has been disqualified and re-queue LHS if necessary. Apart
28051 from that, ignore disqualified RHS.
28053 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28055 * config/s390/s390.c (s390_emit_epilogue): Disable early return
28056 address fetch for z10 or later.
28058 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28060 * config/arc/arc.md (tst_movb): Add guard when splitting.
28062 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28064 * config/arc/arc.c (arc_can_eliminate): Test against
28065 arc_frame_pointer_needed.
28067 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28069 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
28070 to prevent store reordering.
28071 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
28072 (type): Add block type.
28073 (stack_tie): Define special instruction to be used in
28076 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28078 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
28079 constraint. It is not valid for the pattern.
28080 (noncommutative_binary_comparison): Likewise.
28082 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28084 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
28087 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28089 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
28090 be used by the reg-alloc.
28092 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28094 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
28095 reg-alloc when having mul64 or mul32x16 instructions.
28096 (mulsidi3): Likewise.
28097 (umulsidi3): Likewise.
28098 (mulsi32x16): New pattern.
28099 (mulsi64): Likewise.
28100 (mulsidi64): Likewise.
28101 (umulsidi64): Likewise.
28102 (MUL32x16_REG): Define.
28103 (mul64_600): Use MUL32x16_REG.
28104 (mac64_600): Likewise.
28105 (umul64_600): Likewise.
28106 (umac64_600): Likewise.
28108 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28110 * config/arc/arc.md (mulsi3_700): Make it commutative.
28112 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
28114 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
28116 (*sign_extendsidi2_insn): Likewise for movstosw.
28118 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
28120 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
28121 the type of the input discriminant value. Convert the
28122 discriminant value of signedness vary.
28124 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
28126 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
28127 Add to -Wall section.
28129 2017-06-01 Richard Biener <rguenther@suse.de>
28131 PR middle-end/66313
28132 * fold-const.c (fold_plusminus_mult_expr): If the factored
28133 factor may be zero use a wrapping type for the inner operation.
28134 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
28135 and handle moved defs.
28136 (process_assignment): Properly guard the unary op case. Return a
28137 tri-state indicating that moving the stmt before the call may allow
28138 to continue. Pass through to_move.
28139 (find_tail_calls): Handle moving unrelated defs before
28142 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
28145 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
28146 splitter result in the canonical way.
28148 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
28150 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
28151 also for 32bit target. Update insn attributes.
28152 (zero-extendsidi2 splitter): Allow all registers for operand 1.
28154 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
28156 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
28157 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
28158 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
28159 (_mm_maskz_min_ss): New intrinsics.
28161 2017-05-31 Martin Liska <mliska@suse.cz>
28163 * tree-vect-loop.c (vect_create_epilog_for_reduction):
28164 Change comment style to one we normally use.
28165 (vectorizable_reduction): Likewise.
28166 (vectorizable_induction): Likewise.
28167 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
28168 (vectorizable_call): Likewise.
28169 (vectorizable_simd_clone_call): Likewise.
28170 (vectorizable_conversion): Likewise.
28171 (vectorizable_assignment): Likewise.
28172 (vectorizable_shift): Likewise.
28173 (vectorizable_operation): Likewise.
28174 (vectorizable_store): Likewise.
28175 (vectorizable_load): Likewise.
28176 * tree-vectorizer.h: Likewise.
28178 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
28180 * passes.c (emergency_dump_function): New.
28181 * tree-pass.h (emergency_dump_function): Declare.
28182 * plugin.c (plugins_internal_error_function): Remove.
28183 * plugin.h (plugins_internal_error_function): Remove declaration.
28184 * toplev.c (internal_error_function): New static function. Use it...
28185 (general_init): ...here.
28187 2017-05-31 Graham Markall <graham.markall@embecosm.com>
28189 * config/arc/arc.c (arc_print_operand): Handle constant operands.
28190 (arc_rtx_costs): Add costs for new patterns.
28191 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
28192 * config/arc/predicates.md: Add _1_2_3_operand predicate.
28194 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
28196 * tree-ssa-strlen.c (get_next_strinfo): New function.
28197 (get_stridx_plus_constant): Use it.
28198 (zero_length_string): Likewise.
28199 (adjust_related_strinfos): Likewise.
28200 (adjust_last_stmt): Likewise.
28202 2017-05-31 Richard Biener <rguenther@suse.de>
28205 * config/i386/i386.c (ix86_expand_builtin): Remove assert
28206 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
28208 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
28210 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
28211 loop_vinfo argument and use of dependence distance vectors.
28212 Check instead whether the two references differ only in their
28213 initial value and assume that they have the same alignment if the
28214 difference is a multiple of the vector alignment.
28215 (vect_analyze_data_refs_alignment): Update call accordingly.
28217 2017-05-31 Martin Liska <mliska@suse.cz>
28220 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
28222 2017-05-31 Bin Cheng <bin.cheng@arm.com>
28224 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
28225 (create_intersect_range_checks): Move from ...
28226 * tree-data-ref.c (create_intersect_range_checks_index)
28227 (create_intersect_range_checks): ... to here.
28228 (create_runtime_alias_checks): New function factored from ...
28229 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
28230 here. Call above function.
28231 * tree-data-ref.h (create_runtime_alias_checks): New function.
28233 2017-05-31 Bin Cheng <bin.cheng@arm.com>
28235 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
28236 segment length for dr_b and compute it in wide_int.
28238 2017-05-31 Richard Biener <rguenther@suse.de>
28240 PR tree-optimization/80906
28241 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
28242 and pass through iv_map.
28243 (copy_bb_and_scalar_dependences): Adjust.
28244 (translate_pending_phi_nodes): Likewise.
28245 (copy_loop_close_phi_args): Handle code-generating IVs instead
28248 2017-05-30 David Malcolm <dmalcolm@redhat.com>
28250 * diagnostic-color.c (color_dict): Add "type-diff".
28251 (parse_gcc_colors): Update comment.
28252 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
28253 -fdiagnostics-show-template-tree and -fno-elide-type.
28254 (GCC_COLORS): Add type-diff to example.
28256 (-fdiagnostics-show-template-tree): New.
28257 (-fno-elide-type): New.
28258 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
28259 the pp_format_decoder callback. Call any m_format_postprocessor's
28261 (pretty_printer::pretty_printer): Initialize
28262 m_format_postprocessor.
28263 (pretty_printer::~pretty_printer): Delete any
28264 m_format_postprocessor.
28265 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
28266 (class format_postprocessor): New class.
28267 (struct pretty_printer::format_decoder): Document the new parameters.
28268 (struct pretty_printer::m_format_postprocessor): New field.
28269 * tree-diagnostic.c (default_tree_printer): Update for new
28270 bool and const char ** params.
28271 * tree-diagnostic.h (default_tree_printer): Likewise.
28273 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
28275 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
28276 (lwa_operand): Delete rs6000_gen_cell_microcode test.
28277 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
28278 rs6000_gen_cell_microcode code.
28279 (rs6000_final_prescan_insn): Delete.
28280 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
28281 "warn-cell-microcode" entries.
28282 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
28283 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
28284 throughout. Change cc_reg_not_micro_cr0_operand to
28285 cc_reg_not_cr0_operand throughout.
28286 (*extendhi<mode>2_noload): Delete.
28287 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
28288 (mwarn-cell-microcode): Delete.
28289 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
28290 -mgen-cell-microcode and -mwarn-cell-microcode.
28292 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
28295 * config/i386/constraints.md (Yd): New constraint.
28297 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
28298 and (?Yd, r) alternatives. Update insn attributes.
28299 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
28300 and (?*Yd, r) alternatives. Update insn attributes.
28301 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
28303 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
28305 * gimplify.c (gimplify_modify_expr): Don't create a
28306 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
28309 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
28311 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
28313 2017-05-30 Richard Biener <rguenther@suse.de>
28315 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
28316 and reduc_def fields.
28317 (STMT_VINFO_REDUC_TYPE): New define.
28318 (STMT_VINFO_REDUC_DEF): Likewise.
28319 (vect_force_simple_reduction): Adjust prototype.
28320 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
28321 (vect_is_simple_reduction): Remove check_reduction argument.
28322 (vect_force_simple_reduction): Adjust and set
28323 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
28324 (vectorizable_reduction): Do not re-do reduction analysis
28325 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
28326 * tree-parloops.c (gather_scalar_reductions): Adjust.
28328 2017-05-30 Richard Biener <rguenther@suse.de>
28330 PR middle-end/80901
28331 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
28334 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
28336 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
28337 Introduce unknown_misalignment parameter and remove vf.
28338 (vect_peeling_hash_get_lowest_cost):
28339 Pass unknown_misalignment parameter.
28340 (vect_enhance_data_refs_alignment):
28341 Fix unsupportable data ref treatment.
28343 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
28345 * tree-vect-data-refs.c (vect_get_data_access_cost):
28346 Workaround for SLP handling.
28347 (vect_enhance_data_refs_alignment):
28348 Compute costs for doing no peeling at all, compare to the best
28349 peeling costs so far and avoid peeling if cheaper.
28351 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
28353 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
28354 Return peeling info and set costs to zero for unlimited cost
28356 (vect_enhance_data_refs_alignment): Also inspect all datarefs
28357 with unknown misalignment. Compute and costs for unknown
28358 misalignment, compare them to the costs for known misalignment
28359 and choose the cheapest for peeling.
28361 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
28363 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
28364 (vect_get_peeling_costs_all_drs): Create function.
28365 (vect_peeling_hash_get_lowest_cost):
28366 Use vect_get_peeling_costs_all_drs.
28367 (vect_peeling_supportable): Create function.
28368 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
28370 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
28372 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
28373 DR_HAS_NEGATIVE_STEP.
28374 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
28375 (vect_enhance_data_refs_alignment): Use.
28376 (vect_duplicate_ssa_name_ptr_info): Use.
28377 * tree-vectorizer.h (dr_misalignment): Use.
28378 (known_alignment_for_access_p): Use.
28380 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
28383 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
28385 (has_section_name): New function.
28387 2017-05-30 Martin Liska <mliska@suse.cz>
28390 * auto-profile.c (get_function_decl_from_block): Fix
28393 2017-05-30 Richard Biener <rguenther@suse.de>
28395 PR middle-end/80876
28396 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
28398 2017-05-30 Martin Liska <mliska@suse.cz>
28400 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
28401 * dumpfile.h (struct dump_file_info): Remove ctors.
28403 2017-05-30 Martin Liska <mliska@suse.cz>
28405 * predict.def: Fix GNU coding style.
28407 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
28409 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
28410 Mark 'to' argument with ATTRIBUTE_UNUSED.
28412 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
28414 * config/xtensa/xtensa.c (xtensa_emit_call): Use
28415 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
28416 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
28419 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
28421 * doc/install.texi (Options specification): Restore entry of
28422 --enable-sjlj-exceptions.
28424 2017-05-27 Michael Eager <eager@eagercon.com>
28427 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
28429 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
28431 * config/microblaze/microblaze.h
28432 (FIXED_REGISTERS): Update in macro.
28433 (CALL_USED_REGISTERS): Update in macro.
28435 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
28437 * doc/install.texi: Add links to macOS binary distributions.
28439 2017-05-27 Jakub Jelinek <jakub@redhat.com>
28443 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
28445 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
28447 2017-05-26 Martin Liska <mliska@suse.cz>
28449 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
28451 2017-05-26 Martin Liska <mliska@suse.cz>
28453 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
28454 always leading ';; '.
28455 (dump_bb_info): Likewise.
28456 (brief_dump_cfg): Likewise.
28457 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
28458 * dumpfile.c: Remove usage of TDF_VERBOSE.
28459 * dumpfile.h (enum dump_kind): Likewise.
28460 (dump_gimple_bb_header): Do not use TDF_COMMENT.
28461 * print-tree.c (debug_verbose): Remove.
28462 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
28463 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
28464 * tree-diagnostic.c (default_tree_printer): Replace
28465 TDF_DIAGNOSTIC with TDF_SLIM.
28467 2017-05-26 Bin Cheng <bin.cheng@arm.com>
28469 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
28470 in parameter loop, rather than loop_vinfo.
28471 (create_intersect_range_checks): Ditto.
28472 (vect_create_cond_for_alias_checks): Update call to above functions.
28474 2017-05-26 Bin Cheng <bin.cheng@arm.com>
28476 PR tree-optimization/80815
28477 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
28478 for merging runtime alias checks. Handle negative DR_STEPs.
28480 2017-05-26 Bin Cheng <bin.cheng@arm.com>
28482 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
28484 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
28485 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
28486 out code pruning runtime alias checks.
28487 * tree-data-ref.c (prune_runtime_alias_test_list): New function
28488 factored out from above.
28489 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
28491 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
28493 (prune_runtime_alias_test_list): New decalaration.
28495 2017-05-26 Bin Cheng <bin.cheng@arm.com>
28497 * tree-vect-data-refs.c (compare_tree): Rename and move ...
28498 * tree-data-ref.c (data_ref_compare_tree): ... to here.
28499 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
28500 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
28501 (operator==, comp_dr_with_seg_len_pair): Ditto.
28502 (vect_prune_runtime_alias_test_list): Ditto.
28504 2017-05-26 Martin Liska <mliska@suse.cz>
28507 * params.def: Bound partial-inlining-entry-probability param.
28509 2017-05-26 Marek Polacek <polacek@redhat.com>
28512 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
28515 2017-05-26 Richard Biener <rguenther@suse.de>
28517 PR tree-optimization/80842
28518 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
28521 2017-05-26 Richard Biener <rguenther@suse.de>
28523 PR tree-optimization/80844
28524 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
28526 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
28528 * doc/md.texi (Machine Constraints): Update x86 family
28529 machine constraints section to match 'config/i386/constraints.md'.
28531 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
28533 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
28535 2017-05-25 Nathan Sidwell <nathan@acm.org>
28537 * doc/invoke.texi (--enable-languages): Update documentation.
28539 2017-05-25 Martin Liska <mliska@suse.cz>
28541 * dumpfile.c: Add TDF_FOLDING.
28542 * dumpfile.h (enum dump_kind): Likewise.
28543 * genmatch.c (dt_simplify::gen_1): Use it.
28545 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
28547 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
28549 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
28551 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
28552 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
28554 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
28556 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
28557 * match.pd (X == C): Rewrite it here.
28558 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
28559 with_certain_nonzero_bits2): New predicates.
28560 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
28562 2017-05-24 Nathan Sidwell <nathan@acm.org>
28564 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
28567 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
28570 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
28572 * config/powerpcspe: New port. Files are copied from the rs6000
28573 port, with "rs6000" in filenames replaced by "powerpcspe".
28575 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
28577 PR rtl-optimization/80754
28578 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
28580 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
28582 * config/sparc/sparc.md (length): Return the correct value for -mflat
28583 sibcalls to match output_sibcall.
28585 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
28589 * config/rs6000/rs6000.c (struct machine_function): Add new field
28591 (rs6000_get_separate_components): Init that field, use it.
28592 (rs6000_components_for_bb): Use the field.
28594 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
28596 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
28598 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
28600 PR middle-end/80823
28601 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
28603 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28606 * config/s390/s390.c (s390_check_qrst_address): Check incoming
28607 address against address_operand predicate.
28608 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
28610 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
28612 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
28613 parameters passed indirectly.
28615 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
28617 * config/i386/i386.md (*movdi_internal): Remove SSE4
28618 alternative 18 (?r, *v). Update insn attributes.
28619 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
28620 Update insn attributes.
28621 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
28622 Update insn attributes.
28623 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
28624 alternative 1 (r, v). Remove isa attribute.
28625 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
28626 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
28627 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
28629 2017-05-23 Tom de Vries <tom@codesourcery.com>
28631 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
28634 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
28636 * cgraphunit.c (symbol_table::process_new_functions): Update.
28637 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
28638 (inline_generate_summary): Rename to ...
28639 (ipa_fn_summary_generate): ... this one.
28640 (inline_read_summary): Rename to ...
28641 (ipa_fn_summary_read): ... this one.
28642 (inline_write_summary): Rename to ...
28643 (ipa_fn_summary_write): ... this one.
28644 (inline_free_summary): Rename to ...
28645 (ipa_free_fn_summary): ... this one.
28646 (pass_data_local_fn_summary, pass_local_fn_summary,
28647 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
28648 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
28649 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
28650 make_pass_ipa_fn_summary): New.
28651 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
28652 inline_write_summary, inline_free_summary): Remove.
28653 (ipa_free_fn_summary) : New.
28654 * ipa-inline.c (ipa_inline): Update.
28655 (pass_ipa_inline): Do not generate summaries.
28656 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
28658 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
28659 and add pass_ipa_fn_summary.
28660 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
28662 (make_pass_inline_parameters): Remove.
28664 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
28666 * omp-low.c (struct omp_context): Remove "default_kind" member.
28669 * omp-offload.c (execute_oacc_device_lower): Remove the
28670 parallelism dimensions function attributes for unparallelized
28671 OpenACC kernels constructs.
28673 2017-05-23 Martin Liska <mliska@suse.cz>
28675 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
28677 (cgraph_edge::make_speculative): Likewise.
28678 (cgraph_edge::resolve_speculation): Likewise.
28679 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
28680 (cgraph_node::dump): Likewise.
28681 * cgraph.h: Likewise.
28682 * cgraphunit.c (analyze_functions): Likewise.
28683 (symbol_table::compile): Likewise.
28684 * ipa-cp.c (print_all_lattices): Likewise.
28685 (determine_versionability): Likewise.
28686 (initialize_node_lattices): Likewise.
28687 (ipcp_verify_propagated_values): Likewise.
28688 (estimate_local_effects): Likewise.
28689 (update_profiling_info): Likewise.
28690 (create_specialized_node): Likewise.
28691 (perhaps_add_new_callers): Likewise.
28692 (decide_about_value): Likewise.
28693 (decide_whether_version_node): Likewise.
28694 (identify_dead_nodes): Likewise.
28695 (ipcp_store_bits_results): Likewise.
28696 * ipa-devirt.c (dump_targets): Likewise.
28697 (ipa_devirt): Likewise.
28698 * ipa-icf.c (sem_item::dump): Likewise.
28699 (sem_function::equals): Likewise.
28700 (sem_variable::equals): Likewise.
28701 (sem_item_optimizer::read_section): Likewise.
28702 (sem_item_optimizer::execute): Likewise.
28703 (congruence_class::dump): Likewise.
28704 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
28705 (dump_inline_summary): Likewise.
28706 (estimate_node_size_and_time): Likewise.
28707 (inline_analyze_function): Likewise.
28708 * ipa-inline-transform.c (inline_call): Likewise.
28709 * ipa-inline.c (report_inline_failed_reason): Likewise.
28710 (want_early_inline_function_p): Likewise.
28711 (edge_badness): Likewise.
28712 (update_edge_key): Likewise.
28713 (inline_small_functions): Likewise.
28714 * ipa-profile.c (ipa_profile): Likewise.
28715 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
28716 (ipa_make_edge_direct_to_target): Likewise.
28717 (remove_described_reference): Likewise.
28718 (ipa_impossible_devirt_target): Likewise.
28719 (propagate_controlled_uses): Likewise.
28720 (ipa_print_node_params): Likewise.
28721 (ipcp_transform_function): Likewise.
28722 * ipa-pure-const.c (pure_const_read_summary): Likewise.
28723 (propagate_pure_const): Likewise.
28724 * ipa-reference.c (generate_summary): Likewise.
28725 (read_write_all_from_decl): Likewise.
28726 (propagate): Likewise.
28727 (ipa_reference_read_optimization_summary): Likewise.
28728 * ipa-utils.c (ipa_merge_profiles): Likewise.
28729 * ipa.c (walk_polymorphic_call_targets): Likewise.
28730 (symbol_table::remove_unreachable_nodes): Likewise.
28731 (ipa_single_use): Likewise.
28732 * passes.c (execute_todo): Likewise.
28733 * predict.c (drop_profile): Likewise.
28734 * symtab.c (symtab_node::get_dump_name): New function.
28735 (symtab_node::dump_name): Likewise.
28736 (symtab_node::dump_asm_name): Likewise.
28737 (symtab_node::dump_references): Likewise.
28738 (symtab_node::dump_referring): Likewise.
28739 (symtab_node::dump_base): Likewise.
28740 (symtab_node::debug_symtab): Likewise.
28741 * tree-sra.c (convert_callers_for_node): Likewise.
28742 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
28743 * value-prof.c (init_node_map): Likewise.
28745 2017-05-23 Martin Liska <mliska@suse.cz>
28747 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
28748 and symtab_node::debug_symtab to symbol_table::debug.
28749 * cgraphunit.c (analyze_functions): Use the renamed function.
28750 (symbol_table::compile): Likewise.
28751 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
28752 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
28753 * passes.c (execute_todo): Likewise.
28754 * symtab.c (symbol_table::dump): New function.
28755 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
28757 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
28759 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
28760 that nonconst implies exec.
28762 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
28764 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
28765 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
28766 (inline_edge_summary_vec): Turn into ...
28767 (ipa_call_summaries): ... this one.
28768 (redirect_to_unreachable, edge_set_predicate,
28769 evaluate_properties_for_edge, inline_summary_alloc,
28770 reset_ipa_call_summary, reset_inline_summary,
28771 inline_summary_t::duplicate): Update.
28772 (inline_edge_duplication_hook): Turn to ...
28773 (ipa_call_summary_t::duplicate): ... this one.
28774 (inline_edge_removal_hook): Turn to ...
28775 (ipa_call_summary_t::remove): ... this one.
28776 (dump_inline_edge_summary): Turn to ...
28777 (dump_ipa_call_summary): ... this one.
28778 (estimate_function_body_sizes): Update.
28779 (inline_update_callee_summaries): Update.
28780 (remap_edge_change_prob): Update.
28781 (remap_edge_summaries): Update.
28782 (inline_merge_summary): Update.
28783 (do_estimate_edge_time): Update.
28784 (inline_generate_summary): Update.
28785 (inline_read_section): Update.
28786 (inline_read_summary): Update.
28787 (inline_free_summary): Update.
28788 * ipa-inline.c (can_inline_edge_p): Update.
28789 (compute_inlined_call_time): Update.
28790 (want_inline_small_function_p): Update.
28791 (edge_badness): Update.
28792 (early_inliner): Update.
28793 * ipa-inline.h (inline_edge_summary): Turn to ...
28794 (ipa_call_summary): ... this one.
28795 (ipa_call_summary_t): New class.
28796 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
28797 (ipa_call_summaries): New.
28798 (inline_edge_summary): Remove.
28799 (estimate_edge_growth): Update.
28800 * ipa-profile.c (ipa_propagate_frequency_1): Update.
28801 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
28802 * ipa-split.c (execute_split_functions): Update.
28803 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
28805 2017-05-23 Tom de Vries <tom@codesourcery.com>
28807 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
28808 attributes): Document rdrand effective target.
28810 2017-05-23 Tom de Vries <tom@codesourcery.com>
28812 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
28813 attributes): Sort alphabetically.
28815 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
28817 * config/avr/genmultilib.awk: Use gsub instead of gensub.
28819 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
28822 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
28823 V2DF/V2DI splat into two separate patterns, one that handles
28824 registers, and the other that only handles memory. Drop support
28825 for splatting from a GPR on ISA 2.07 and then splitting the
28826 splat into direct move and splat.
28827 (vsx_splat_<mode>_reg): Likewise.
28828 (vsx_splat_<mode>_mem): Likewise.
28830 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
28832 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
28834 2017-05-22 Jakub Jelinek <jakub@redhat.com>
28836 PR middle-end/80809
28837 * omp-low.c (finish_taskreg_remap): New function.
28838 (finish_taskreg_scan): If unit size of ctx->record_type
28839 is non-constant, unshare the size expression and replace
28840 decls in it with possible outer var refs.
28842 PR middle-end/80809
28843 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
28844 GOVD_SHARED rather than GOVD_PRIVATE with it.
28845 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
28846 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
28848 PR middle-end/80853
28849 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
28850 as last argument to build_outer_var_ref for pointer bases of array
28851 section reductions.
28853 2017-05-19 Martin Sebor <msebor@redhat.com>
28855 * print-tree.c (print_node): Print DECL_READ_P flag.
28857 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
28859 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
28860 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
28861 * cgraph.c: Likewise.
28862 * cgraphunit.c: Likewise.
28863 * gengtype.c: Likewise.
28864 * ipa-cp.c: Likewise.
28865 * ipa-devirt.c: Likewise.
28866 * ipa-icf.c: Likewise.
28867 * ipa-predicate.c: Likewise.
28868 * ipa-profile.c: Likewise.
28869 * ipa-prop.c: Likewise.
28870 * ipa-split.c: Likewise.
28872 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
28873 edge_predicate_pool, dump_inline_hints,
28874 inline_summary::account_size_time, redirect_to_unreachable,
28875 edge_set_predicate, set_hint_predicate,
28876 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
28877 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
28878 inline_summary_t::remove, remap_hint_predicate_after_duplication,
28879 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
28880 ipa_call_summary_t::remove, initialize_growth_caches,
28881 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
28882 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
28883 mark_modified, unmodified_parm_1, unmodified_parm,
28884 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
28885 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
28886 compute_bb_predicates, will_be_nonconstant_expr_predicate,
28887 will_be_nonconstant_predicate, record_modified_bb_info,
28888 get_minimal_bb, record_modified, param_change_prob,
28889 phi_result_unknown_predicate, predicate_for_phi_result,
28890 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
28891 estimate_function_body_sizes, compute_inline_parameters,
28892 compute_inline_parameters_for_curren, pass_data_inline_parameters,
28893 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
28894 inline_update_callee_summaries, remap_edge_change_prob,
28895 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
28896 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
28897 inline_analyze_function, inline_summary_t::insert,
28898 inline_generate_summary, read_ipa_call_summary, inline_read_section,
28899 inline_read_summary, write_ipa_call_summary, inline_write_summary,
28900 inline_free_summary): Move to ipa-fnsummary.h
28901 (predicate_t): Remove.
28902 * ipa-fnsummary.c: New file.
28903 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
28904 (enum inline_hints_vals, inline_hints, agg_position_info,
28905 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
28906 inline_summaries, ipa_call_summary, ipa_call_summary_t,
28907 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
28908 dump_inline_summary, dump_inline_hints, inline_generate_summary,
28909 inline_read_summary, inline_write_summary, inline_free_summary,
28910 inline_analyze_function, initialize_inline_failed,
28911 inline_merge_summary, inline_update_overall_summary,
28912 compute_inline_parameters): Move to ipa-fnsummary.h
28913 * ipa-fnsummary.h: New file.
28914 * ipa-inline-transform.h: Include ipa-inline.h.
28915 * ipa-inline.c: LIkewise.
28917 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
28919 * ipa-inline.c (edge_badness): Use inlined_time instead of
28920 inline_summaries->get.
28922 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
28924 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
28926 2017-05-22 Nathan Sidwell <nathan@acm.org>
28928 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
28929 (fdump-lang): Document 'raw' option.
28930 * dumpfile.h (TDI_tu): Delete.
28931 * dumpfile.c (dump_files): Remove translation-unit.
28932 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
28934 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
28936 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
28937 command option from $(AWK) call.
28938 * config/avr/genmultilib.awk: Simplify and rewrite so that it
28939 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
28940 [FORMAT]: Remove handling of variable.
28941 * config/avr/t-multilib: Regenerate.
28943 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
28945 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
28947 (dump_inline_summary): Do not print self_time.
28948 (estimate_function_body_sizes): Do not set self_time.
28949 (compute_inline_parameters): Likewise.
28950 (inline_read_section, inline_write_summary): Do not stream self_time.
28951 * ipa-inline.h (inline_summary): Drop self_time.
28953 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
28955 * ipa-inline-analysis.c (account_size_time): Rename to ...
28956 (inline_summary::account_size_time): ... this one.
28957 (reset_ipa_call_summary): Turn to ...
28958 (ipa_call_summary::reset): ... this one.
28959 (reset_inline_summary): Turn to ...
28960 (inline_summary::reset): ... this one.
28961 (inline_summary_t::remove): Update.
28962 (inline_summary_t::duplicate): Update.
28963 (ipa_call_summary_t::remove): Update.
28964 (dump_inline_summary): Update.
28965 (estimate_function_body_sizes): Update.
28966 (compute_inline_parameters): Update.
28967 (estimate_node_size_and_time): Update.
28968 (inline_merge_summary): Update.
28969 (inline_update_overall_summary): Update.
28970 (inline_read_section): Update.
28971 (inline_write_summary): Update.
28972 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
28973 add account_size_time and reset member functions.
28974 (ipa_call_summary): Add reset function.
28975 * ipa-predicate.h (predicate::operator &): Constify.
28977 2017-05-22 Richard Biener <rguenther@suse.de>
28979 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
28981 2017-05-19 Jason Merrill <jason@redhat.com>
28983 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
28985 2017-05-19 Marek Polacek <polacek@redhat.com>
28988 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
28989 TYPE_OVERFLOW_WRAPS checks.
28991 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
28993 * tree-core.h (enum omp_clause_default_kind): Add
28994 "OMP_CLAUSE_DEFAULT_PRESENT".
28995 * tree-pretty-print.c (dump_omp_clause): Handle it.
28996 * gimplify.c (enum gimplify_omp_var_data): Add
28997 "GOVD_MAP_FORCE_PRESENT".
28998 (gimplify_adjust_omp_clauses_1): Map it to
28999 "GOMP_MAP_FORCE_PRESENT".
29000 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
29002 * gimplify.c (oacc_default_clause): Clarify.
29004 2017-05-19 Nathan Sidwell <nathan@acm.org>
29006 LANG_HOOK_REGISTER_DUMPS
29007 * toplev.c (general_init): Call register dump lang hook.
29008 * doc/invoke.texi: Document -fdump-lang option family.
29009 * dumpfile.c (dump_files): Remove class dump here.
29010 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
29011 * dumpfile.h (tree_dump_index): Remove TDI_class.
29012 * langhooks-def.h (lhd_register_dumps): Declare.
29013 (LANG_HOOKS_REGISTER_DUMPS): Define.
29014 (LANG_HOOKS_INITIALIZER): Add it.
29015 * langhooks.c (lhd_register_dumps): Define.
29016 * langhooks.h (struct lang_hooks): Add register_dumps.
29018 2017-05-19 Nathan Sidwell <nathan@acm.org>
29020 * context.h (context::set_passes): New.
29021 * context.c (context::context): Do not create pass manager.
29022 * toplev.c (general_init): Create pass manager here.
29024 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
29026 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
29027 use this splitter if two add or or instructions would also work for
29028 the constant we want to generate.
29030 2017-05-19 Richard Biener <rguenther@suse.de>
29033 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
29034 predicate evaluation.
29036 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
29038 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
29040 * ipa-inline.c (want_inline_small_function_p): Do not cast to
29043 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
29045 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
29046 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
29047 (inline_edge_summary_vec): Turn into ...
29048 (ipa_call_summaries): ... this one.
29049 (redirect_to_unreachable, edge_set_predicate,
29050 evaluate_properties_for_edge, inline_summary_alloc,
29051 reset_ipa_call_summary, reset_inline_summary,
29052 inline_summary_t::duplicate): Update.
29053 (inline_edge_duplication_hook): Turn to ...
29054 (ipa_call_summary_t::duplicate): ... this one.
29055 (inline_edge_removal_hook): Turn to ...
29056 (ipa_call_summary_t::remove): ... this one.
29057 (dump_inline_edge_summary): Turn to ...
29058 (dump_ipa_call_summary): ... this one.
29059 (estimate_function_body_sizes): Update.
29060 (inline_update_callee_summaries): Update.
29061 (remap_edge_change_prob): Update.
29062 (remap_edge_summaries): Update.
29063 (inline_merge_summary): Update.
29064 (do_estimate_edge_time): Update.
29065 (inline_generate_summary): Update.
29066 (inline_read_section): Update.
29067 (inline_read_summary): Update.
29068 (inline_free_summary): Update.
29069 * ipa-inline.c (can_inline_edge_p): Update.
29070 (compute_inlined_call_time): Update.
29071 (want_inline_small_function_p): Update.
29072 (edge_badness): Update.
29073 (early_inliner): Update.
29074 * ipa-inline.h (inline_edge_summary): Turn to ...
29075 (ipa_call_summary): ... this one.
29076 (ipa_call_summary_t): New class.
29077 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
29078 (ipa_call_summaries): New.
29079 (inline_edge_summary): Remove.
29080 (estimate_edge_growth): Update.
29081 * ipa-profile.c (ipa_propagate_frequency_1): Update.
29082 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
29083 * ipa-split.c (execute_split_functions): Update.
29084 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
29086 2017-05-19 Richard Biener <rguenther@suse.de>
29088 PR middle-end/80764
29089 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
29091 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
29093 * config/rs6000/rs6000.c (struct machine_function): Add field
29094 fpr_is_wrapped_separately.
29095 (rs6000_get_separate_components): Use 64 components. Handle the
29096 new FPR components.
29097 (rs6000_components_for_bb): Handle the FPR components.
29098 (rs6000_emit_prologue_components): Handle the FPR components.
29099 (rs6000_emit_epilogue_components): Handle the FPR components.
29100 (rs6000_set_handled_components): Handle the FPR components.
29101 (rs6000_emit_prologue): Don't output prologue code for those FPRs
29102 that are already separately shrink-wrapped.
29103 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
29104 that are already separately shrink-wrapped.
29106 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
29109 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
29112 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
29113 (define_peephole2 for Altivec d-form load): Add peepholes to catch
29114 cases where the register allocator uses a move and an offsettable
29115 memory operation to/from a FPR register on ISA 2.06/2.07.
29116 (define_peephole2 for Altivec d-form store): Likewise.
29118 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
29121 * config/i386/mmx.md (*mov<mode>_internal): Enable
29122 alternatives 11, 12, 13 and 14 also for 32bit targets.
29123 Remove alternatives 15, 16, 17 and 18.
29124 * config/i386/sse.md (vec_concatv2di): Change
29125 alternative (!x, *y) to (x, ?!*Yn).
29127 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
29129 * dumpfile.h (enum dump_kind): Remove stray comma.
29131 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
29133 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
29134 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
29135 predicate::num_conditions
29136 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
29137 (CHANGED): turn into predicate::changed.
29138 (agg_position_info): Move to ipa-predicate.h
29139 (add_condition, predicate::add_clause, predicate::operator &=,
29140 predicate::or_with, predicate::evaluate, predicate::probability,
29141 dump_condition, dump_clause, predicate::dump,
29142 predicate::remap_after_duplication, predicate::remap_after_inlining,
29143 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
29144 (evaluate_conditions_for_known_args): Update.
29145 (set_cond_stmt_execution_predicate): Update.
29146 * ipa-inline.h: Include ipa-predicate.h
29147 (condition, inline_param_summary, conditions, agg_position_info,
29148 predicate): Move to ipa-predicate.h
29149 * ipa-predicate.c: New file.
29150 * ipa-predicate.h: New file.
29152 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
29154 * final.c (leaf_function_p): Check we are not in a sequence.
29156 2017-05-18 Martin Liska <mliska@suse.cz>
29158 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
29159 * dumpfile.c (dump_register): Use new enum dump_kind.
29160 (get_dump_file_name): Likewise.
29161 (dump_enable_all): Likewise.
29162 (dump_switch_p_1): Likewise.
29163 (enable_rtl_dump_file): Remove usage of TDF_RTL.
29164 * dumpfile.h (enum dump_kind): New enum type.
29165 (struct dump_file_info): Create constructor and
29166 format fields and comments.
29167 * passes.c (pass_manager::register_one_dump_file):
29169 * statistics.c (statistics_early_init): Likewise.
29170 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
29171 TDF_TREE with TDF_SLIM.
29172 (gather_memory_references_ref): Likewise.
29174 2017-05-18 Martin Liska <mliska@suse.cz>
29176 * vec.h (struct vnull): Use it.
29178 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
29180 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
29181 (true_predicate, false_predicate, true_predicate_p,
29182 false_predicate_p): Remove.
29183 (single_cond_predicate, not_inlined_predicate): Turn to member function
29185 (add_condition): Update.
29186 (add_clause): Turn to...
29187 (predicate::add_clause): ... this one; update; allow passing NULL
29189 (and_predicates): Turn to ...
29190 (predicate::operator &=): ... this one.
29191 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
29192 (or_predicates): Turn to ...
29193 (predicate::or_with): ... this one.
29194 (evaluate_predicate): Turn to ...
29195 (predicate::evaluate): ... this one.
29196 (predicate_probability): Turn to ...
29197 (predicate::probability): ... this one.
29198 (dump_condition): Update.
29199 (dump_predicate): Turn to ...
29200 (predicate::dump): ... this one.
29201 (account_size_time): Update.
29202 (edge_set_predicate): Update.
29203 (set_hint_predicate): UPdate.
29204 (evaluate_conditions_for_known_args): Update.
29205 (evaluate_properties_for_edge): Update.
29206 (remap_predicate_after_duplication): Turn to...
29207 (predicate::remap_after_duplication): ... this one.
29208 (remap_hint_predicate_after_duplication): Update.
29209 (inline_summary_t::duplicate): UPdate.
29210 (dump_inline_edge_summary): Update.
29211 (dump_inline_summary): Update.
29212 (set_cond_stmt_execution_predicate): Update.
29213 (set_switch_stmt_execution_predicate): Update.
29214 (compute_bb_predicates): Update.
29215 (will_be_nonconstant_expr_predicate): Update.
29216 (will_be_nonconstant_predicate): Update.
29217 (phi_result_unknown_predicate): Update.
29218 (predicate_for_phi_result): Update.
29219 (array_index_predicate): Update.
29220 (estimate_function_body_sizes): Update.
29221 (estimate_node_size_and_time): Update.
29222 (estimate_ipcp_clone_size_and_time): Update.
29223 (remap_predicate): Rename to ...
29224 (predicate::remap_after_inlining): ... this one.
29225 (remap_hint_predicate): Update.
29226 (inline_merge_summary): Update.
29227 (inline_update_overall_summary): Update.
29228 (estimate_size_after_inlining): Update.
29229 (read_predicate): Rename to ...
29230 (predicate::stream_in): ... this one.
29231 (read_inline_edge_summary): Update.
29232 (write_predicate): Rename to ...
29233 (predicate::stream_out): ... this one.
29234 (write_inline_edge_summary): Update.
29235 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
29236 (clause_t): Turn to uint32_t
29237 (predicate): Turn to class; implement constructor and operators
29239 (size_time_entry): Update.
29240 (inline_summary): Update.
29241 (inline_edge_summary): Update.
29243 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
29245 * fold-const.c (fold_binary_loc): Move transformation...
29246 * match.pd (C - X CMP X): ... here.
29248 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
29250 * config/sparc/sparc.c (sparc_option_override): Set function
29251 alignment for -mcpu=niagara7 to 64 to match the I$ line.
29252 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
29254 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
29256 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
29258 2017-05-18 Marek Polacek <polacek@redhat.com>
29261 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
29262 (pass_ubsan::execute): Call gimple_assign_single_p instead of
29263 gimple_assign_load_p.
29265 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
29267 PR middle-end/80692
29268 * real.c (do_compare): Give decimal_do_compare preference over
29269 comparing just the signs.
29271 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
29273 * doc/md.texi (Canonicalization of Instructions): Describe the
29274 canonical form of instructions that inherently set a condition
29277 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
29279 PR middle-end/80775
29280 * tree-cfg.c: Move deletion of unreachable case statements to after
29281 the merging of consecutive case labels.
29283 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
29285 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
29286 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
29287 restoring of callee-saved registers.
29289 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
29291 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
29292 * config/visium/visium.c (single_set_and_flags): Likewise.
29293 * config/visium/visium.md (Substitutions): Likewise.
29295 2017-05-17 Martin Liska <mliska@suse.cz>
29297 * cfg.c: Introduce dump_flags_t type and
29298 use it instead of int type.
29300 * cfghooks.c: Likewise.
29301 * cfghooks.h (struct cfg_hooks): Likewise.
29302 * cfgrtl.c: Likewise.
29303 * cfgrtl.h: Likewise.
29304 * cgraph.c (cgraph_node::get_body): Likewise.
29305 * coretypes.h: Likewise.
29306 * domwalk.c: Likewise.
29307 * domwalk.h: Likewise.
29308 * dumpfile.c (struct dump_option_value_info): Likewise.
29309 (dump_enable_all): Likewise.
29310 (dump_switch_p_1): Likewise.
29311 (opt_info_switch_p): Likewise.
29312 * dumpfile.h (enum tree_dump_index): Likewise.
29313 (struct dump_file_info): Likewise.
29314 * genemit.c: Likewise.
29315 * generic-match-head.c: Likewise.
29316 * gengtype.c (open_base_files): Likewise.
29317 * gimple-pretty-print.c: Likewise.
29318 * gimple-pretty-print.h: Likewise.
29319 * graph.c (print_graph_cfg): Likewise.
29320 * graphite-scop-detection.c (dot_all_sese): Likewise.
29321 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
29322 * loop-unroll.c (report_unroll): Likewise.
29323 * passes.c (pass_manager::register_one_dump_file): Likewise.
29324 * print-tree.c: Likewise.
29325 * statistics.c: Likewise.
29326 * tree-cfg.c: Likewise.
29327 * tree-cfg.h: Likewise.
29328 * tree-dfa.c: Likewise.
29329 * tree-dfa.h: Likewise.
29330 * tree-dump.c (dump_function): Likewise.
29331 * tree-dump.h (struct dump_info): Likewise.
29332 * tree-pretty-print.c: Likewise.
29333 * tree-pretty-print.h: Likewise.
29334 * tree-ssa-live.c: Likewise.
29335 * tree-ssa-live.h: Likewise.
29336 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
29337 * tree-vect-loop.c: Likewise.
29338 * tree-vect-slp.c: Likewise.
29340 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
29341 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29343 PR tree-optimization/80457
29344 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
29345 of all arguments to a statement as scalar_to_vec operations.
29346 (vectorizable_call): Adjust call to vect_model_simple_cost for
29348 (vectorizable_conversion): Likewise.
29349 (vectorizable_assignment): Likewise.
29350 (vectorizable_shift): Likewise.
29351 (vectorizable_operation): Likewise.
29352 (vectorizable_comparison): Likewise.
29353 (vect_is_simple_cond): Record the def types for operands.
29354 (vectorizable_condition): Likewise, call vect_model_simple_cost.
29355 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
29356 for statement argument count.
29358 2017-05-16 Carl Love <cel@us.ibm.com>
29360 * config/rs6000/rs6000-c: Add support for built-in functions
29361 vector unsigned long long vec_bperm (vector unsigned long long,
29362 vector unsigned char)
29363 vector signed long long vec_mule (vector signed int,
29365 vector unsigned long long vec_mule (vector unsigned int,
29366 vector unsigned int)
29367 vector signed long long vec_mulo (vector signed int,
29369 vector unsigned long long vec_mulo (vector unsigned int,
29370 vector unsigned int)
29371 vector signed char vec_sldw (vector signed char,
29372 vector signed char,
29374 vector unsigned char vec_sldw (vector unsigned char,
29375 vector unsigned char,
29377 vector signed short vec_sldw (vector signed short,
29378 vector signed short,
29380 vector unsigned short vec_sldw (vector unsigned short,
29381 vector unsigned short,
29383 vector signed int vec_sldw (vector signed int,
29386 vector unsigned int vec_sldw (vector unsigned int,
29387 vector unsigned int,
29389 vector signed long long vec_sldw (vector signed long long,
29390 vector signed long long,
29392 vector unsigned long long vec_sldw (vector unsigned long long,
29393 vector unsigned long long,
29395 * config/rs6000/rs6000-c: Add support for built-in functions
29396 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
29397 * config/rs6000/altivec.h: Add defintion for vec_sldw.
29398 * doc/extend.texi: Update the built-in documentation for the
29399 new built-in functions.
29401 2017-05-16 Marek Polacek <polacek@redhat.com>
29405 * tree.c (save_expr): Don't fold the expression.
29407 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
29409 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
29410 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
29411 and (?*y,m). Update insn attributes.
29413 2017-05-16 Martin Liska <mliska@suse.cz>
29415 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
29416 flags argument of print_gimple_stmt, print_gimple_expr,
29417 print_generic_stmt and print_generic_expr.
29418 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
29419 * coretypes.h: Likewise.
29420 * except.c (dump_eh_tree): Likewise.
29421 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
29422 * gimple-pretty-print.h: Likewise.
29423 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
29424 (backprop::push_to_worklist): Likewise.
29425 (backprop::pop_from_worklist): Likewise.
29426 (backprop::process_use): Likewise.
29427 (backprop::intersect_uses): Likewise.
29428 (note_replacement): Likewise.
29429 * gimple-ssa-store-merging.c
29430 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
29431 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
29432 (pass_store_merging::execute): Likewise.
29433 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
29434 (ssa_base_cand_dump_callback): Likewise.
29435 (dump_incr_vec): Likewise.
29436 (replace_refs): Likewise.
29437 (replace_mult_candidate): Likewise.
29438 (create_add_on_incoming_edge): Likewise.
29439 (create_phi_basis): Likewise.
29440 (insert_initializers): Likewise.
29441 (all_phi_incrs_profitable): Likewise.
29442 (introduce_cast_before_cand): Likewise.
29443 (replace_one_candidate): Likewise.
29444 * gimplify.c (gimplify_expr): Likewise.
29445 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
29446 (set_rename): Likewise.
29447 (rename_uses): Likewise.
29448 (copy_loop_phi_nodes): Likewise.
29449 (add_close_phis_to_merge_points): Likewise.
29450 (copy_loop_close_phi_args): Likewise.
29451 (copy_cond_phi_args): Likewise.
29452 (graphite_copy_stmts_from_block): Likewise.
29453 (translate_pending_phi_nodes): Likewise.
29454 * graphite-poly.c (print_pdr): Likewise.
29455 (dump_gbb_cases): Likewise.
29456 (dump_gbb_conditions): Likewise.
29457 (print_scop_params): Likewise.
29458 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
29459 (build_cross_bb_scalars_use): Likewise.
29460 (gather_bbs::before_dom_children): Likewise.
29461 * hsa-dump.c (dump_hsa_immed): Likewise.
29462 * ipa-cp.c (print_ipcp_constant_value): Likewise.
29463 (get_replacement_map): Likewise.
29464 * ipa-inline-analysis.c (dump_condition): Likewise.
29465 (estimate_function_body_sizes): Likewise.
29466 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
29467 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
29468 * ipa-prop.c (ipa_dump_param): Likewise.
29469 (ipa_print_node_jump_functions_for_edge): Likewise.
29470 (ipa_modify_call_arguments): Likewise.
29471 (ipa_modify_expr): Likewise.
29472 (ipa_dump_param_adjustments): Likewise.
29473 (ipa_dump_agg_replacement_values): Likewise.
29474 (ipcp_modif_dom_walker::before_dom_children): Likewise.
29475 * ipa-pure-const.c (check_stmt): Likewise.
29476 (pass_nothrow::execute): Likewise.
29477 * ipa-split.c (execute_split_functions): Likewise.
29478 * omp-offload.c (dump_oacc_loop_part): Likewise.
29479 (dump_oacc_loop): Likewise.
29480 * trans-mem.c (tm_log_emit): Likewise.
29481 (tm_memopt_accumulate_memops): Likewise.
29482 (dump_tm_memopt_set): Likewise.
29483 (dump_tm_memopt_transform): Likewise.
29484 * tree-cfg.c (gimple_verify_flow_info): Likewise.
29485 (print_loop): Likewise.
29486 * tree-chkp-opt.c (chkp_print_addr): Likewise.
29487 (chkp_gather_checks_info): Likewise.
29488 (chkp_get_check_result): Likewise.
29489 (chkp_remove_check_if_pass): Likewise.
29490 (chkp_use_outer_bounds_if_possible): Likewise.
29491 (chkp_reduce_bounds_lifetime): Likewise.
29492 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
29493 (chkp_mark_completed_bounds): Likewise.
29494 (chkp_register_incomplete_bounds): Likewise.
29495 (chkp_mark_invalid_bounds): Likewise.
29496 (chkp_maybe_copy_and_register_bounds): Likewise.
29497 (chkp_build_returned_bound): Likewise.
29498 (chkp_get_bound_for_parm): Likewise.
29499 (chkp_build_bndldx): Likewise.
29500 (chkp_get_bounds_by_definition): Likewise.
29501 (chkp_generate_extern_var_bounds): Likewise.
29502 (chkp_get_bounds_for_decl_addr): Likewise.
29503 * tree-chrec.c (chrec_apply): Likewise.
29504 * tree-data-ref.c (dump_data_reference): Likewise.
29505 (dump_subscript): Likewise.
29506 (dump_data_dependence_relation): Likewise.
29507 (analyze_overlapping_iterations): Likewise.
29508 * tree-inline.c (expand_call_inline): Likewise.
29509 (tree_function_versioning): Likewise.
29510 * tree-into-ssa.c (dump_defs_stack): Likewise.
29511 (dump_currdefs): Likewise.
29512 (dump_names_replaced_by): Likewise.
29513 (dump_update_ssa): Likewise.
29514 (update_ssa): Likewise.
29515 * tree-object-size.c (pass_object_sizes::execute): Likewise.
29516 * tree-parloops.c (build_new_reduction): Likewise.
29517 (try_create_reduction_list): Likewise.
29518 (ref_conflicts_with_region): Likewise.
29519 (oacc_entry_exit_ok_1): Likewise.
29520 (oacc_entry_exit_single_gang): Likewise.
29521 * tree-pretty-print.h: Likewise.
29522 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
29523 (get_scalar_evolution): Likewise.
29524 (add_to_evolution): Likewise.
29525 (get_loop_exit_condition): Likewise.
29526 (analyze_evolution_in_loop): Likewise.
29527 (analyze_initial_condition): Likewise.
29528 (analyze_scalar_evolution): Likewise.
29529 (instantiate_scev): Likewise.
29530 (number_of_latch_executions): Likewise.
29531 (gather_chrec_stats): Likewise.
29532 (final_value_replacement_loop): Likewise.
29533 (scev_const_prop): Likewise.
29534 * tree-sra.c (dump_access): Likewise.
29535 (disqualify_candidate): Likewise.
29536 (create_access): Likewise.
29537 (reject): Likewise.
29538 (maybe_add_sra_candidate): Likewise.
29539 (create_access_replacement): Likewise.
29540 (analyze_access_subtree): Likewise.
29541 (analyze_all_variable_accesses): Likewise.
29542 (sra_modify_assign): Likewise.
29543 (initialize_constant_pool_replacements): Likewise.
29544 (find_param_candidates): Likewise.
29545 (decide_one_param_reduction): Likewise.
29546 (replace_removed_params_ssa_names): Likewise.
29547 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
29548 * tree-ssa-copy.c (dump_copy_of): Likewise.
29549 (copy_prop_visit_cond_stmt): Likewise.
29550 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
29551 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
29552 (record_equivalences_from_stmt): Likewise.
29553 * tree-ssa-dse.c (compute_trims): Likewise.
29554 (delete_dead_call): Likewise.
29555 (delete_dead_assignment): Likewise.
29556 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
29557 (forward_propagate_into_cond): Likewise.
29558 (pass_forwprop::execute): Likewise.
29559 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
29560 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
29562 (move_computations_worker): Likewise.
29563 (execute_sm): Likewise.
29564 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
29565 (remove_exits_and_undefined_stmts): Likewise.
29566 (remove_redundant_iv_tests): Likewise.
29567 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
29568 (adjust_iv_update_pos): Likewise.
29569 * tree-ssa-math-opts.c (bswap_replace): Likewise.
29570 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
29571 (value_replacement): Likewise.
29572 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
29573 * tree-ssa-pre.c (print_pre_expr): Likewise.
29574 (get_representative_for): Likewise.
29575 (create_expression_by_pieces): Likewise.
29576 (insert_into_preds_of_block): Likewise.
29577 (eliminate_insert): Likewise.
29578 (eliminate_dom_walker::before_dom_children): Likewise.
29579 (eliminate): Likewise.
29580 (remove_dead_inserted_code): Likewise.
29581 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
29582 * tree-ssa-reassoc.c (get_rank): Likewise.
29583 (eliminate_duplicate_pair): Likewise.
29584 (eliminate_plus_minus_pair): Likewise.
29585 (eliminate_not_pairs): Likewise.
29586 (undistribute_ops_list): Likewise.
29587 (eliminate_redundant_comparison): Likewise.
29588 (update_range_test): Likewise.
29589 (optimize_range_tests_var_bound): Likewise.
29590 (optimize_vec_cond_expr): Likewise.
29591 (rewrite_expr_tree): Likewise.
29592 (rewrite_expr_tree_parallel): Likewise.
29593 (linearize_expr): Likewise.
29594 (break_up_subtract): Likewise.
29595 (linearize_expr_tree): Likewise.
29596 (attempt_builtin_powi): Likewise.
29597 (attempt_builtin_copysign): Likewise.
29598 (transform_stmt_to_copy): Likewise.
29599 (transform_stmt_to_multiply): Likewise.
29600 (dump_ops_vector): Likewise.
29601 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
29602 (print_scc): Likewise.
29603 (set_ssa_val_to): Likewise.
29604 (visit_reference_op_store): Likewise.
29605 (visit_use): Likewise.
29606 (sccvn_dom_walker::before_dom_children): Likewise.
29607 (run_scc_vn): Likewise.
29608 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
29610 (expr_hash_elt::print): Likewise.
29611 (const_and_copies::pop_to_marker): Likewise.
29612 (const_and_copies::record_const_or_copy_raw): Likewise.
29613 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
29614 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
29615 (dump_predicates): Likewise.
29616 (find_uninit_use): Likewise.
29617 (warn_uninitialized_phi): Likewise.
29618 (pass_late_warn_uninitialized::execute): Likewise.
29619 * tree-ssa.c (verify_vssa): Likewise.
29620 (verify_ssa): Likewise.
29621 (maybe_optimize_var): Likewise.
29622 * tree-vrp.c (dump_value_range): Likewise.
29623 (dump_all_value_ranges): Likewise.
29624 (dump_asserts_for): Likewise.
29625 (register_edge_assert_for_2): Likewise.
29626 (vrp_visit_cond_stmt): Likewise.
29627 (vrp_visit_switch_stmt): Likewise.
29628 (vrp_visit_stmt): Likewise.
29629 (vrp_visit_phi_node): Likewise.
29630 (simplify_cond_using_ranges_1): Likewise.
29631 (fold_predicate_in): Likewise.
29632 (evrp_dom_walker::before_dom_children): Likewise.
29633 (evrp_dom_walker::push_value_range): Likewise.
29634 (evrp_dom_walker::pop_value_range): Likewise.
29635 (execute_early_vrp): Likewise.
29637 2017-05-16 Richard Biener <rguenther@suse.de>
29639 * dwarf2out.c (loc_list_from_tree_1): Do not create
29640 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
29642 2017-05-16 Richard Biener <rguenther@suse.de>
29644 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
29646 (note_variable_value_in_expr): If we resolved the decl ref
29647 do not push to the stack.
29649 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
29651 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
29652 operations in fast-math mode.
29653 (vaddq_f16): Likewise.
29654 (vmul_f16): Likewise.
29655 (vmulq_f16): Likewise.
29656 (vsub_f16): Likewise.
29657 (vsubq_f16): Likewise.
29658 * config/arm/neon.md (add<mode>3): New.
29660 (fma:<VH:mode>3): New. Also remove outdated comment.
29663 2017-05-16 Martin Liska <mliska@suse.cz>
29667 * ipa-devirt.c (warn_types_mismatch): Fix typo.
29668 (odr_types_equivalent_p): Likewise.
29670 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
29672 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
29674 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
29677 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
29678 non-interunit SSE move alternatives with '?'.
29679 (zero-extendsidi peephole2): New peephole to skip intermediate
29680 general register in SSE zero-extend sequence.
29682 2017-05-15 Jeff Law <law@redhat.com>
29684 * reorg.c (relax_delay_slots): Create a new variable to hold
29685 the temporary target rather than clobbering TARGET_LABEL.
29687 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
29688 missing argument to extract_bit_field call.
29689 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
29691 2017-05-15 Martin Liska <mliska@suse.cz>
29694 * gcc.c (process_command): Do not allow empty argument of -o option.
29696 2017-05-15 Renlin Li <renlin.li@arm.com>
29698 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
29699 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
29700 * config/aarch64/constraints.md (Usf): Add long call check.
29701 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
29702 (call_value): Likewise.
29703 (sibcall): Likewise.
29704 (sibcall_value): Likewise.
29706 (call_value_insn): New.
29707 (sibcall_insn): Update rtx pattern.
29708 (sibcall_value_insn): Likewise.
29709 (call_internal): Remove.
29710 (call_value_internal): Likewise.
29711 (sibcall_internal): Likewise.
29712 (sibcall_value_internal): Likewise.
29713 (call_reg): Likewise.
29714 (call_symbol): Likewise.
29715 (call_value_reg): Likewise.
29716 (call_value_symbol): Likewise.
29718 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
29721 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
29723 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
29725 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
29726 compatible with CCGOCmode and with CCZmode.
29728 2017-05-14 Martin Sebor <msebor@redhat.com>
29730 PR middle-end/77671
29731 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
29732 (gimple_fold_builtin_snprintf): Same.
29733 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
29734 (gimple_fold_builtin_snprintf): Same.
29735 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
29736 of character types.
29737 (is_call_safe): New function.
29738 (try_substitute_return_value): Call it.
29739 (try_simplify_call): New function.
29740 (pass_sprintf_length::handle_gimple_call): Call it.
29742 2017-05-14 Martin Sebor <msebor@redhat.com>
29744 PR middle-end/80669
29745 * builtins.c (expand_builtin_stpncpy): Simplify.
29747 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
29749 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
29750 * config/i386/i386.h
29751 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
29752 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
29753 (struct machine_function): Add new members call_ms2sysv,
29754 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
29755 (struct machine_frame_state): New fields sp_realigned and
29756 sp_realigned_offset.
29757 * config/i386/i386.c
29758 (enum xlogue_stub): New enum.
29759 (enum xlogue_stub_sets): New enum.
29760 (class xlogue_layout): New class.
29761 (struct ix86_frame): New fields stack_realign_allocate_offset,
29762 stack_realign_offset and outlined_save_offset. Modify comments to
29763 detail stack layout when using out-of-line stubs.
29764 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
29765 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
29766 -mcall-ms2sysv-xlogues.
29767 (stub_managed_regs): New static variable.
29768 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
29769 registers managed by out-of-line stub.
29770 (disable_call_ms2sysv_xlogues): New function.
29771 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
29772 m->call_ms2sysv when appropriate and compute frame layout for
29774 (sp_valid_at, fp_valid_at): New inline functions.
29775 (choose_basereg): New function.
29776 (choose_baseaddr): Add align parameter, use choose_basereg and modify
29778 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
29779 Use align parameter of choose_baseaddr to generated aligned SSE movs
29781 (pro_epilogue_adjust_stack): Modify to track
29782 machine_frame_state::sp_realigned.
29783 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
29784 (ix86_nsaved_sseregs): Likewise.
29785 (ix86_emit_save_regs): Likewise.
29786 (ix86_emit_save_regs_using_mov): Likewise.
29787 (ix86_emit_save_sse_regs_using_mov): Likewise.
29788 (get_scratch_register_on_entry): Likewise.
29789 (gen_frame_set): New function.
29790 (gen_frame_load): Likewise.
29791 (gen_frame_store): Likewise.
29792 (emit_outlined_ms2sysv_save): Likewise.
29793 (emit_outlined_ms2sysv_restore): Likewise.
29794 (ix86_expand_prologue): Modify stack re-alignment code and call
29795 emit_outlined_ms2sysv_save when appropriate.
29796 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
29797 parameter rtx_insn *insn, which allows the function to be used to only
29798 generate the notes.
29799 (ix86_expand_epilogue): Modify validity checks of frame and stack
29800 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
29801 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
29802 * config/i386/predicates.md
29803 (save_multiple): New predicate.
29804 (restore_multiple): Likewise.
29805 * config/i386/sse.md
29806 (save_multiple<mode>): New pattern.
29807 (save_multiple_realign<mode>): Likewise.
29808 (restore_multiple<mode>): Likewise.
29809 (restore_multiple_and_return<mode>): Likewise.
29810 (restore_multiple_leave_return<mode>): Likewise.
29811 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
29813 2017-05-14 Julia Koval <julia.koval@intel.com>
29815 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
29816 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
29817 (__builtin_ia32_xsetbv): New builtins.
29818 * config/i386/i386.c (ix86_expand_special_args_builtin):
29820 (ix86_expand_builtin): Special expand for new intrinsics.
29821 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
29822 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
29823 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
29825 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29827 * cfganal.c (inverted_post_order_compute): Change argument type
29829 * cfganal.h (inverted_post_order_compute): Adjust prototype.
29830 * df-core.c (rest_of_handle_df_initialize): Adjust.
29831 (rest_of_handle_df_finish): Likewise.
29832 (df_analyze_1): Likewise.
29833 (df_analyze): Likewise.
29834 (loop_inverted_post_order_compute): Change argument to be a vec *.
29835 (df_analyze_loop): Adjust.
29836 (df_get_n_blocks): Likewise.
29837 (df_get_postorder): Likewise.
29838 * df.h (struct df_d): Change field to be a vec.
29839 * lcm.c (compute_laterin): Adjust.
29840 (compute_available): Likewise.
29841 * lra-lives.c (lra_create_live_ranges_1): Likewise.
29842 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
29843 * tree-ssa-pre.c (compute_antic): Likewise.
29845 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29847 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
29848 (depth_first_search::depth_first_search): Change structure init
29849 function to this constructor.
29850 (depth_first_search::add_bb): Rename function to this member.
29851 (depth_first_search::execute): Likewise.
29852 (flow_dfs_compute_reverse_finish): Adjust.
29854 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29856 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
29857 (longest_simple_path): Likewise.
29858 * shrink-wrap.c (spread_components): Likewise.
29859 (disqualify_problematic_components): Likewise.
29860 (emit_common_heads_for_components): Likewise.
29861 (emit_common_tails_for_components): Likewise.
29862 (insert_prologue_epilogue_for_components): Likewise.
29864 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29866 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
29869 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29871 * df-core.c (df_set_blocks): Start using auto_bitmap.
29872 (df_compact_blocks): Likewise.
29873 * df-problems.c (df_rd_confluence_n): Likewise.
29874 * df-scan.c (df_insn_rescan_all): Likewise.
29875 (df_process_deferred_rescans): Likewise.
29876 (df_update_entry_block_defs): Likewise.
29877 (df_update_exit_block_uses): Likewise.
29878 (df_entry_block_bitmap_verify): Likewise.
29879 (df_exit_block_bitmap_verify): Likewise.
29880 (df_scan_verify): Likewise.
29881 * lra-constraints.c (lra_constraints): Likewise.
29882 (undo_optional_reloads): Likewise.
29883 (lra_undo_inheritance): Likewise.
29884 * lra-remat.c (calculate_gen_cands): Likewise.
29885 (do_remat): Likewise.
29886 * lra-spills.c (assign_spill_hard_regs): Likewise.
29887 (spill_pseudos): Likewise.
29888 * tree-ssa-pre.c (bitmap_set_and): Likewise.
29889 (bitmap_set_subtract_values): Likewise.
29891 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29893 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
29894 management with auto_bitmap.
29895 (fix_inter_tick): Likewise.
29896 (fix_recovery_deps): Likewise.
29897 * ira.c (add_store_equivs): Likewise.
29898 (find_moveable_pseudos): Likewise.
29899 (split_live_ranges_for_shrink_wrap): Likewise.
29900 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
29901 (rtx_reuse_manager::seen_def_p): Likewise.
29902 (rtx_reuse_manager::set_seen_def): Likewise.
29903 * print-rtl.h (class rtx_reuse_manager): Likewise.
29905 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29907 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
29909 (migrate_btr_def): Likewise.
29910 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
29911 * df-core.c (loop_post_order_compute): Likewise.
29912 (loop_inverted_post_order_compute): Likewise.
29913 * hsa-common.h: Likewise.
29914 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
29915 * init-regs.c (initialize_uninitialized_regs): Likewise.
29916 * ipa-inline.c (resolve_noninline_speculation): Likewise.
29917 (inline_small_functions): Likewise.
29918 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
29919 * ira.c (combine_and_move_insns): Likewise.
29920 (build_insn_chain): Likewise.
29921 * loop-invariant.c (find_invariants): Likewise.
29922 * lower-subreg.c (propagate_pseudo_copies): Likewise.
29923 * predict.c (tree_predict_by_opcode): Likewise.
29924 (predict_paths_leading_to): Likewise.
29925 (predict_paths_leading_to_edge): Likewise.
29926 (estimate_loops_at_level): Likewise.
29927 (estimate_loops): Likewise.
29928 * shrink-wrap.c (try_shrink_wrapping): Likewise.
29929 (spread_components): Likewise.
29930 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
29931 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
29932 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
29933 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
29934 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
29935 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
29936 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
29937 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
29938 (mark_threaded_blocks): Likewise.
29939 (thread_through_all_blocks): Likewise.
29940 * tree-ssa.c (verify_ssa): Likewise.
29941 (execute_update_addresses_taken): Likewise.
29942 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
29944 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29946 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
29948 (post_order_compute): Likewise.
29949 (inverted_post_order_compute): Likewise.
29950 (pre_and_rev_post_order_compute_fn): Likewise.
29952 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29954 * genrecog.c (int_set::int_set): Explicitly construct our
29955 auto_vec base class.
29956 * vec.h (auto_vec::auto_vec): New constructor.
29958 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29960 * bitmap.h (class auto_bitmap): New constructor taking
29961 bitmap_obstack * argument.
29963 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29965 * bitmap.h (class auto_bitmap): Change type of m_bits to
29966 bitmap_head, and adjust ctor / dtor and member operators.
29968 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
29970 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
29971 when returned register mode doesn't match original mode.
29973 2017-05-12 Jeff Law <law@redhat.com>
29974 Jakub Jelinek <jakub@redhat.com>
29976 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
29977 we look for cc setter after the compare-elim changes.
29978 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
29979 within the vector to match what compare-elim now expects.
29980 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
29981 (xorsi3_flags, one_cmplsi2_flags): Likewise.
29983 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
29984 after the compare-elim changes.
29985 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
29986 the vector to match what compare-elim now expects.
29987 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
29988 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
29989 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
29990 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
29991 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
29993 * config/visium/visium.c (single_set_and_flags): Fix where
29994 we look for cc setter after the compare-elim changes.
29995 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
29996 with the vector to match what compare-elim now expects.
29997 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
29998 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
29999 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
30000 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
30001 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
30002 (neg<mode>2_insn_set_overflow): Likewise.
30004 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
30006 PR middle-end/79794
30007 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
30008 maybe_expand_insn call, set ops[0].target. If still set after call,
30009 set alt_rtl. Add extra arg to recursive calls.
30010 (extract_bit_field): Add alt_rtl argument. Pass to
30012 * expmed.h (extract_bit_field): Fix prototype.
30013 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
30014 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
30015 to extract_bit_field_calls.
30016 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
30017 Pass alt_rtl to extract_bit_field calls.
30018 * calls.c (store_unaligned_arguments_into_psuedos)
30019 load_register_parameters): Pass extra NULL to extract_bit_field calls.
30020 * optabs.c (maybe_legitimize_operand): Clear op->target when call
30022 * optabs.h (struct expand_operand): Add target bitfield.
30024 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
30026 * compare-elim.c (try_eliminate_compare): Canonicalize
30027 operation with embedded compare to
30028 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
30029 (set (reg) (operation)].
30031 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
30033 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
30036 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
30037 cost of adding a carry flag for ADC instruction.
30038 [case MINUS]: Ignore the cost of subtracting a carry flag
30039 for SBB instruction.
30041 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
30043 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
30045 * config/rs6000/bmiintrin.h: New file.
30046 * config/rs6000/bmi2intrin.h: New file.
30047 * config/rs6000/x86intrin.h: New file.
30049 2017-05-12 Jeff Law <law@redhat.com>
30051 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
30054 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
30056 PR middle-end/80707
30057 * tree-cfg.c: Remove cfg edges of unreachable case statements.
30059 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30061 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
30062 early expansion of vector divide builtins.
30063 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
30064 builtins identified as having unsigned arguments.
30066 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30068 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
30069 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
30070 expansion of vector logical operations (and, andc, or, xor,
30073 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30075 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
30076 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
30078 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30080 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
30081 early GIMPLE expansion of vector multiplies.
30083 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
30085 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
30086 TARGET_HAVE_MOVT conditional.
30087 (movt splitter): Likewise.
30089 2017-05-12 Richard Biener <rguenther@suse.de>
30091 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
30092 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
30093 Fold all stmts not inplace.
30095 2017-05-12 Richard Biener <rguenther@suse.de>
30097 PR tree-optimization/80713
30098 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
30099 inserted_exprs bit for not removed stmts.
30101 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
30103 PR middle-end/69921
30104 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
30105 parallelized" attribute for parallelized OpenACC kernels.
30106 * omp-offload.c (execute_oacc_device_lower): Use it.
30108 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
30109 Set "oacc kernels" attribute.
30110 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
30111 parameter. Adjust all users.
30112 (oacc_fn_attrib_kernels_p): Remove function.
30113 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
30114 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
30115 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
30116 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
30117 assert "oacc kernels" attribute is set.
30119 2017-05-11 Carl Love <cel@us.ibm.com>
30121 * config/rs6000/rs6000-c: Add support for built-in functions
30122 vector unsigned char vec_popcnt (vector signed char)
30123 vector unsigned char vec_popcnt (vector unsigned char)
30124 vector unsigned short vec_popcnt (vector signed short)
30125 vector unsigned short vec_popcnt (vector unsigned short)
30126 vector unsigned int vec_popcnt (vector signed int)
30127 vector unsigned int vec_popcnt (vector unsigned int)
30128 vector unsigned long long vec_popcnt (vector signed long long)
30129 vector unsigned long long vec_popcnt (vector unsigned long long)
30130 vector signed long long vec_slo (vector signed long long,
30131 vector signed char)
30132 vector signed long long vec_slo (vector signed long long,
30133 vector unsigned char)
30134 vector unsigned long long vec_slo (vector unsigned long long,
30135 vector signed char)
30136 vector unsigned long long vec_slo (vector unsigned long long,
30137 vector unsigned char)
30138 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
30139 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
30140 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
30141 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
30142 * doc/extend.texi: Update the built-in documentation file for the
30143 new built-in functions.
30145 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
30147 * attribs.h (sorted_attr_string): Move machine independent
30148 functions for target clone support from the i386 port to common
30149 code. Rename ix86_function_versions to common_function_versions.
30150 Rename make_name to make_unique_name.
30151 (common_function_versions): Likewise.
30152 (make_unique_name): Likewise.
30153 (make_dispatcher_decl): Likewise.
30154 (is_function_default_version): Likewise.
30155 * attribs.c (attr_strcmp): Likewise.
30156 (sorted_attr_string): Likewise.
30157 (common_function_versions): Likewise.
30158 (make_unique_name): Likewise.
30159 (make_dispatcher_decl): Likewise.
30160 (is_function_default_version): Likewise.
30161 * config/i386/i386.c (attr_strcmp): Likewise.
30162 (sorted_attr_string): Likewise.
30163 (ix86_function_versions): Likewise.
30164 (make_name): Likewise.
30165 (make_dispatcher_decl): Likewise.
30166 (is_function_default_version): Likewise.
30167 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
30169 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30172 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
30173 Account for direct move costs for vec_construct of integer
30176 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
30179 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
30180 (UNSPEC_STX_ATOMIC): Ditto.
30181 (loaddi_via_sse): New insn.
30182 (storedi_via_sse): Ditto.
30183 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
30184 Update corresponding peephole2 patterns.
30185 (atomic_storedi_fpu): Ditto.
30187 2017-05-11 Julia Koval <julia.koval@intel.com>
30189 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
30190 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
30192 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
30193 (__builtin_ia32_rsqrt14ss_mask): New builtins.
30194 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
30196 2017-05-11 Nathan Sidwell <nathan@acm.org>
30198 * graphite-poly.c: Include dumpfile.h.
30200 * dumpfle.h (dump_function): Declare here ...
30201 * tree-dump.h (dump_function): ... not here.
30202 * dumpfile.c: #include tree-cfg.h.
30203 (dump_function): Move here from ...
30204 * tree-dump.c (dump_function): ... here.
30205 * gimplify.c: #include splay-tree.h, not tree-dump.h.
30206 * graphite-poly.c: Don't include tree-dump.h.
30207 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
30208 * print-tree.c: Likewise.
30209 * stor-layout.c: Likewise.
30210 * tree-nested.c: Likewise.
30212 * dumpfile.c (dump_start): Use TDF_FLAGS.
30213 (dump_enable_all): Fix TDF_KIND check thinko.
30215 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
30217 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
30218 array entries to represent two legal parameterizations of the
30219 overloaded __builtin_cmpb function, as represented by the
30220 P6_OV_BUILTIN_CMPB constant.
30221 (altivec_resolve_overloaded_builtin): Add special case handling
30222 for the __builtin_cmpb function, as represented by the
30223 P6_OV_BUILTIN_CMPB constant.
30224 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
30225 (BU_P6_64BIT_2): New macro.
30226 (BU_P6_OVERLOAD_2): New macro
30227 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
30228 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
30229 (CMPB): Add overload support to represent both 32-bit and 64-bit
30230 compare-bytes function.
30231 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
30232 support for TARGET_CMPB.
30233 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
30234 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
30235 documentation of the __builtin_cmpb overloaded built-in function.
30237 2017-05-11 Richard Biener <rguenther@suse.de>
30239 PR tree-optimization/80705
30240 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
30241 bases are not vectorizable.
30243 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30245 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
30246 when counting register pressure.
30248 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30250 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
30251 (iv_ca_more_deps): Renamed to ...
30252 (iv_ca_compare_deps): ... this.
30253 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
30255 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30257 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
30259 (determine_group_iv_costs): ... here.
30260 (find_inv_vars_cb): Record inv var if it's not recorded before.
30262 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30264 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
30265 (get_shiftadd_cost): Ditto.
30267 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30269 * tree-ssa-address.c: Include header file.
30270 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
30272 (add_to_parts): Refactor.
30273 (addr_to_parts): New parameter. Update use of move_hint_to_base.
30274 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
30277 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30279 PR tree-optimization/53090
30280 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
30282 (extract_cond_operands): Detect condition with IV on both sides
30283 and return COMP_IV_EXPR_2.
30284 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
30285 (rewrite_use_compare): Simplify by removing call to function
30286 extract_cond_operands.
30288 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30290 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
30291 (extract_cond_operands): Detect condition comparing against non-
30292 invariant bound and return appropriate enum value.
30293 (find_interesting_uses_cond): Update use of extract_cond_operands.
30294 Handle its return value accordingly.
30295 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
30297 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30299 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
30300 nonlinear iv_use computation in loop invariant sensitive way.
30302 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30304 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
30305 (find_iv_candidates): Call relate_compare_use_with_all_cands.
30307 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30309 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
30310 (dump_cand): Support iv_cand.inv_exprs.
30311 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
30313 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
30316 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30318 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
30320 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
30321 as local function. Include necessary header files.
30322 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
30324 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30326 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
30328 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30330 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
30331 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
30332 RSHIFT_EXPR and BIT_NOT_EXPR.
30334 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30336 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
30337 (adjust_setup_cost): New parameter supporting round up adjustment.
30338 (struct address_cost_data): Delete.
30339 (force_expr_to_var_cost): Don't bound cost with spill_cost.
30340 (split_address_cost, ptr_difference_cost): Delete.
30341 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
30342 (struct ainc_cost_data): New struct.
30343 (get_address_cost_ainc): New function.
30344 (get_address_cost, get_computation_cost): Reimplement.
30345 (determine_group_iv_cost_address): Record inv_expr for all uses of
30347 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
30348 (iv_ca_has_deps): Reimplemented to ...
30349 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
30351 (iv_ca_extend): Call iv_ca_more_deps.
30353 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30355 * tree-ssa-address.c (struct mem_address): Move to header file.
30356 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
30357 * tree-ssa-address.h (struct mem_address): Move from C file.
30358 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
30360 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30362 * tree-affine.h (aff_combination_type): New interface.
30363 (aff_combination_zero_p): Remove static.
30364 (aff_combination_const_p): New interface.
30365 (aff_combination_singleton_var_p): New interfaces.
30367 2017-05-11 Richard Biener <rguenther@suse.de>
30369 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
30370 Skip unreachable blocks and destinations.
30371 (eliminate): Move stmt removal and fixup ...
30372 (fini_eliminate): ... here. Skip inserted exprs.
30373 (pass_pre::execute): Move fini_pre after fini_eliminate.
30374 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
30375 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
30376 PRE to get rid of dead code that has invalid SSA form and
30377 split critical edges again.
30379 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30381 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
30383 2017-05-11 Richard Biener <rguenther@suse.de>
30385 * passes.c (execute_function_todo): Verify loops if they are
30386 said to be up-to-date.
30387 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
30388 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
30390 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
30393 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
30394 handle calling assemble_external ourself.
30397 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
30398 modes with zero size. Enhance comment.
30400 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30402 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
30403 built-ins for vec_xl and vec_xst with short and char pointer
30406 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
30408 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
30409 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
30410 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
30411 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
30412 (_mm_maskz_min_round_ss): New intrinsics.
30413 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
30414 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
30415 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
30416 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
30417 (__builtin_ia32_minss_mask_round): New builtins.
30418 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
30419 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
30420 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
30422 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
30423 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
30425 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
30428 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
30430 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
30431 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
30432 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
30433 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
30434 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
30435 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
30436 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
30437 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
30438 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
30439 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
30440 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
30441 (__builtin_ia32_mulss_mask_round): New builtins.
30442 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
30443 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
30444 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
30446 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
30447 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
30449 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
30452 2017-05-10 Julia Koval <julia.koval@intel.com>
30454 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
30455 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
30456 (_mm256_setr_m128i): New intrinsics.
30458 2017-05-10 Julia Koval <julia.koval@intel.com>
30460 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
30461 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
30462 (_mm_maskz_rcp14_ss): New intrinsics.
30463 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
30464 (__builtin_ia32_rcp14ss_mask): New builtins.
30465 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
30467 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
30469 PR tree-optimization/51513
30470 * tree-cfg.c (gimple_seq_unreachable_p): New function.
30471 (assert_unreachable_fallthru_edge_p): Use it.
30472 (group_case_labels_stmt): Likewise.
30473 * tree-cfg.h: Prototype it.
30474 * stmt.c: Include cfghooks.h and tree-cfg.h.
30475 (emit_case_dispatch_table) <gap_label>: New local variable.
30476 Use it to fill dispatch table gaps.
30477 Test for default_label before updating probabilities.
30478 (expand_case) <default_label>: Remove unneeded initialization.
30479 Test for unreachable default case statement and remove its edge.
30480 Set default_label accordingly.
30481 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
30483 2017-05-10 Carl Love <cel@us.ibm.com>
30485 * config/rs6000/rs6000-c: Add support for built-in functions
30486 vector signed char vec_neg (vector signed char)
30487 vector signed short int vec_neg (vector short int)
30488 vector signed int vec_neg (vector signed int)
30489 vector signed long long vec_neg (vector signed long long)
30490 vector float vec_neg (vector float)
30491 vector double vec_neg (vector double)
30492 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
30494 * config/rs6000/altivec.h: Add define for vec_neg
30495 * doc/extend.texi: Update the built-in documentation for the
30496 new built-in functions.
30498 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30500 PR tree-optimization/77644
30501 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
30503 2017-05-10 Nathan Sidwell <nathan@acm.org>
30505 * dumpfile.h (TDI_lang_all): New.
30506 (TDF_KIND): New. Renumber others
30507 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
30509 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
30511 (get_dump_file_name): Adjust suffix generation.
30512 (dump_enable_all): Use TDF_KIND.
30513 * doc/invoke.texi (-fdump-lang-all): Document.
30515 * dumpfile.h: Tabify.
30517 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
30520 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
30521 Move member access before delete.
30523 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
30525 * tree-inline.c (expand_call_inline): Split block at stmt
30528 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
30531 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
30532 are now unused after splitting mov{sf,sd}_hardfloat.
30533 (f32_lr2): Likewise.
30534 (f32_lm): Likewise.
30535 (f32_lm2): Likewise.
30536 (f32_li): Likewise.
30537 (f32_li2): Likewise.
30538 (f32_lv): Likewise.
30539 (f32_sr): Likewise.
30540 (f32_sr2): Likewise.
30541 (f32_sm): Likewise.
30542 (f32_sm2): Likewise.
30543 (f32_si): Likewise.
30544 (f32_si2): Likewise.
30545 (f32_sv): Likewise.
30546 (f32_dm): Likewise.
30547 (f32_vsx): Likewise.
30548 (f32_av): Likewise.
30549 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
30550 For movsf, order stores so the VSX stores occur before the GPR
30551 store which encourages the register allocator to use a traditional
30552 FPR instead of a GPR. For movsd, order the stores so that the GPR
30553 store comes before the VSX stores to allow the power6 to work.
30554 This is due to the power6 not having a 32-bit integer store
30555 instruction from a FPR.
30556 (movsf_hardfloat): Likewise.
30557 (movsd_hardfloat): Likewise.
30559 2017-05-09 Martin Sebor <msebor@redhat.com>
30561 PR translation/80280
30562 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
30565 PR translation/80280
30566 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
30567 data member added in r247778.
30568 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
30570 2017-05-09 Nathan Sidwell <nathan@acm.org>
30572 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
30574 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
30577 2017-05-09 Marek Polacek <polacek@redhat.com>
30579 * doc/invoke.texi: Fix typo.
30581 2017-05-09 Richard Biener <rguenther@suse.de>
30583 * tree-vrp.c (vrp_val_is_max): Adjust comment.
30584 (vrp_val_is_min): Likewise.
30585 (set_value_range_to_value): Likewise.
30586 (set_value_range_to_nonnegative): Likewise.
30587 (gimple_assign_nonzero_p): Likewise.
30588 (gimple_stmt_nonzero_p): Likewise.
30589 (vrp_int_const_binop): Likewise. Remove unreachable case.
30590 (adjust_range_with_scev): Adjust comments.
30591 (compare_range_with_value): Likewise.
30592 (extract_range_from_phi_node): Likewise.
30593 (test_for_singularity): Likewise.
30595 2017-05-09 Richard Biener <rguenther@suse.de>
30597 * tree-vrp.c (get_single_symbol): Add assert that we don't
30598 get overflowed constants as invariant part.
30599 (compare_values_warnv): Add comment before the TREE_NO_WARNING
30600 checks. Use wi::cmp instead of recursing for integer constants.
30601 (compare_values): Just ignore whether we assumed undefined
30602 overflow instead of failing the compare.
30603 (extract_range_for_var_from_comparison_expr): Add comment before the
30604 TREE_NO_WARNING sets.
30605 (test_for_singularity): Likewise.
30606 (extract_range_from_comparison): Do not disable optimization
30607 when we assumed undefined overflow.
30608 (extract_range_basic): Remove init of unused var.
30610 2017-05-09 Richard Biener <rguenther@suse.de>
30612 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
30613 (extract_range_from_multiplicative_op_1): Adjust.
30614 (extract_range_from_binary_expr_1): Use int_const_binop.
30616 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
30619 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
30620 rs6000_store_data_bypass_p in seven define_bypass directives and
30621 in several comments.
30622 * config/rs6000/rs6000-protos.h: Add prototype for
30623 rs6000_store_data_bypass_p function.
30624 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
30625 function implements slightly different (rs6000-specific) semantics
30626 than store_data_bypass_p, returning false rather than aborting
30627 with assertion error when arguments do not satisfy the
30628 requirements of store data bypass.
30629 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
30630 rs6000_store_data_bypass_p.
30632 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
30634 * config/xtensa/xtensa-protos.h
30635 (xtensa_initial_elimination_offset): New declaration.
30636 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
30637 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
30638 macro definition, add case for FRAME_POINTER_REGNUM when
30639 FRAME_GROWS_DOWNWARD.
30640 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
30641 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
30642 xtensa_initial_elimination_offset.
30644 2017-05-08 Nathan Sidwell <nathan@acm.org>
30646 * doc/invoke.texi: Alphabetize -fdump options.
30648 2017-05-08 Martin Sebor <msebor@redhat.com>
30650 PR translation/80280
30651 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
30653 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
30655 * target.def (compute_frame_layout): New optional target hook.
30656 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
30657 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
30658 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
30660 * reload1.c (verify_initial_elim_offsets): Likewise.
30661 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
30662 (use_simple_return_p): Call arm_compute_frame_layout if needed.
30663 (arm_get_frame_offsets): Split up into this ...
30664 (arm_compute_frame_layout): ... and this function.
30666 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
30668 * config/aarch64/constraints.md (Usa): New constraint.
30669 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
30671 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
30673 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
30674 with_multilib_list after it has been checked.
30676 2017-05-08 Richard Biener <rguenther@suse.de>
30678 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
30679 (bitmap_set_subtract_values): Likewise.
30681 2017-05-08 Richard Biener <rguenther@suse.de>
30683 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
30684 (gimple_assign_nonzero): ... this and remove strict_overflow_p
30686 (gimple_stmt_nonzero_warnv_p): Rename to ...
30687 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
30689 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
30690 (extract_range_basic): Adjust, do not disable propagation on
30691 strict overflow sensitive simplification.
30692 (vrp_visit_cond_stmt): Likewise.
30694 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
30696 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
30697 body size unconditionally.
30699 2017-05-07 Jeff Law <law@redhat.com>
30702 2017-05-06 Jeff Law <law@redhat.com>
30703 PR tree-optimization/78496
30704 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
30707 PR tree-optimization/78496
30708 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
30709 (simplify_stmt_using_ranges): Call it.
30710 (vrp_dom_walker::before_dom_children): Extract equivalences
30711 from an ASSERT_EXPR with an equality comparison against a
30714 2017-05-06 Jeff Law <law@redhat.com>
30716 PR tree-optimization/78496
30717 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
30720 PR tree-optimization/78496
30721 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
30722 (simplify_stmt_using_ranges): Call it.
30723 (vrp_dom_walker::before_dom_children): Extract equivalences
30724 from an ASSERT_EXPR with an equality comparison against a
30727 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
30729 * lra-constraints.c (lra_copy_reg_equiv): New function.
30730 (split_reg): Use it to copy equivalence information from the
30731 original register to the spill register.
30733 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
30735 PR rtl-optimization/75964
30736 * simplify-rtx.c (simplify_const_relational_operation): Remove
30737 invalid handling of comparisons of integer ABS.
30739 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
30741 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
30742 initialize to zero.
30743 (init_regs): Remove declaration.
30744 (function_arg_advance_32): Initialize error_p as boolean variable.
30746 2017-05-05 Nathan Sidwell <nathan@acm.org>
30748 * store-motion.c (remove_reachable_equiv_notes): Reformat long
30749 lines. Use for (;;).
30751 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30753 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
30754 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
30755 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
30756 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
30757 VF=2 that require versioning.
30759 2017-05-05 David Malcolm <dmalcolm@redhat.com>
30761 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
30764 2017-05-05 David Malcolm <dmalcolm@redhat.com>
30766 * diagnostic.h (diagnostic_override_option_index): Convert from
30767 macro to inline function.
30769 2017-05-05 David Malcolm <dmalcolm@redhat.com>
30771 * diagnostic.c (last_module_changed_p): New function.
30772 (set_last_module): New function.
30773 (diagnostic_report_current_module): Convert macro usage to
30774 the above functions.
30775 * diagnostic.h (diagnostic_context::last_module): Strengthen
30776 from const line_map * to const line_map_ordinary *.
30777 (diagnostic_last_module_changed): Delete macro.
30778 (diagnostic_set_last_module): Delete macro.
30780 2017-05-05 David Malcolm <dmalcolm@redhat.com>
30782 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
30783 with diagnostic_report_diagnostic.
30784 (diagnostic_n_impl_richloc): Likewise.
30785 * diagnostic.h (report_diagnostic): Delete macro.
30786 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
30787 with diagnostic_report_diagnostic.
30788 * substring-locations.c (format_warning_va): Likewise.
30790 2017-05-05 David Malcolm <dmalcolm@redhat.com>
30792 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
30793 save/restor of format_spec. Move option-printing code to...
30794 (print_option_information): ...this new function, and
30795 reimplement by simply printing to the pretty_printer,
30796 rather than appending to the format string.
30798 2017-05-05 David Malcolm <dmalcolm@redhat.com>
30800 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
30801 handling logic into...
30802 (update_effective_level_from_pragmas): ...this new function.
30804 2017-05-04 Andrew Waterman <andrew@sifive.com>
30806 * config/riscv/riscv.opt (mstrict-align): New option.
30807 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
30808 (SLOW_UNALIGNED_ACCESS): Define.
30809 (riscv_slow_unaligned_access): Declare.
30810 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
30812 (riscv_slow_unaligned_access): New variable.
30813 (rocket_tune_info): Set slow_unaligned_access to true.
30814 (optimize_size_tune_info): Set slow_unaligned_access to false.
30815 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
30816 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
30817 (riscv_option_override): Set riscv_slow_unaligned_access.
30818 * doc/invoke.texi: Add -mstrict-align to RISC-V.
30820 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
30822 * config/riscv/riscv.md: Unify indentation.
30824 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
30829 * config/rs6000/rs6000.md (u code attribute): Add FIX and
30831 (extendsi<mode>2): Add support for doing sign extension via
30832 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
30833 don't have ISA 3.0 instructions.
30834 (extendsi<mode>2 splitter): Likewise.
30835 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
30836 generate the normal insns since SImode can now go in vector
30837 registers. Disallow the special UNSPECs needed for previous
30838 machines to hide SImode being used. Add new insns
30839 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
30840 (fix_trunc<mode>si2_stfiwx): Likewise.
30841 (fix_trunc<mode>si2_internal): Likewise.
30842 (fixuns_trunc<mode>si2): Likewise.
30843 (fixuns_trunc<mode>si2_stfiwx): Likewise.
30844 (fctiw<u>z_<mode>_smallint): Likewise.
30845 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
30846 of floating point to 32-bit integer from doing a direct move to
30847 the GPR registers to do a store.
30848 (fctiwz_<mode>): Break long line.
30850 2017-05-05 Bin Cheng <bin.cheng@arm.com>
30852 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
30853 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
30854 (addr_list, addr_offset_valid_p): New.
30855 (split_address_groups): Check offset validity with above function.
30856 (gt-tree-ssa-loop-ivopts.h): Include header file.
30858 2017-05-05 Nathan Sidwell <nathan@acm.org>
30860 * config.gcc (arm*-*-*): Add missing 'fi'.
30862 2017-05-05 Steve Ellcey <sellcey@cavium.com>
30864 * doc/invoke.texi (-fopt-info): Explicitly say order of options
30865 included in -fopt-info does not matter.
30866 * doc/optinfo.texi (-fopt-info): Fix description of default
30867 behavour. Explicitly say order of options included in -fopt-info
30870 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
30872 * config.gcc: Allow combinations of aprofile and rmprofile values for
30873 --with-multilib-list.
30874 * config/arm/t-multilib: New file.
30875 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
30876 variables. Remove setting of ISA and floating-point ABI in
30877 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
30878 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
30879 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
30880 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
30882 * config/arm/t-rmprofile: Likewise except for the matches changes.
30883 * doc/install.texi (--with-multilib-list): Document the combination of
30884 aprofile and rmprofile values and warn about pitfalls in doing that.
30886 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
30888 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
30889 (movdi_aarch64): Likewise.
30891 2017-05-05 Jakub Jelinek <jakub@redhat.com>
30893 PR tree-optimization/80632
30894 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
30896 (build_arrays): Initialize it for virtual phis.
30897 (fix_phi_nodes): Use it for virtual phis.
30899 PR tree-optimization/80558
30900 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
30901 [x, y] op z into [x op, y op z] for op & or | if conditions
30904 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
30905 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
30908 * config/arm/arm.md (use_literal_pool): Remove.
30909 (64-bit immediate split): No longer takes cost into consideration
30910 if arm_disable_literal_pool is enabled.
30911 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
30912 used when arm_disable_literal_pool is enabled.
30913 (arm_max_const_double_inline_cost): Remove use of
30914 arm_disable_literal_pool.
30915 (push_minipool_fix): Add assert.
30916 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
30917 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
30918 (no_literal_pool_sf_immediate): New.
30920 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30922 PR tree-optimization/80613
30923 * tree-ssa-dce.c (propagate_necessity): Remove cases for
30924 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
30926 2017-05-05 Richard Biener <rguenther@suse.de>
30928 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
30930 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
30932 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
30933 of this flag from insn conditions due to removal from r247495.
30935 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
30937 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
30939 (arm_early_store_addr_dep_ptr): Likewise.
30940 * config/arm/aarch-common-protos.h
30941 (arm_early_load_addr_dep_ptr): Add prototype.
30942 (arm_early_store_addr_dep_ptr): Likewise.
30943 * config/arm/cortex-a53.md: Add new bypasses.
30945 2017-05-05 Jakub Jelinek <jakub@redhat.com>
30947 * tree.c (next_type_uid): Change type to unsigned.
30948 (type_hash_canon): Decrement back next_type_uid if
30949 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
30950 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
30953 2017-05-04 Martin Sebor <msebor@redhat.com>
30955 * builtins.c: Fix a trivial typo in a comment.
30957 PR middle-end/79234
30958 * builtins.c (check_sizes): Adjust to handle reading past the end.
30959 Avoid printing excessive upper bound of ranges. Use %E to print
30960 tree nodes instead of converting them to %wu.
30961 (expand_builtin_memchr): New function.
30962 (compute_dest_size): Rename...
30963 (compute_objsize): ...to this.
30964 (expand_builtin_memcpy): Adjust.
30965 (expand_builtin_mempcpy): Adjust.
30966 (expand_builtin_strcat): Adjust.
30967 (expand_builtin_strcpy): Adjust.
30968 (check_strncat_sizes): Adjust.
30969 (expand_builtin_strncat): Adjust.
30970 (expand_builtin_strncpy): Adjust and simplify.
30971 (expand_builtin_memset): Adjust.
30972 (expand_builtin_bzero): Adjust.
30973 (expand_builtin_memcmp): Adjust.
30974 (expand_builtin): Handle memcmp.
30975 (maybe_emit_chk_warning): Check strncat just once.
30977 2017-05-04 Martin Sebor <msebor@redhat.com>
30979 PR preprocessor/79214
30980 PR middle-end/79222
30981 PR middle-end/79223
30982 * builtins.c (check_sizes): Add inlining context and issue
30983 warnings even when -Wno-system-headers is set.
30984 (check_strncat_sizes): Same.
30985 (expand_builtin_strncat): Same.
30986 (expand_builtin_memmove): New function.
30987 (expand_builtin_stpncpy): Same.
30988 (expand_builtin): Handle memmove and stpncpy.
30990 2017-05-04 Bin Cheng <bin.cheng@arm.com>
30992 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
30993 which is not used any more.
30995 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
30997 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
30999 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31001 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
31002 (cortexa53_tunings): Likewise.
31003 (cortexa57_tunings): Likewise.
31004 (cortexa72_tunings): Likewise.
31005 (cortexa73_tunings): Likewise.
31007 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31009 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
31010 Set loop alignment to 8.
31012 2017-05-04 Martin Sebor <msebor@redhat.com>
31014 PR translation/80280
31015 * builtins.c (expand_builtin_object_size): Add missing quoting to
31016 %D and like directives.
31017 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
31018 (hsa_type_for_tree_type): Same.
31019 (verify_function_arguments): Same.
31020 * symtab.c (symbol_table::change_decl_assembler_name): Same.
31021 * varasm.c (get_section): Same.
31024 2017-05-04 Martin Sebor <msebor@redhat.com>
31026 PR translation/80280
31027 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
31029 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31031 * config/aarch64/aarch64.c (generic_addrcost_table):
31032 Change HI/TI mode setting.
31034 2017-05-04 Martin Jambor <mjambor@suse.cz>
31036 PR tree-optimization/80622
31037 * tree-sra.c (comes_initialized_p): New function.
31038 (build_accesses_from_assign): Only set write lazily when
31039 comes_initialized_p is false.
31040 (analyze_access_subtree): Use comes_initialized_p.
31041 (propagate_subaccesses_across_link): Assert !comes_initialized_p
31042 instead of testing for PARM_DECL.
31044 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31046 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
31047 constraint on operand 0 to allow more general addressing modes.
31048 Adjust output template.
31049 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
31051 * config/aarch64/aarch64-protos.h
31052 (aarch64_address_valid_for_prefetch_p): Declare prototype.
31053 * config/aarch64/constraints.md (Dp): New address constraint.
31054 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
31057 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
31059 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
31060 update use of estimate_ipcp_clone_size_and_time.
31061 (estimate_local_effects): Update use of
31062 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
31063 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
31064 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
31065 Return nonspecialized time.
31067 2017-05-04 Richard Biener <rguenther@suse.de>
31069 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
31070 for the last VUSE which def dominates the PHI. Directly call
31072 (get_continuation_for_phi_1): Remove.
31074 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
31076 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
31077 to explain the use of truncating division. Cap the number of
31078 iterations to the maximum given by nb_iterations_upper_bound,
31081 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
31083 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
31084 * configure: Regenerate.
31085 * config.in: Regenerate.
31086 * config/i386/driver-mingw32.c: new file.
31087 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
31088 * config.host: Link driver-mingw32.o on MinGW host.
31089 * doc/install.texi: Document new --enable-mingw-wildcard configure
31092 2017-05-04 Marek Polacek <polacek@redhat.com>
31094 PR tree-optimization/80612
31095 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
31097 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31098 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
31100 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
31101 (movt splitter): Likewise.
31102 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
31103 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
31104 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
31105 block for Thumb-1 with MOVT.
31106 (thumb2_legitimate_address_p): Move code block ...
31107 (can_avoid_literal_pool_for_label_p): ... into this new function.
31108 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
31110 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
31111 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
31112 "M-profile targets with the MOVT instruction".
31114 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31116 * config/arm/arm-builtins.c (arm_init_builtins): Rename
31117 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
31118 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
31120 2017-05-04 Martin Liska <mliska@suse.cz>
31122 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
31123 variable cond_code.
31125 2017-05-04 Richard Biener <rguenther@suse.de>
31127 * tree.c (array_at_struct_end_p): Handle arrays at struct
31128 end with flexarrays more conservatively. Refactor and treat
31129 arrays of arrays or aggregates more strict. Fix
31130 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
31131 * tree.c (array_at_struct_end_p): Adjust prototype.
31132 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
31133 * gimple-fold.c (get_range_strlen): Likewise.
31134 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
31136 2017-05-04 Richard Biener <rguenther@suse.de>
31138 PR tree-optimization/31130
31139 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
31141 (supports_overflow_infinity): Likewise.
31142 (is_negative_overflow_infinity): Likewise.
31143 (is_positive_overflow_infinity): Likewise.
31144 (is_overflow_infinity): Likewise.
31145 (stmt_overflow_infinity): Likewise.
31146 (overflow_infinity_range_p): Likewise.
31147 (usable_range_p): Remove as always returning true.
31148 (make_overflow_infinity): Remove.
31149 (negative_overflow_infinity): Likewise.
31150 (positive_overflow_infinity): Likewise.
31151 (avoid_overflow_infinity): Likewise.
31152 (set_value_range): Adjust accordingly.
31153 (set_value_range_to_nonnegative): Likewise, remove now unused
31154 overflow_infinity arg.
31155 (vrp_operand_equal_p): Adjust.
31156 (update_value_range): Likewise.
31157 (range_int_cst_singleton_p): Likewise.
31158 (operand_less_p): Likewise.
31159 (compare_values_warnv): Likewise.
31160 (extract_range_for_var_from_comparison_expr): Likewise.
31161 (vrp_int_const_binop): Likewise.
31162 (zero_nonzero_bits_from_vr): Likewise.
31163 (extract_range_from_multiplicative_op_1): Likewise.
31164 (extract_range_from_binary_expr_1): Likewise.
31165 (extract_range_from_unary_expr): Likewise.
31166 (extract_range_from_comparison): Likewise.
31167 (extract_range_basic): Likewise.
31168 (adjust_range_with_scev): Likewise.
31169 (compare_ranges): Likewise.
31170 (compare_range_with_value): Likewise.
31171 (dump_value_range): Likewise.
31172 (test_for_singularity): Likewise, remove strict_overflow_p parameter
31174 (simplify_cond_using_ranges): Adjust.
31176 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
31178 * brig-builtins.def: Added a builtin for class_f64.
31179 * builtin-types.def: Added a builtin type needed by class_f64.
31181 2017-05-03 Jason Merrill <jason@redhat.com>
31183 * timevar.def: Add TV_CONSTEXPR.
31185 2017-05-03 David Malcolm <dmalcolm@redhat.com>
31187 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
31189 2017-05-03 Martin Jambor <mjambor@suse.cz>
31191 * ipa-prop.c (ipa_update_after_lto_read): Removed.
31192 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
31193 * ipa-cp.c (ipcp_propagate_stage): Do not call
31194 ipa_update_after_lto_read.
31195 * ipa-inline.c (ipa_inline): Likewise.
31197 2017-05-03 Martin Jambor <mjambor@suse.cz>
31199 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
31200 tag. Added a default constructor and a destructor.
31201 (ipa_edge_args_sum_t): New class;
31202 (ipa_edge_args_sum): Declare.
31203 (ipa_edge_args_vector): Remove declaration.
31204 (IPA_EDGE_REF): Use ipa_edge_args_sum.
31205 (ipa_free_edge_args_substructures): Remove declaration.
31206 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
31207 (ipa_edge_args_info_available_for_edge_p): Likewise.
31208 * ipa-prop.c (ipa_edge_args_vector): Removed.
31209 (edge_removal_hook_holder): Likewise.
31210 (edge_duplication_hook_holder): Likewise.
31211 (ipa_edge_args_sum): New variable.
31212 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
31213 ipa_edge_args_vector.
31214 (ipa_free_edge_args_substructures): Likewise.
31215 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
31216 ipa_edge_args_vector.
31217 (ipa_edge_removal_hook): Turned into method
31218 ipa_edge_args_sum_t::remove.
31219 (ipa_edge_duplication_hook): Turned into method
31220 ipa_edge_args_sum_t::duplicate.
31221 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
31222 registering edge hooks.
31223 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
31224 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
31225 ipa_edge_args_sum instead of ipa_edge_args_vector.
31226 * ipa-profile.c (ipa_profile): Likewise.
31228 2017-05-03 Martin Jambor <mjambor@suse.cz>
31230 * symbol-summary.h (function_summary): New method exists.
31231 (function_summary::symtab_removal): Deallocate through release.
31232 (call_summary): New class.
31233 (gt_ggc_mx): New overload.
31234 (gt_pch_nx): Likewise.
31235 (gt_pch_nx): Likewise.
31237 2017-05-03 Jeff Law <law@redhat.com>
31239 PR tree-optimization/78496
31240 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
31241 from simplify_cond_using_ranges. Split off code to walk
31242 backwards through casts into ...
31243 (simplify_cond_using_ranges_2): New function.
31244 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
31245 (execute_vrp): After identifying jump threads, call
31246 simplify_cond_using_ranges_2.
31248 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
31251 * ipa-inline.h (inline_summary): Add ctor.
31252 (create_ggc): Do not use ggc_cleared_alloc.
31254 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
31255 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
31257 * gcc.c (handle_braces): Support escaping in switch matching
31259 * doc/invoke.texi (Spec Files): Document it.
31260 Remove superfluous @code markup in items.
31262 2017-05-03 David Malcolm <dmalcolm@redhat.com>
31264 * diagnostic-show-locus.c (struct column_range): New struct.
31265 (get_affected_columns): New function.
31266 (get_printed_columns): New function.
31267 (struct correction): New struct.
31268 (correction::ensure_capacity): New function.
31269 (correction::ensure_terminated): New function.
31270 (struct line_corrections): New struct.
31271 (line_corrections::~line_corrections): New dtor.
31272 (line_corrections::add_hint): New function.
31273 (layout::print_trailing_fixits): Reimplement in terms of the new
31275 (selftest::test_overlapped_fixit_printing): New function.
31276 (selftest::diagnostic_show_locus_c_tests): Call it.
31278 2017-05-03 Nathan Sidwell <nathan@acm.org>
31280 Canonicalize canonical type hashing
31281 * tree.h (type_hash_canon_hash): Declare.
31282 * tree.c (type_hash_list, attribute_hash_list): Move into
31283 type_hash_canon_hash.
31284 (build_type_attribute_qual_variant): Break out hash code calc into
31285 type_hash_canon_hash.
31286 (type_hash_canon_hash): New. Generic type hash computation.
31287 (build_range_type_1, build_array_type_1, build_function_type,
31288 build_method_type_directly, build_offset_type, build_complex_type,
31289 make_vector_type): Call it.
31291 2017-05-03 Richard Biener <rguenther@suse.de>
31293 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
31294 When all DRs have unknown misaligned do not always peel
31295 when there is a store but apply the same costing model as if
31296 there were only loads.
31298 2017-05-03 Richard Biener <rguenther@suse.de>
31301 PR tree-optimization/80492
31302 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
31303 compare_base_decls returning dont-know properly.
31305 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
31307 * config/arm/iterators.md (CCSI): New mode iterator.
31308 (arch): New mode attribute.
31309 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
31310 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
31311 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
31312 code iterator for success result mode.
31313 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
31314 the corresponding new insn generators.
31316 2017-05-03 Bin Cheng <bin.cheng@arm.com>
31319 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31320 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
31322 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
31324 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
31325 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
31326 (DDR_A): Wrap DDR argument in brackets.
31327 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
31328 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
31329 (DDR_REVERSED_P): Likewise.
31331 2017-05-03 Jakub Jelinek <jakub@redhat.com>
31333 PR tree-optimization/79472
31334 * tree-switch-conversion.c (struct switch_conv_info): Add
31335 contiguous_range and default_case_nonstandard fields.
31336 (collect_switch_conv_info): Compute contiguous_range and
31337 default_case_nonstandard fields, don't clear final_bb if
31338 contiguous_range and only the default case doesn't have the required
31340 (check_all_empty_except_final): Set default_case_nonstandard instead
31341 of failing if contiguous_range and the default case doesn't have empty
31343 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
31344 and only the default case doesn't have the required constants. Skip
31346 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
31347 if default_case_nonstandard.
31348 (build_constructors): Build constant 1 just once. Assert that default
31349 values aren't inserted in between cases if contiguous_range. Skip
31351 (build_arrays): Skip virtual phis.
31352 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
31353 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
31354 Handle virtual phis.
31355 (gen_inbound_check): Handle default_case_nonstandard case.
31356 (process_switch): Adjust check_final_bb caller. Call
31357 gather_default_values with the first non-default case instead of
31358 default case if default_case_nonstandard.
31360 2017-05-02 Nathan Sidwell <nathan@acm.org>
31362 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
31363 check. Fix formatting.
31365 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
31367 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
31368 errors when comparing specialized and unspecialized times.
31370 2017-05-02 David Malcolm <dmalcolm@redhat.com>
31372 * diagnostic-show-locus.c
31373 (layout::should_print_annotation_line_p): Make private.
31374 (layout::print_annotation_line): Make private.
31375 (layout::annotation_line_showed_range_p): Make private.
31376 (layout::show_ruler): Make private.
31377 (layout::print_source_line): Make private. Pass in line and
31378 line_width, rather than calling location_get_source_line. Drop
31380 (layout::print_leading_fixits): New method.
31381 (layout::print_any_fixits): Rename to...
31382 (layout::print_trailing_fixits): ...this, and make private.
31383 Don't print newline fixits.
31384 (diagnostic_show_locus): Move logic for printing one row into...
31385 (layout::print_line): ...this new function. Move the
31386 location_get_source_line call and error-handling from
31387 print_source_line to here. Call print_leading_fixits, and rename
31388 print_any_fixits to print_trailing_fixits.
31389 (selftest::test_fixit_insert_containing_newline): Update now that
31390 newlines are partially supported.
31391 (selftest::test_fixit_insert_containing_newline_2): New test.
31392 (selftest::test_fixit_replace_containing_newline): Update comments.
31393 (selftest::diagnostic_show_locus_c_tests): Call the new test.
31394 * edit-context.c (class added_line): New class.
31395 (class edited_line): Describe newline handling in comment.
31396 (edited_line::actually_edited_p): New method.
31397 (edited_line::print_content): Delete redundant decl.
31398 (edited_line::m_predecessors): New field.
31399 (edited_file::print_content): Call edited_line::print_content.
31400 (edited_file::print_diff): Update to support newlines.
31401 (edited_file::print_diff_hunk): Likewise.
31402 (edited_file::print_run_of_changed_lines): New function.
31403 (edited_file::print_diff_line): Convert to...
31404 (print_diff_line): ...this.
31405 (edited_file::get_effective_line_count): New function.
31406 (edited_line::edited_line): Initialize new field m_predecessors.
31407 (edited_line::~edited_line): Clean up m_predecessors.
31408 (edited_line::apply_fixit): Handle newlines.
31409 (edited_line::get_effective_line_count): New function.
31410 (edited_line::print_content): New function.
31411 (edited_line::print_diff_lines): New function.
31412 (selftest::test_applying_fixits_insert_containing_newline): New
31414 (selftest::test_applying_fixits_replace_containing_newline): New
31416 (selftest::insert_line): New function.
31417 (selftest::test_applying_fixits_multiple_lines): Add example of
31419 (selftest::edit_context_c_tests): Call the new tests.
31421 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31423 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
31424 parameter cand. Update dump information.
31425 (get_computation_cost): Update uses.
31427 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31429 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
31430 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
31431 (get_computation_at, rewrite_use_address): Update use of
31432 get_computation_aff.
31434 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31436 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
31437 (get_computation): Delete.
31438 (get_computation_cost): Implement like get_computation_cost_at.
31439 Use get_computation_at.
31440 (get_computation_cost_at): Delete.
31441 (rewrite_use_nonlinear_expr): Use get_computation_at.
31442 (rewrite_use_compare, remove_unused_ivs): Ditto.
31444 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31446 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
31448 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31450 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
31451 (ivopts_global_cost_for_size): Rename parameter and update uses.
31452 (iv_ca_recount_cost): Update uses.
31453 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
31454 candidates seperately in n_invs and n_cands.
31455 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
31457 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31459 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
31460 (find_inv_vars_cb): New.
31461 (find_depends): Renamed to ...
31462 (find_inv_vars): ... this.
31463 (add_candidate_1, force_var_cost): Call find_inv_vars.
31464 (split_address_cost, determine_group_iv_cost_cond): Ditto.
31466 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31468 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
31469 inv_vars. Add inv_exprs.
31470 (struct iv_cand): Rename depends_on to inv_vars.
31471 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
31472 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
31473 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
31474 (dump_cand): Dump inv_vars.
31475 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
31476 (record_invariant, find_depends, add_candidate_1): Ditto.
31477 (set_group_iv_cost, force_var_cost): Ditto.
31478 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
31479 (get_computation_cost_at, get_computation_cost): Ditto.
31480 (determine_group_iv_cost_generic): Ditto.
31481 (determine_group_iv_cost_address): Ditto.
31482 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
31483 (determine_group_iv_costs): Ditto.
31484 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
31485 (iv_ca_set_remove_invariants): Renamed to ...
31486 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
31487 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
31488 (iv_ca_set_add_invariants): Renamed to ...
31489 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
31490 (iv_ca_set_cp): Use iv_ca_set_add_invs.
31491 (iv_ca_has_deps): Support inv_vars and inv_exprs.
31492 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
31493 (create_new_ivs): Remove useless dump.
31495 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31497 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
31499 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
31500 (iv_ca_set_no_cp, create_new_iv): Ditto.
31502 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31504 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
31506 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
31508 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
31509 function tree_check2.
31511 2017-05-02 Martin Liska <mliska@suse.cz>
31513 * doc/gcov.texi: Add missing preposition.
31514 * gcov.c (function_info::function_info): Properly fill up
31515 all member variables.
31517 2017-05-02 Tamar Christina <tamar.christina@arm.com>
31519 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
31521 2017-05-02 Tamar Christina <tamar.christina@arm.com>
31523 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
31525 2017-05-02 Martin Liska <mliska@suse.cz>
31528 * lto-streamer-in.c (lto_read_tree_1): Remove
31529 LTO_STREAMER_DEBUG.
31530 * lto-streamer.c (struct tree_hash_entry): Likewise.
31531 (struct tree_entry_hasher): Likewise.
31532 (tree_entry_hasher::hash): Likewise.
31533 (tree_entry_hasher::equal): Likewise.
31534 (lto_streamer_init): Likewise.
31535 (lto_orig_address_map): Likewise.
31536 (lto_orig_address_get): Likewise.
31537 (lto_orig_address_remove): Likewise.
31538 * lto-streamer.h: Likewise.
31539 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
31540 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
31542 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
31544 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
31545 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
31546 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
31547 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
31548 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
31549 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
31550 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
31551 (mm_maskz_sub_ss): New intrinsics.
31552 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31553 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
31554 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
31555 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
31556 (__builtin_ia32_subss_mask_round): New builtins.
31557 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31558 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
31559 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
31561 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
31562 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
31564 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
31567 2017-05-02 Martin Jambor <mjambor@suse.cz>
31569 PR tree-optimization/78687
31570 * tree-sra.c (access): New field parent.
31571 (process_subtree_disqualification): New function.
31572 (disqualify_candidate): Call it.
31573 (build_accesses_from_assign): Reset write flag if creating an
31575 (build_access_subtree): Fill in parent field and also prpagate
31576 down grp_write flag.
31577 (create_artificial_child_access): New parameter set_grp_write, set
31578 grp_write to its value.
31579 (propagate_subaccesses_across_link): Also propagate grp_write flag
31581 (propagate_all_subaccesses): Push the closest parent back to work
31582 queue if add_access_to_work_queue returned true.
31584 2017-05-02 Richard Biener <rguenther@suse.de>
31586 * common.opt (fstrict-overflow): Alias negative to fwrapv.
31587 * doc/invoke.texi (fstrict-overflow): Remove all traces of
31588 -fstrict-overflow documentation.
31589 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
31590 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
31591 flag_strict_overflow.
31592 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
31593 * lto-opts.c (lto_write_options): Do not stream it.
31594 * lto-wrapper.c (merge_and_complain): Do not handle it.
31595 * opts.c (default_options_table): Do not set -fstrict-overflow.
31596 (finish_options): Likewise do not clear it when sanitizing.
31597 * simplify-rtx.c (simplify_const_relational_operation): Do not
31598 test flag_strict_overflow.
31600 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
31602 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
31603 using enabled attribute.
31604 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
31605 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
31606 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
31607 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
31608 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
31609 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
31610 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
31611 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
31612 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
31613 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
31615 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
31617 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
31619 2017-05-02 Richard Biener <rguenther@suse.de>
31621 PR tree-optimization/80591
31623 2017-04-10 Richard Biener <rguenther@suse.de>
31625 * tree-ssa-structalias.c (find_func_aliases): Properly handle
31628 2017-05-02 Richard Biener <rguenther@suse.de>
31630 PR tree-optimization/80549
31631 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
31632 (cleanup_tree_cfg_noloop): Create forwarders to known loop
31633 headers if they do not have a preheader.
31635 2017-05-02 Martin Liska <mliska@suse.cz>
31638 * common.opt: Fix typo.
31639 * doc/invoke.texi: Likewise.
31641 2017-05-01 Jan Beulich <jbeulich@suse.com>
31643 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
31644 swapping, add (x,x,m,x,n) alternative.
31646 2017-05-01 Nathan Sidwell <nathan@acm.org>
31648 * calls.c (combine_pending_stack_adjustment_and_call): Remove
31649 unnecessary unadjusted_alignment check.
31651 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
31654 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
31656 * gimplify.c (gimplify_cilk_detach): New function.
31657 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
31658 * tree-core.h: Document EXPR_CILK_SPAWN.
31659 * tree.h (EXPR_CILK_SPAWN): Define.
31661 2017-05-01 David Malcolm <dmalcolm@redhat.com>
31663 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
31664 to use new fixit_hint representation, using the "replace" logic.
31665 (get_line_span_for_fixit_hint): Likewise.
31666 (layout::print_any_fixits): Likewise.
31667 (selftest::test_one_liner_many_fixits): Rename to...
31668 (selftest::test_one_liner_many_fixits_1): ...this, and update
31669 comment and expected output to reflect that the multiple fix-it
31670 hints are now consolidated into one insertion.
31671 (selftest::test_one_liner_many_fixits_2): New test.
31672 (selftest::test_diagnostic_show_locus_one_liner): Update for
31674 (selftest::test_fixit_consolidation): Update for fix-it API
31676 * diagnostic.c (print_parseable_fixits): Likewise.
31677 * edit-context.c (edited_line::m_line_events): Convert from
31678 auto_vec <line_event *> to auto_vec <line_event>.
31679 (class line_event): Convert from abstract base class to a concrete
31680 class, taking over the role of replace_event.
31681 (class insert_event): Delete.
31682 (class replace_event): Rename to class line_event. Convert to
31684 (edit_context::add_fixits): Reimplement.
31685 (edit_context::apply_insert): Delete.
31686 (edit_context::apply_replace): Rename to...
31687 (edit_context::apply_fixit): ...this. Convert to half-open range.
31688 (edited_file::apply_insert): Delete.
31689 (edited_file::apply_replace): Rename to...
31690 (edited_file::apply_fixit): ...this.
31691 (edited_line::~edited_line): Drop deletion of events.
31692 (edited_line::apply_insert): Delete.
31693 (edited_line::apply_replace): Rename to...
31694 (edited_line::apply_fixit): ...this. Convert to half-open range.
31695 Update for change to type of m_line_events.
31696 * edit-context.h (edit_context::apply_insert): Delete.
31697 (edit_context::apply_replace): Rename to...
31698 (edit_context::apply_fixit): ...this.
31700 2017-05-01 Martin Sebor <msebor@redhat.com>
31702 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
31705 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
31708 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
31709 __get_cpuid_max returns 0.
31710 (__get_cpuid_count): Ditto.
31712 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
31714 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
31715 replacement expression is another instance of one of its arguments.
31717 2017-05-01 Jakub Jelinek <jakub@redhat.com>
31720 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
31721 check for stack push/pop autoinc.
31722 * config/i386/i386.c (ix86_agi_dependent): Return false
31723 if the only reason why modified_in_p returned true is that
31724 addr is SP based and set_insn is a push or pop.
31726 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
31728 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
31731 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
31734 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
31735 (account_size_time): Use two predicates - exec_pred and
31737 (evaluate_conditions_for_known_args): Compute both clause and
31739 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
31740 (inline_summary_t::duplicate): Update.
31741 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
31743 (compute_inline_parameters): Likewise.
31744 (estimate_edge_size_and_time): Update caluclation of time.
31745 (estimate_node_size_and_time): Compute both time and nonspecialized
31747 (estimate_ipcp_clone_size_and_time): Update.
31748 (inline_merge_summary): Update.
31749 (do_estimate_edge_time): Update.
31750 (do_estimate_edge_size): Update.
31751 (do_estimate_edge_hints): Update.
31752 (inline_read_section, inline_write_summary): Stream both new predicates.
31753 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
31755 (compute_inlined_call_time): Cleanup.
31756 (big_speedup_p): Update.
31757 (edge_badness): Update.
31758 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
31759 (size_time_entry): Replace predicate by exec_predicate and
31760 nonconst_predicate.
31761 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
31762 (estimate_edge_time): Return also nonspec_time.
31763 (reset_edge_growth_cache): Update.
31765 2017-04-29 Jakub Jelinek <jakub@redhat.com>
31767 PR rtl-optimization/80491
31768 * ifcvt.c (noce_process_if_block): When looking for x setter
31769 with missing else_bb, don't check only the insn right before
31770 cond_earliest, but look for the last insn that x is modified in
31771 within the same bb.
31773 PR rtl-optimization/80491
31774 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
31776 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
31778 PR tree-optimization/80487
31779 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
31781 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31783 PR tree-optimization/79697
31784 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
31785 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
31786 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
31788 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
31789 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
31791 2017-04-28 Martin Sebor <msebor@redhat.com>
31793 PR tree-optimization/80523
31794 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
31795 (init_target_to_host_charmap, target_to_host, target_strtol10): New
31797 (maybe_warn, format_directive, parse_directive): Use new functions.
31798 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
31800 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
31802 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
31804 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
31806 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
31807 target_header_dir): Set correctly.
31808 * configure: Regenerated.
31809 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
31810 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
31811 instead of SYSTEM_HEADER_DIR.
31813 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
31815 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
31816 (estimate_local_effects): Likewise.
31817 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
31818 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
31819 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
31820 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
31821 do_estimate_edge_time, estimate_edge_time): Likewise.
31822 * ipa-inline-analysis.c (estimate_node_size_and_time,
31823 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
31824 (estimate_time_after_inlining): Remove.
31826 2017-04-28 Martin Liska <mliska@suse.cz>
31828 * doc/gcov.texi: Enhance documentation of gcov.
31830 2017-04-28 Martin Liska <mliska@suse.cz>
31832 * doc/gcov.texi: Sort options in alphabetic order.
31833 * doc/gcov-dump.texi: Likewise.
31834 * doc/gcov-tool.texi: Likewise.
31835 * gcov.c (print_usage): Likewise.
31836 * gcov-dump.c (print_usage): Likewise.
31837 * gcov-tool.c (print_merge_usage_message): Likewise.
31838 (print_rewrite_usage_message): Likewise.
31839 (print_overlap_usage_message): Likewise.
31841 2017-04-28 Martin Liska <mliska@suse.cz>
31843 PR gcov-profile/53915
31844 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
31846 2017-04-28 Martin Liska <mliska@suse.cz>
31848 PR gcov-profile/79891
31849 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
31850 is marked by compiler as living on a line.
31851 (get_cycles_count): Remove usage of the union.
31852 (output_intermediate_file): Likewise.
31853 (find_source): Fix GNU coding style.
31854 (accumulate_line_counts): Remove old non-all block mode.
31855 (output_lines): Remove usage of the union.
31856 * profile.c (output_location): Include all BBs, even if
31857 belonging to a same line (and file) as a previous BB.
31859 2017-04-28 Martin Liska <mliska@suse.cz>
31861 * gcov.c (process_args): Handle new argument 'w'.
31862 (read_graph_file): Assign ID to BBs.
31863 (output_branch_count): Display BB # if verbose flag is set.
31864 (output_lines): Likewise for arcs.
31865 (print_usage): Add '--verbose' option help.
31866 * doc/gcov.texi: Document --verbose (-w) option.
31868 2017-04-28 Martin Liska <mliska@suse.cz>
31870 * gcov.c (struct block_location_info): New struct.
31871 (process_file): Fill up the new structure.
31872 (read_graph_file): Replace usage of encoding by the newly added
31874 (add_line_counts): Likewise.
31875 (accumulate_line_counts): Remove usage of the union.
31876 (function_info::function_info): New function.
31877 (function_info::~function_info): Likewise.
31878 (process_file): Call delete instead of release_function.
31879 (release_function): Release the function.
31880 (release_structures): Call delete instead of release_function.
31881 (solve_flow_graph): Replace usage of num_blocks.
31882 (find_exception_blocks): Likewise.
31883 (output_lines): Fix GNU coding style.
31885 2017-04-28 Martin Liska <mliska@suse.cz>
31888 * coverage.c (coverage_remove_note_file): New function.
31889 * coverage.h: Declare the function.
31890 * toplev.c (finalize): Clean if an error has been seen.
31892 2017-04-28 Martin Liska <mliska@suse.cz>
31894 PR gcov-profile/80031
31895 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
31896 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
31897 * gcov.c (read_graph_file): Read just number of blocks.
31898 * profile.c (branch_prob): Do not stream 0 flags per a basic
31901 2017-04-28 Martin Liska <mliska@suse.cz>
31903 * gcov-dump.c (tag_*): Add new argument to declarations.
31904 (dump_gcov_file): Likewise.
31905 (tag_blocks): Add and use new argument depth.
31906 (tag_arcs): Likewise.
31907 (tag_lines): Likewise.
31908 (tag_counters): Likewise.
31909 (tag_summary): Likewise.
31910 (dump_working_sets): Use depth to do a proper indentation.
31912 2017-04-28 Jakub Jelinek <jakub@redhat.com>
31915 * cgraph.h (symtab_node::debug_symtab): No longer inline.
31916 * symtab.c (symtab_node::debug_symtab): Move definition here.
31918 2017-04-28 Richard Biener <rguenther@suse.de>
31920 * lto-streamer.h (LTO_major_version): Bump to 7.
31922 2017-04-28 Richard Biener <rguenther@suse.de>
31924 * tree-vrp.c (assert_info): New struct.
31925 (add_assert_info): New helper.
31926 (register_edge_assert_for_2): Refactor to add asserts to a vector
31928 (register_edge_assert_for_1): Likewise.
31929 (register_edge_assert_for): Likewise.
31930 (finish_register_edge_assert_for): New helper actually registering
31931 asserts where live on edge.
31932 (find_conditional_asserts): Adjust.
31933 (find_switch_asserts): Likewise.
31934 (evrp_dom_walker::try_find_new_range): Generalize.
31935 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
31937 2017-04-27 Marek Polacek <polacek@redhat.com>
31940 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
31941 arg10 and arg11 to itype.
31943 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
31945 * doc/extend.texi (Object Size Checking): Improve grammar.
31947 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
31950 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
31951 that the logic for permitting reciprocal estimates matches that
31954 2017-04-27 Jakub Jelinek <jakub@redhat.com>
31957 * tree.c (type_cache_hasher::equal): Only compare
31958 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
31959 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
31960 non-aggregate element types.
31961 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
31962 about the flag on ARRAY_TYPEs in the comment, formatting fix.
31964 2017-04-27 Richard Biener <rguenther@suse.de>
31966 PR middle-end/80533
31967 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
31968 stripping ARRAY_REFs from MEM_EXPR make sure we're not
31969 keeping a reference to a trailing array.
31971 2017-04-27 Richard Biener <rguenther@suse.de>
31973 PR middle-end/80539
31974 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
31975 being in loop-closed SSA form conservatively.
31976 (chrec_fold_multiply_poly_poly): Likewise.
31978 2017-04-27 Tamar Christina <tamar.christina@arm.com>
31980 PR middle-end/79665
31981 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
31982 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
31984 2017-04-27 Jakub Jelinek <jakub@redhat.com>
31987 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
31988 (aarch64_function_arg_alignment): Return unsigned int again, but still
31989 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
31990 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
31991 Don't emit -Wpsabi note.
31992 (aarch64_function_arg_boundary): Likewise.
31993 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
31996 2017-04-26 Nathan Sidwell <nathan@acm.org>
31998 * tree.h (crc32_unsigned_n): Declare.
31999 (crc32_unsigned, crc32_unsigned): Make inline.
32000 * tree.c (crc32_unsigned_bits): Replace with ...
32001 (crc32_unsigned_n): ... this.
32002 (crc32_unsigned, crc32_byte): Remove.
32003 (crc32_string): Remove unnecessary braces.
32005 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
32007 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
32008 * ipa-inline-analysis.c (MAX_TIME): Remove.
32009 (account_size_time): Use sreal for time.
32010 (dump_inline_summary): Update.
32011 (estimate_function_body_sizes): Update.
32012 (estimate_edge_size_and_time): Update.
32013 (estimate_calls_size_and_time): Update.
32014 (estimate_node_size_and_time): Update.
32015 (inline_merge_summary): Update.
32016 (inline_update_overall_summary): Update.
32017 (estimate_time_after_inlining): Update.
32018 (inline_read_section): Update.
32019 (inline_write_summary): Update.
32020 * ipa-inline.c (compute_uninlined_call_time): Update.
32021 (compute_inlined_call_time): Update.
32022 (recursive_inlining): Update.
32023 (inline_small_functions): Update.
32024 (dump_overall_stats): Update.
32025 * ipa-inline.h: Include sreal.h.
32026 (size_time_entry): Turn time to sreal.
32027 (inline_summary): Turn self_time nad time to sreal.
32029 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
32031 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
32033 (sreal::stream_out, sreal::stream_in): New.
32034 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
32036 2017-04-25 Jakub Jelinek <jakub@redhat.com>
32038 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
32041 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
32044 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
32046 (dimode_scalar_chain::compute_convert_gain): Ditto.
32047 (dimode_scalar_chain::make_vector_copies): Ditto.
32048 (dimode_scalar_chain::convert_reg): Ditto.
32049 (dimode_scalar_chain::convert_insn): Ditto.
32050 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
32051 (VI248_AVX512BW_1): New mode iterator.
32052 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
32053 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
32056 2017-04-25 Martin Sebor <msebor@redhat.com>
32058 PR tree-optimization/80497
32059 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
32060 constants are representable in HOST_WIDE_INT.
32061 (parse_directive): Ditto.
32063 2017-04-25 Martin Sebor <msebor@redhat.com>
32066 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
32067 (new_zero_array): Adjust signature.
32068 (dom_info::dom_init): Used unsigned rather that size_t.
32069 (dom_info::dom_info): Same.
32071 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
32072 Jakub Jelinek <jakub@redhat.com>
32075 * config/arm/arm.c: Include gimple.h.
32076 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
32077 returns negative, increment ncrn only if it returned positive.
32078 (arm_needs_doubleword_align): Return int instead of bool,
32079 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
32080 members, but if there is any such non-FIELD_DECL
32081 > PARM_BOUNDARY aligned decl, return -1 instead of false.
32082 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
32083 returns negative, increment nregs only if it returned positive.
32084 (arm_setup_incoming_varargs): Likewise.
32085 (arm_function_arg_boundary): Emit -Wpsabi note if
32086 arm_needs_doubleword_align returns negative, return
32087 DOUBLEWORD_ALIGNMENT only if it returned positive.
32089 2017-04-25 Marek Polacek <polacek@redhat.com>
32092 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
32093 first argument to type.
32095 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
32098 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
32099 type checks to test for compatibility instead of equality.
32101 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
32102 Jakub Jelinek <jakub@redhat.com>
32105 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
32107 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
32108 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
32109 the alignment computation, but return their maximum in warn_alignment.
32110 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
32111 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
32113 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
32114 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
32117 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32119 * config/arc/simdext.md (dmpyh): Fix typo.
32121 2017-04-25 Richard Biener <rguenther@suse.de>
32123 PR tree-optimization/80492
32124 * alias.c (compare_base_decls): Handle registers with asm
32125 specification conservatively.
32126 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
32127 compare_base_decls returning dont-know properly.
32129 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32131 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
32132 (legitimate_offset_address_p): New function.
32133 (arc_legitimate_address_p): Use above function.
32135 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32137 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
32139 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32141 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
32142 ACCH registers whenever they are available.
32144 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32146 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
32147 double regs fix when not used.
32149 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32151 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
32153 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
32154 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
32156 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32158 * config/arc/arc.c (arc_output_addsi): Check for h-register class
32159 when emitting short ADD instructions.
32161 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32163 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
32165 (cmpsi_cc_c_insn): Likewise.
32166 (cbranchsi4_scratch): Compute proper instruction length using
32167 compact_hreg_operand.
32168 * config/arc/predicates.md (compact_hreg_operand): New predicate.
32170 2017-04-25 Richard Biener <rguenther@suse.de>
32172 PR middle-end/80509
32173 * passes.c (pass_manager::pass_manager): Initialize
32174 m_name_to_pass_map.
32176 2017-04-25 Richard Biener <rguenther@suse.de>
32178 PR tree-optimization/79201
32179 * tree-ssa-sink.c (statement_sink_location): Handle calls.
32181 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32184 * config/s390/vector.md: Split MEM->GPR vector moves for
32185 non-s_operand addresses.
32187 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32190 * config/s390/predicates.md (reload_const_wide_int_operand): New
32192 * config/s390/s390.md ("movti"): Remove d/P alternative.
32193 ("movti_bigconst"): New pattern definition.
32195 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
32198 * s390-protos.h (s390_expand_cs_hqi): Removed.
32199 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
32200 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
32201 modes as well as CCZ1mode and CCZmode.
32202 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
32203 signature of s390_emit_compare_and_swap.
32204 (s390_expand_cs_hqi): Likewise, make static.
32205 (s390_expand_cs_tdsi): Generate an explicit compare before trying
32206 compare-and-swap, in some cases.
32207 (s390_expand_cs): Wrapper function.
32208 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
32210 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
32211 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
32212 patterns for small and large integers. Forbid symref memory operands.
32213 Move expander to s390.c. Require cc register.
32214 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
32215 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
32216 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
32217 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
32218 symref memory operands. Remove CC mode and call s390_match_ccmode
32220 ("atomic_exchange<mode>"): Allow and implement all integer modes.
32222 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
32224 * config/s390/s390.md (define_peephole2): New peephole to help
32225 combining the load-and-test pattern with volatile memory.
32227 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
32229 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
32230 with CCZmode for TARGET_Z196.
32232 2017-04-25 Jakub Jelinek <jakub@redhat.com>
32234 PR rtl-optimization/80501
32235 * combine.c (make_compound_operation_int): Set subreg_code to SET
32236 even for AND with mask of the sign bit of mode.
32238 PR rtl-optimization/80500
32239 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
32240 sum's initial value.
32242 2017-04-25 Julian Brown <julian@codesourcery.com>
32243 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32245 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
32247 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
32249 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
32251 2017-04-25 Julian Brown <julian@codesourcery.com>
32252 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32254 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
32255 (thunderx2t99_sha): New Reservation.
32257 2017-04-25 Julian Brown <julian@codesourcery.com>
32258 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32260 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
32261 type for 1-element load.
32263 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
32265 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
32267 2017-04-24 Martin Jambor <mjambor@suse.cz>
32269 PR tree-optimization/80293
32270 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
32271 char arrays not totally scalarizable if it is false.
32272 (analyze_all_variable_accesses): Pass correct value in the new
32273 parameter. Add a statistics counter.
32275 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
32277 PR middle-end/79931
32278 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
32280 2017-04-24 Richard Biener <rguenther@suse.de>
32282 PR tree-optimization/80494
32283 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
32284 out for complex types.
32286 2017-04-24 Richard Biener <rguenther@suse.de>
32288 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
32289 * tree-ssa-sccvn.c (print_scc): Print SCC size.
32290 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
32291 (DFS): Adjust and never fail.
32292 (sccvn_dom_walker::fail): Remove.
32293 (sccvn_dom_walker::before_dom_children): Adjust.
32294 (run_scc_vn): Likewise and never fail.
32295 * tree-ssa-pre.c (pass_pre::execute): Adjust.
32296 (pass_fre::execute): Likewise.
32298 2017-04-24 Richard Biener <rguenther@suse.de>
32300 PR tree-optimization/79725
32301 * tree-ssa-sink.c (statement_sink_location): Return whether
32302 failure reason was zero uses. Move that check later.
32303 (sink_code_in_bb): Deal with zero uses by removing the stmt
32306 2017-04-24 Richard Biener <rguenther@suse.de>
32309 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
32310 pointer-based references.
32312 2017-04-24 Richard Biener <rguenther@suse.de>
32315 * pass_manager.h (pass_manager::operator new): Remove.
32316 (pass_manager::operator delete): Likewise.
32317 * passes.c (pass_manager::operator new): Remove.
32318 (pass_manager::operator delete): Likewise.
32319 (pass_manager::pass_manager): Zero individual pass members.
32321 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
32324 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
32325 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
32326 Check "XEXP (src, 1)" operand here.
32327 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
32328 Check "XEXP (src, 1)" operand here.
32329 (dimode_scalar_chain::make_vector_copies): Detect count register
32330 of a shift instruction. Zero extend count register from QImode
32331 to DImode to satisfy vector shift pattern count operand predicate.
32332 Substitute vector shift count operand with a DImode copy.
32333 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
32336 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
32338 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
32339 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
32340 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
32341 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
32342 (UNSPEC_NOREX_MEM): Remove definition.
32344 2017-04-21 Richard Biener <rguenther@suse.de>
32346 PR tree-optimization/79547
32347 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
32348 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
32349 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
32350 without any constraints.
32352 2017-04-21 Richard Biener <rguenther@suse.de>
32354 PR tree-optimization/78847
32355 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
32357 2017-04-21 Richard Biener <rguenther@suse.de>
32359 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
32360 (build_distinct_type_copy): Likewise.
32361 (build_variant_type_copy): Likewise.
32362 * tree.c (build_qualified_type): Pass down mem-stat info.
32363 (build_distinct_type_copy): Likewise.
32364 (build_variant_type_copy): Likewise.
32366 2017-04-21 Richard Biener <rguenther@suse.de>
32368 PR tree-optimization/80237
32369 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
32371 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
32372 for a simplified result.
32374 2016-04-21 Richard Biener <rguenther@suse.de>
32376 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
32377 sth as strict as a simple_iv but a chrec without symbols and an
32378 operand defined in the loop we are peeling (and not some subloop).
32379 (propagate_constants_for_unrolling): Propagate all constants.
32381 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
32384 * config/i386/i386.c (print_reg): Remove assert for disalowed
32385 regno values, call output_operand_lossage instead.
32387 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
32390 * config/i386/constraints.md (Yc): New register constraint.
32391 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
32392 Use Yc constraint for alternative 2 of operand 0. Remove
32393 preferred_for_speed attribute.
32395 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
32397 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
32398 lastprivate clauses in SIMT case.
32400 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
32402 * doc/invoke.texi (-Wextra-semi): Document new warning option.
32404 2017-04-20 Richard Biener <rguenther@suse.de>
32406 PR tree-optimization/57796
32407 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
32408 as N scalar stores.
32409 (vect_model_load_cost): Cost gathers as N scalar loads.
32411 2017-04-20 Richard Biener <rguenther@suse.de>
32413 * ggc-page.c (ggc_allocated_p): Rename to ...
32414 (safe_lookup_page_table_entry): ... this and return the lookup
32416 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
32418 2017-04-20 Richard Biener <rguenther@suse.de>
32420 PR tree-optimization/80453
32421 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
32422 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
32423 from the conditions.
32424 (vn_phi_eq): Pass them down.
32425 (vn_phi_lookup): Record them.
32426 (vn_phi_insert): Likewise.
32428 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
32430 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
32431 uninitialized variable warning to avoid buffer overrun.
32433 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
32436 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
32437 is suppressed for '{ 0 }' in C.
32439 2017-04-20 Jakub Jelinek <jakub@redhat.com>
32441 * BASE-VER: Set to 8.0.0.
32443 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
32445 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
32446 priority .init_array and .fini_array section with SECTION_NOTYPE
32449 2017-04-20 Jakub Jelinek <jakub@redhat.com>
32451 PR middle-end/80423
32452 * tree.h (build_array_type): Add typeless_storage default argument.
32453 * tree.c (type_cache_hasher::equal): Also compare
32454 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
32455 (build_array_type): Add typeless_storage argument, set
32456 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
32458 (build_nonshared_array_type): Adjust build_array_type_1 caller.
32459 (build_array_type): Likewise. Add typeless_storage argument.
32461 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
32462 Jakub Jelinek <jakub@redhat.com>
32464 PR tree-optimization/80426
32465 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
32466 operation on symbolic operands, also compute the overflow for the
32467 invariant part when the operation degenerates into a negation.
32469 2017-04-19 Jakub Jelinek <jakub@redhat.com>
32472 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
32473 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
32476 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
32478 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
32481 * config/avr/avr.c (tree.h): Include it.
32482 (cgraph.h): Include it.
32483 (avr_encode_section_info): Don't warn for uninitialized progmem
32484 variable if it's just an alias.
32486 2017-04-19 Richard Biener <rguenther@suse.de>
32489 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
32490 when needed by AutoPGO.
32492 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
32495 * doc/lto.texi: Remove an extra 'that'.
32497 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
32499 PR rtl-optimization/80429
32500 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
32501 are only used in debug insns.
32503 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
32504 Vladimir Makarov <vmakarov@redhat.com>
32506 * config/sparc/predicates.md (input_operand): Add comment. Return
32507 true for any memory operand when LRA is in progress.
32508 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
32510 2017-04-18 Jeff Law <law@redhat.com>
32513 * mips.md ({return,simple_return}_internal): Do not overwrite
32516 2017-04-18 Jakub Jelinek <jakub@redhat.com>
32518 PR tree-optimization/80443
32519 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
32520 instead of adding 1, subtract -1 and similarly instead of subtracting
32523 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
32525 PR rtl-optimization/80357
32526 * haifa-sched.c (tmp_bitmap): New variable.
32527 (model_recompute): Handle duplicate use records.
32528 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
32529 (free_global_sched_pressure_data): Free it.
32531 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
32534 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
32535 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
32536 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
32537 instead of SYSTEM_HEADER_DIR.
32539 2017-04-18 Jeff Law <law@redhat.com>
32541 PR middle-end/80422
32542 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
32543 predecessors after walking up the insn chain.
32545 2017-04-18 Jakub Jelinek <jakub@redhat.com>
32548 * dwarf2out.c (modified_type_die): Try harder not to emit internal
32549 sizetype type into debug info.
32551 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
32554 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
32555 unneeded test for TARGET_UPPER_REGS_SF.
32556 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
32558 2017-04-18 Jakub Jelinek <jakub@redhat.com>
32561 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
32562 instead of gsi_after_labels.
32564 2017-04-18 Jeff Law <law@redhat.com>
32566 * regcprop.c (maybe_mode_change): Avoid creating copies of the
32570 2017-04-13 Jeff Law <law@redhat.com>
32571 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
32572 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
32574 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
32577 * config/avr/avr.c (intl.h): Include it.
32578 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
32580 2017-04-18 Martin Liska <mliska@suse.cz>
32582 PR gcov-profile/78783
32583 * gcov-tool.c (gcov_output_files): Validate that destination
32584 file is either removed by the tool or by a user.
32586 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
32587 Guy Benyei <guybe@mellanox.com>
32589 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
32590 block, and do not negate it, the stored id is already negative.
32592 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
32594 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
32596 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
32599 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
32600 masks of options that should be turned off if the VSX vector
32601 options are turned off.
32602 (OTHER_P8_VECTOR_MASKS): Likewise.
32603 (OTHER_VSX_VECTOR_MASKS): Likewise.
32604 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
32605 rs6000_disable_incompatible_switches to validate no type switches
32607 (rs6000_incompatible_switch): New function to disallow turning on
32608 other vector options if -mno-vsx, -mno-power8-vector, or
32609 -mno-power9-vector are specified.
32611 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
32613 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
32615 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
32617 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
32618 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
32619 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
32620 (ARG_POINTER_CFA_OFFSET): Likewise.
32622 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
32624 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
32625 conditions to take advantage of various optimizations.
32627 2017-04-13 Jeff Law <law@redhat.com>
32629 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
32630 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
32631 (zero_extendsidi2_dext): Likewise.
32633 2017-04-13 Jakub Jelinek <jakub@redhat.com>
32636 * fold-const.c (fold_ternary_loc): Revert
32637 use op0 instead of fold_convert_loc (loc, type, arg0) part of
32640 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
32642 PR rtl-optimization/80343
32643 * lra-remat.c (update_scratch_ops): Assign original hard reg to
32644 new scratch pseudo.
32646 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
32649 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
32650 to ubsan_encode_value.
32652 2017-04-13 Jeff Law <law@redhat.com>
32654 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
32655 appearing in DEBUG_INSNs.
32657 2017-04-13 Martin Liska <mliska@suse.cz>
32659 PR gcov-profile/80413
32660 * gcov-io.c (gcov_write_string): Copy to buffer just when
32661 allocated size is greater than zero.
32663 2017-04-13 Jakub Jelinek <jakub@redhat.com>
32666 * dwarf2out.c (decls_for_scope): Ignore declarations of
32667 current_function_decl in BLOCK_NONLOCALIZED_VARS.
32669 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
32672 * ipa-visibility.c (non_local_p): Fix typos.
32673 (localize_node): When localizing symbol in same comdat group,
32674 dissolve the group only when we know external symbols are going
32676 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
32678 2017-04-12 Jakub Jelinek <jakub@redhat.com>
32680 PR tree-optimization/79390
32681 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
32682 order does not result in usable sequence, retry with reversed operand
32688 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
32689 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
32690 op0 instead of fold_convert_loc (loc, type, arg0).
32692 2017-04-12 Jeff Law <law@redhat.com>
32694 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
32695 has a delay slot in the generated code.
32697 * config/cris/cris.md (cris_preferred_reload_class): Return
32698 GENNONACR_REGS rather than GENERAL_REGS.
32700 2017-04-12 Jakub Jelinek <jakub@redhat.com>
32703 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
32704 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
32705 signedness of the result type.
32707 2017-04-12 Richard Biener <rguenther@suse.de>
32708 Jeff Law <law@redhat.com>
32710 PR tree-optimization/80359
32711 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
32712 trim stores to TARGET_MEM_REFs.
32714 2017-04-12 Richard Biener <rguenther@suse.de>
32716 PR tree-optimization/79390
32717 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
32718 threading case even more.
32720 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
32723 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
32724 for quad_address_p for TImode, instead of just not indexed_address.
32726 2017-04-12 Richard Biener <rguenther@suse.de>
32727 Bernd Edlinger <bernd.edlinger@hotmail.de>
32729 PR middle-end/79671
32730 * alias.c (component_uses_parent_alias_set_from): Handle
32731 TYPE_TYPELESS_STORAGE.
32732 (get_alias_set): Likewise.
32733 * tree-core.h (tree_type_common): Add typeless_storage flag.
32734 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
32735 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
32736 for types containing members with TYPE_TYPELESS_STORAGE.
32737 (place_field): Likewise.
32738 (layout_type): Likewise for ARRAY_TYPE.
32739 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
32740 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
32741 TYPE_TYPELESS_STORAGE.
32742 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
32744 2017-04-12 Jakub Jelinek <jakub@redhat.com>
32747 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
32748 first argument to type.
32750 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32754 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
32755 CONST0_RTX (mode) rather than const0_rtx where appropriate.
32756 (rs6000_expand_binop_builtin): Likewise.
32757 (rs6000_expand_ternop_builtin): Likewise; also add missing
32758 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
32759 vshasigma built-ins.
32760 * doc/extend.texi: Document that vec_xxpermdi's third argument
32761 must be a constant.
32763 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
32765 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
32766 Use shift_const cost parameter when calculating gain of STV shifts.
32768 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
32770 PR rtl-optimization/70478
32771 * lra-constraints.c (process_alt_operands): Check memory for
32772 disfavoring memory insn operand.
32774 2017-04-11 Jakub Jelinek <jakub@redhat.com>
32776 PR middle-end/80100
32777 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
32778 left shift in unsigned HOST_WIDE_INT type.
32780 PR rtl-optimization/80385
32781 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
32782 (not (neg X)) into (plus X -1) for complex or non-integral modes.
32785 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
32786 if they have any depend clauses.
32788 2017-04-11 Martin Liska <mliska@suse.cz>
32791 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
32792 * ipa-split.c (split_function): Create a local comdat symbol
32793 if caller is in a comdat group.
32795 2017-04-11 Martin Liska <mliska@suse.cz>
32798 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
32801 2017-04-11 Martin Sebor <msebor@redhat.com>
32803 PR middle-end/80364
32804 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
32805 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
32807 (directive::set_width, directive::set_precision, format_character):
32809 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
32812 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
32815 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
32816 conflict, set target->arch_name instead of target->cpu_name.
32818 2017-04-11 Richard Biener <rguenther@suse.de>
32820 PR tree-optimization/80374
32821 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
32822 build_zero_cst, remove fold_convertible_p check again.
32824 2017-04-11 Martin Liska <mliska@suse.cz>
32827 * ubsan.c (instrument_object_size): Do not instrument register
32830 2017-04-11 Jakub Jelinek <jakub@redhat.com>
32833 * config/i386/i386-builtin-types.def
32834 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
32835 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
32836 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
32837 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
32838 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
32839 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
32840 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
32841 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
32842 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
32843 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
32844 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
32845 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
32846 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
32847 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
32848 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
32849 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
32850 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
32851 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
32852 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
32853 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
32854 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
32855 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
32856 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
32857 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
32858 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
32859 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
32860 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
32861 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
32862 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
32863 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
32864 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
32865 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
32866 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
32867 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
32868 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
32869 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
32870 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
32871 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
32872 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
32873 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
32874 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
32875 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
32876 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
32877 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
32878 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
32879 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
32881 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
32882 flag to second_arg_count, handle 4 argument function type _COUNT
32883 aliases, handle second_arg_count on second argument rather than last.
32885 2017-04-10 Jeff Law <law@redhat.com>
32887 PR tree-optimization/80374
32888 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
32889 record anything if we can not convert integer_zero_node to the
32892 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
32895 * config/rs6000/rs6000.c (rs6000_option_override_internal):
32896 Enhance special handling given to the TARGET_P9_MINMAX option in
32897 relation to certain other options.
32899 2017-04-10 Bin Cheng <bin.cheng@arm.com>
32901 PR tree-optimization/80153
32902 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
32903 remove POINTER_PLUS_EXPR's base part directly, rather than through
32906 2017-04-10 Richard Biener <rguenther@suse.de>
32907 Bin Cheng <bin.cheng@arm.com>
32909 PR tree-optimization/80153
32910 * tree-affine.c (aff_combination_to_tree): Get base pointer from
32911 the first element of pointer type aff_tree. Build result expr in
32913 (add_elt_to_tree): Convert to type unconditionally. Remove other
32914 fold_convert calls.
32915 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
32916 (rewrite_use_nonlinear_expr): Check invariant using iv information.
32918 2017-04-10 Richard Biener <rguenther@suse.de>
32920 * tree-ssa-structalias.c (find_func_aliases): Properly handle
32923 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
32925 PR rtl-optimization/70478
32926 * lra-constraints.c (curr_small_class_check): New.
32927 (update_and_check_small_class_inputs): New.
32928 (process_alt_operands): Update curr_small_class_check. Disfavor
32929 alternative insn memory operands. Check available regs for small
32932 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
32935 * config/mips/mips.opt (-mvirt): Update description.
32936 * doc/invoke.texi (-mvirt): Likewise.
32938 2017-04-10 Richard Biener <rguenther@suse.de>
32940 PR middle-end/80362
32941 * fold-const.c (fold_binary_loc): Look at unstripped ops when
32942 looking for NEGATE_EXPR in -A / -B to A / B folding.
32944 2017-04-10 Martin Liska <mliska@suse.cz>
32946 PR gcov-profile/80224
32947 * gcov.c (print_usage): Fix usage string.
32948 (get_gcov_intermediate_filename): Remove.
32949 (output_gcov_file): Use both for normal and intermediate format.
32950 (generate_results): Do not initialize special file for
32951 intermediate format.
32953 2017-04-10 Richard Biener <rguenther@suse.de>
32955 PR tree-optimization/80304
32956 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
32959 2017-04-10 Nathan Sidwell <nathan@acm.org>
32962 * config/rs6000/rs6000.c (rs6000_vector_type): New.
32963 (rs6000_init_builtins): Use it.
32965 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32967 * config/arm/arm.md (<mrc>): Add mode to SET source.
32968 (<mrrc>): Likewise.
32970 2017-04-10 Richard Biener <rguenther@suse.de>
32972 PR middle-end/80344
32973 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
32975 2017-04-10 Jakub Jelinek <jakub@redhat.com>
32978 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
32979 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
32980 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
32981 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
32982 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
32983 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
32984 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
32985 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
32986 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
32987 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
32988 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
32989 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
32990 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
32991 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
32992 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
32993 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
32994 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
32995 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
32996 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
32997 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
32998 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
32999 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
33000 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
33002 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
33004 PR rtl-optimization/70478
33005 * lra-constraints.c: Reverse the last patch.
33007 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
33009 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
33010 Add comment for WCHAR_T.
33012 2017-04-08 Martin Liska <mliska@suse.cz>
33015 2017-04-07 Martin Liska <mliska@suse.cz>
33018 * ipa-split.c (split_function): Add function part to a same comdat
33021 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33024 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
33026 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
33028 * rs6000/rs6000.c (vec_load_pendulum): Rename...
33029 (vec_pairing): ...to this.
33030 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
33031 (rs6000_sched_init): Adjust for name change.
33032 (struct rs6000_sched_context): Likewise.
33033 (rs6000_init_sched_context): Likewise.
33034 (rs6000_set_sched_context): Likewise.
33036 2017-04-07 Jakub Jelinek <jakub@redhat.com>
33042 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
33044 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
33045 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
33046 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
33048 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
33050 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
33052 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
33054 PR rtl-optimization/70703
33055 * ira-color.c (update_conflict_hard_regno_costs): Use
33056 int64_t instead of HOST_WIDE_INT.
33058 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
33060 PR rtl-optimization/70478
33061 * lra-constraints.c (process_alt_operands): Disfavor alternative
33062 insn memory operands.
33064 2017-04-07 Jeff Law <law@redhat.com>
33066 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
33067 CALL and NOTE_INSN_CALL_ARG_LOCATION.
33069 2017-04-07 Martin Liska <mliska@suse.cz>
33072 * config/aarch64/aarch64.c (aarch64_process_target_attr):
33073 Show error message instead of an ICE.
33075 2017-04-07 Martin Liska <mliska@suse.cz>
33078 * ipa-split.c (split_function): Add function part to a same comdat
33081 2017-04-07 Richard Biener <rguenther@suse.de>
33083 PR middle-end/80341
33084 * tree.c (get_unwidened): Also handle ! for_type case for
33086 * convert.c (do_narrow): Split out from ...
33087 (convert_to_integer_1): ... here. Do not pass final truncation
33088 type to get_unwidened for TRUNC_DIV_EXPR.
33090 2017-04-07 Richard Biener <rguenther@suse.de>
33092 * tree-affine.c (wide_int_ext_for_comb): Take type rather
33094 (aff_combination_const): Adjust.
33095 (aff_combination_scale): Likewise.
33096 (aff_combination_add_elt): Likewise.
33097 (aff_combination_add_cst): Likewise.
33098 (aff_combination_convert): Likewise.
33099 (add_elt_to_tree): Likewise. Remove unused argument.
33100 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
33102 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
33104 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
33106 * config/arm/arm.c (arm_default_short_enums): Use
33107 ARM_DEFAULT_SHORT_ENUMS.
33108 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
33110 2017-04-06 Jakub Jelinek <jakub@redhat.com>
33113 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
33114 members with redundant out-of-class redeclaration.
33116 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
33119 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
33120 * config/i386/i386.md (*zero_extendsidi2):
33121 Add (?*x,*x) and (?*v,*v) alternatives.
33123 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
33126 * config/i386/i386.c (ix86_expand_builtin)
33127 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
33128 mode from insn data. Convert operands to insn operand mode.
33129 Copy operands that don't satisfy insn predicate to a register.
33131 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
33133 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
33136 2017-04-06 Richard Biener <rguenther@suse.de>
33138 PR tree-optimization/80334
33139 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
33140 preserve alignment of accesses.
33142 2017-04-06 Richard Biener <rguenther@suse.de>
33144 PR tree-optimization/80262
33145 * tree-sra.c (build_ref_for_offset): Preserve address-space
33147 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
33148 Drop useless address-space information on MEM_REF offsets.
33150 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
33152 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
33154 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
33156 PR rtl-optimization/70703
33157 * ira-color.c (update_conflict_hard_regno_costs): Use
33158 HOST_WIDE_INT instead of long.
33160 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
33163 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
33164 not defined for x86_64 target. Add -mmmx target option when __SSE2__
33166 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
33167 for x86_64 target. Handle -m3dnowa option.
33169 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
33171 PR rtl-optimization/70703
33172 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
33173 (update_conflict_hard_regno_costs): Use long instead of unsigned
33174 arithmetic for cost calculation.
33176 2017-04-05 Jakub Jelinek <jakub@redhat.com>
33177 Bernd Edlinger <bernd.edlinger@hotmail.de>
33180 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
33183 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
33186 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
33187 ptr_mode with Pmode throughout.
33188 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
33189 into probe_stack_range and use DImode.
33191 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
33194 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
33195 call_eh_return is true.
33197 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33199 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
33200 Initialize last_match_fntype_index.
33202 2017-04-05 Jakub Jelinek <jakub@redhat.com>
33205 * tree-nvr.c: Include internal-fn.h.
33206 (pass_return_slot::execute): Ignore internal calls without
33209 2017-04-04 Jakub Jelinek <jakub@redhat.com>
33210 Richard Biener <rguenther@suse.de>
33213 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
33214 captures used multiple times, except for the last use.
33215 * generic-match-head.c: Include gimplify.h.
33217 2017-04-04 Jakub Jelinek <jakub@redhat.com>
33219 PR tree-optimization/79390
33220 * target.h (struct noce_if_info): Declare.
33221 * targhooks.h (default_noce_conversion_profitable_p): Declare.
33222 * target.def (noce_conversion_profitable_p): New target hook.
33223 * ifcvt.h (struct noce_if_info): New type, moved from ...
33224 * ifcvt.c (struct noce_if_info): ... here.
33225 (noce_conversion_profitable_p): Renamed to ...
33226 (default_noce_conversion_profitable_p): ... this. No longer
33228 (noce_try_store_flag_constants, noce_try_addcc,
33229 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
33230 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
33231 instead of noce_conversion_profitable_p.
33232 * config/i386/i386.c: Include ifcvt.h.
33233 (ix86_option_override_internal): Don't override
33234 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
33235 (ix86_noce_conversion_profitable_p): New function.
33236 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
33237 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
33238 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
33239 * doc/tm.texi: Regenerated.
33241 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33243 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
33246 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
33249 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
33250 instructions for small multiply cores.
33252 2017-04-04 Jeff Law <law@redhat.com>
33254 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
33256 (mips_expand_vec_perm_const): Initialize elements in orig_perm
33257 that are not set by the loop over the elements.
33259 2017-04-04 Jakub Jelinek <jakub@redhat.com>
33262 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
33263 int mode, convert_modes it to mode as unsigned, otherwise use
33264 lowpart_subreg to mode rather than SImode.
33265 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
33266 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
33267 Use DImode instead of SImode for the shift count operand.
33268 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
33271 2017-04-04 Richard Biener <rguenther@suse.de>
33273 PR middle-end/80281
33274 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
33275 arithmetic done for the negate or the plus. Simplify.
33276 (A - (-B) -> A + B): Likewise.
33277 * fold-const.c (split_tree): Make sure to not negate pointers.
33279 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
33281 PR rtl-optimization/60818
33282 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
33283 a compare of comparisons with the thing compared if this results
33284 in a different machine mode.
33286 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
33288 * alias.c (base_alias_check): Fix typo in comment.
33289 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
33290 * cgraphunit.c (symbol_table::compile): Likewise.
33291 * collect2.c (maybe_run_lto_and_relink): Likewise.
33292 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
33293 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
33294 * config/avr/avr.c (avr_map_op_t): Likewise.
33295 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
33296 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
33297 * config/epiphany/epiphany.md (movcc): Likewise.
33298 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
33299 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
33301 * config/mips/mips.c (mips_save_restore_reg): Likewise.
33302 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
33303 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
33304 * config/sh/sh.c (sh_rtx_costs): Likewise.
33305 * fold-const.c (fold_truth_andor): Likewise.
33306 * genautomata.c (collapse_flag): Likewise.
33307 * gengtype.h (struct type::u::s): Likewise.
33308 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
33309 * input.c (FORMAT_AMOUNT): Likewise.
33310 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
33311 (known_aggs_to_agg_replacement_list): Likewise.
33312 * ipa-inline-analysis.c: Likewise.
33313 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
33314 * ipa-polymorphic-call.c
33315 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
33316 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
33317 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
33319 * modulo-sched.c (apply_reg_moves): Likewise.
33320 * omp-expand.c (build_omp_regions_1): Likewise.
33321 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
33322 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
33323 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
33324 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
33325 * value-prof.c: Likewise.
33326 * var-tracking.c (val_reset): Likewise.
33328 2017-04-03 Richard Biener <rguenther@suse.de>
33330 PR tree-optimization/80275
33331 * fold-const.c (split_address_to_core_and_offset): Handle
33334 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
33336 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
33337 descriptors is at least equal to that of functions.
33339 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
33341 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
33343 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
33346 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
33347 (mov<IMOD4:mode>): New expander.
33348 (*mov<IMOD4:mode>_internal): New insn and split pattern.
33350 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
33352 PR rtl-optimization/79405
33353 * fwprop.c (propagations_left): New variable.
33354 (forward_propagate_into): Decrement it.
33355 (fwprop_init): Initialize it.
33356 (fw_prop): If the variable has reached zero, stop propagating.
33357 (fwprop_addr): Ditto.
33359 2017-03-31 Jakub Jelinek <jakub@redhat.com>
33362 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
33363 a FUNCTION_DECL, pass it as decl instead of origin to
33366 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
33368 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
33371 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
33374 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
33375 TARGET_VSX_SMALL_INTEGER.
33377 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33379 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
33380 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
33382 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
33384 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
33385 extraction from odd-numbered MSA register.
33387 2017-03-31 Jakub Jelinek <jakub@redhat.com>
33389 PR middle-end/80173
33390 * expmed.c (store_bit_field_1): Don't attempt to create
33391 a word subreg out of hard registers wider than word if they
33392 have HARD_REGNO_NREGS of 1 for their mode.
33394 PR middle-end/80163
33395 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
33396 conversions to integer types wider than word and pointer.
33399 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
33400 (rtx_equal_for_cselib_p): Pass 0 to it.
33401 * cselib.c (cselib_hasher::equal): Likewise.
33402 (rtx_equal_for_cselib_1): Add depth argument. If depth
33403 is 128, don't look up VALUE locs and punt. Increment
33404 depth in recursive calls when walking VALUE locs.
33406 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
33408 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
33409 (make_gcov_file_name): Use the canonical path name for generating
33411 (read_line): Fix handling of files with ascii null bytes.
33413 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
33415 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
33416 to initialise a vector register instead
33417 of using a const_int.
33419 2017-03-30 Jakub Jelinek <jakub@redhat.com>
33421 PR translation/80189
33422 * gimplify.c (omp_default_clause): Use %qs instead of %s in
33423 diagnostic messages.
33425 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
33428 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
33429 (dfp_diex_<mode>): Update mode of operand 1.
33430 * doc/extend.texi (dxex, dxexq): Document change to return type.
33431 (diex, diexq): Document change to argument type.
33433 2017-03-30 Martin Jambor <mjambor@suse.cz>
33436 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
33437 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
33438 it reflects the signature changes performed at the callee side.
33439 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
33440 to cgraph_build_function_type_skip_args.
33441 (build_function_decl_skip_args): Adjust call to the above function.
33443 2017-03-30 Jakub Jelinek <jakub@redhat.com>
33446 * config/i386/sse.md
33447 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
33448 register as dest whenever it is a MEM not rtx_equal_p to the
33449 corresponding dup operand, and when forcing into reg move the
33450 reg into the memory afterwards.
33451 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
33452 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
33453 for the force_reg mode.
33454 (avx512vl_vextractf128<mode>): Use register as dest either
33455 always when a MEM, or when it is a MEM not rtx_equal_p to the
33456 corresponding dup operand, or even not when it is a CONST_VECTOR
33457 depending on the mode and lo vs. hi.
33458 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
33460 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
33461 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
33462 Likewise. Require that operands[2] is even.
33463 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
33464 Remove extraneous parens. Require that operands[2] is a multiple
33466 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
33467 operands[0] is a MEM if <mask_applied>, the predicates/constraints
33468 disallow memory then.
33470 2017-03-30 Richard Biener <rguenther@suse.de>
33472 PR tree-optimization/77498
33473 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
33474 to non-constants over backedges.
33476 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
33478 PR rtl-optimization/80233
33479 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
33480 as last_combined_insn. Do not test for BARRIER_P separately.
33482 2017-03-29 Andreas Schwab <schwab@suse.de>
33485 * calls.c (prepare_call_address): Convert funexp to Pmode before
33486 copying to temp reg.
33488 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33490 PR tree-optimization/80158
33491 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
33492 Handle possible future case of more than one alternate
33494 (replace_rhs_if_not_dup): Likewise.
33495 (replace_one_candidate): Likewise.
33497 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
33499 PR rtl-optimization/80193
33500 * ira.c (ira): Do not check allocation for LRA.
33502 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
33504 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
33505 (nvptx_output_simt_exit): Declare.
33506 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
33507 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
33508 (init_softstack_frame): Move initialization of crtl->is_leaf to...
33509 (nvptx_declare_function_name): ...here. Emit declaration of local
33510 memory space buffer for omp_simt_enter insn.
33511 (nvptx_output_unisimt_switch): New.
33512 (nvptx_output_softstack_switch): New.
33513 (nvptx_output_simt_enter): New.
33514 (nvptx_output_simt_exit): New.
33515 * config/nvptx/nvptx.h (struct machine_function): New fields
33516 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
33517 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
33518 (UNSPECV_SIMT_EXIT): Ditto.
33519 (omp_simt_enter_insn): New insn.
33520 (omp_simt_enter): New expansion.
33521 (omp_simt_exit): New insn.
33522 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
33524 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
33525 (expand_GOMP_SIMT_ENTER_ALLOC): New.
33526 (expand_GOMP_SIMT_EXIT): New.
33527 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
33528 (GOMP_SIMT_ENTER_ALLOC): Ditto.
33529 (GOMP_SIMT_EXIT): Ditto.
33530 * target-insns.def (omp_simt_enter): New insn.
33531 (omp_simt_exit): Ditto.
33532 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
33534 (lower_rec_simd_input_clauses): Implement SIMT privatization.
33535 (lower_rec_input_clauses): Likewise.
33536 (lower_lastprivate_clauses): Handle SIMT privatization.
33538 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
33539 (ompdevlow_adjust_simt_enter): New.
33540 (find_simtpriv_var_op): New.
33541 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
33542 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
33544 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
33545 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
33546 (copy_decl_for_dup_finish): Ditto.
33548 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
33550 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
33553 * config/i386/i386.c (ix86_option_override_internal): Always
33554 allow -mpreferred-stack-boundary=3 for 64-bit targets.
33556 2017-03-28 Bin Cheng <bin.cheng@arm.com>
33558 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
33560 2017-03-28 Bin Cheng <bin.cheng@arm.com>
33562 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
33563 mark new edge's irreducible flag accordign to it.
33564 (vect_do_peeling): Check loop preheader edge's irreducible flag
33565 and pass it to function slpeel_add_loop_guard.
33567 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
33569 PR tree-optimization/80218
33570 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
33571 Update block frequencies and counts.
33573 2017-03-28 Richard Biener <rguenther@suse.de>
33575 PR tree-optimization/78644
33576 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
33577 of a simplification result we may not use it at all.
33579 2017-03-28 Richard Biener <rguenther@suse.de>
33582 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
33583 without arguments, generate default definition of a SSA name.
33585 2017-03-28 Richard Biener <rguenther@suse.de>
33587 PR middle-end/80222
33588 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
33589 TYPE_REF_CAN_ALIAS_ALL references.
33590 * fold-const.c (fold_indirect_ref_1): Likewise.
33592 2017-03-28 Martin Liska <mliska@suse.cz>
33595 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
33596 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
33598 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
33599 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
33601 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
33602 (EXTRA_SPECS): Define.
33603 (SUBTARGET_EXTRA_SPECS): Likewise.
33604 (SUBTARGET_CPP_SPEC): Likewise.
33605 * config/arc/elf.h (EXTRA_SPECS): Renamed to
33606 SUBTARGET_EXTRA_SPECS.
33607 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
33609 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
33611 * config/arc/simdext.md (vst64_insn): Update pattern.
33612 (vld32wh_insn): Likewise.
33613 (vld32wl_insn): Likewise.
33614 (vld64_insn): Likewise.
33615 (vld32_insn): Likewise.
33617 2017-03-28 Marek Polacek <polacek@redhat.com>
33620 * fold-const.c (fold_comparison): Use protected_set_expr_location
33621 instead of SET_EXPR_LOCATION.
33623 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
33625 * tree.c (add_expr): Avoid name lookup warning.
33627 2017-03-27 Jeff Law <law@redhat.com>
33629 PR tree-optimization/80216
33630 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
33631 function name. Limit recursion depth.
33632 (record_temporary_equivalences): Corresponding changes.
33634 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
33636 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
33639 2017-03-27 Jakub Jelinek <jakub@redhat.com>
33642 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
33644 * cfgcleanup.c (reg_note_cfa_p): New array.
33645 (insns_have_identical_cfa_notes): New function.
33646 (old_insns_match_p): Don't cross-jump in between /f
33647 and non-/f instructions. If both i1 and i2 are frame related,
33648 verify all CFA notes, their order and content.
33650 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
33653 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
33654 HImode and SImode with zero extend to DImode to one insn.
33655 (bswap<mode>2_extenddi): Likewise.
33656 (bswapsi2_extenddi): Likewise.
33657 (bswaphi2_extendsi): Likewise.
33658 (bswaphi2): Combine bswap HImode and SImode into one insn.
33659 Separate memory insns from swapping register.
33660 (bswapsi2): Likewise.
33661 (bswap<mode>2): Likewise.
33662 (bswaphi2_internal): Delete, no longer used.
33663 (bswapsi2_internal): Likewise.
33664 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
33665 store, and gpr<-gpr swap insns.
33666 (bswap<mode>2_store): Likewise.
33667 (bswaphi2_reg): Register only splitter, combine with the splitter.
33668 (bswaphi2 splitter): Likewise.
33669 (bswapsi2_reg): Likewise.
33670 (bswapsi2 splitter): Likewise.
33671 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
33672 the insns into load, store, and register/register insns.
33673 (bswapdi2_ldbrx): Likewise.
33674 (bswapdi2_load): Likewise.
33675 (bswapdi2_store): Likewise.
33676 (bswapdi2_reg): Likewise.
33678 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
33680 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
33681 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
33683 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
33686 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
33688 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
33689 special handling for target option conflicts between dform
33690 options (-mpower9-dform, -mpower9-dform-vector,
33691 -mpower9-dform-scalar) and -mno-direct-move.
33693 2017-03-27 Richard Biener <rguenther@suse.de>
33695 PR tree-optimization/80181
33696 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
33698 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
33700 * config/arc/predicates.md (move_double_src_operand): Replace the
33701 call to move_double_src_operand with a call to address_operand.
33703 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
33705 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
33706 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
33707 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
33709 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
33711 * config/arc/predicates.md (long_immediate_loadstore_operand):
33712 Consider scaled addresses cases.
33714 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
33716 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
33717 restored when in interrupt.
33718 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
33719 doesn't have delay slot.
33721 2017-03-27 Richard Biener <rguenther@suse.de>
33724 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
33725 inlined thunk clones.
33727 2017-03-27 Jakub Jelinek <jakub@redhat.com>
33730 * asan.c (instrument_derefs): Copy over last operand from
33731 original COMPONENT_REF to the new COMPONENT_REF with
33732 DECL_BIT_FIELD_REPRESENTATIVE.
33733 * ubsan.c (instrument_object_size): Likewise.
33735 2017-03-27 Richard Biener <rguenther@suse.de>
33737 PR tree-optimization/80170
33738 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
33739 sure DR/SCEV didnt fold in constants we do not see when looking
33740 at the reference base alignment.
33742 2017-03-27 Richard Biener <rguenther@suse.de>
33744 PR middle-end/80171
33745 * gimple-fold.c (fold_ctor_reference): Properly guard against
33746 NULL return value from canonicalize_constructor_val.
33748 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
33751 * config/i386/i386.c (ix86_expand_builtin)
33752 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
33753 flags reg setting and flags reg using instructions.
33754 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
33755 clobbering instructions to zero extend op2.
33757 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
33759 * doc/install.texi (Configuration) <--with-aix-soname>:
33760 Update link to AIX ld.
33762 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
33764 PR rtl-optimization/80160
33765 PR rtl-optimization/80159
33766 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
33767 reg_alternate_class into account.
33769 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
33772 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
33773 to consider in curr_insn_transform.
33775 2017-03-24 Jakub Jelinek <jakub@redhat.com>
33777 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
33778 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
33779 and emit_mode_inner.
33781 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33783 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
33784 argument to the overloaded builtin variants. Use the new flag to
33785 deprecate certain builtin variants.
33786 * config/s390/s390-builtin-types.def: Add new builtin types.
33787 * config/s390/s390-builtins.h: Support new flags field for
33788 overloaded builtins.
33789 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
33790 (s390_macro_to_expand): Enable vector float data type.
33791 (s390_cpu_cpp_builtins_internal): Indicate support of the new
33792 builtins by incrementing the __VEC__ version number.
33793 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
33795 (s390_resolve_overloaded_builtin): Emit error messages depending
33796 on the builtin flags.
33797 * config/s390/s390.c (s390_expand_builtin): Support additional
33798 flags argument. Change error message to match the messages
33799 emitted in s390-c.c.
33800 * config/s390/s390.md: New UNSPEC_* constants.
33801 (op_type): Add new instruction types.
33802 * config/s390/vecintrin.h: Add new builtins and test data class
33804 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
33805 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
33806 (VEC_INEXACT, VEC_NOINEXACT): New constants.
33807 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
33808 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
33809 ("vec_mergel<mode>"): V_HW -> VEC_HW.
33811 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
33812 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
33813 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
33814 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
33816 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
33817 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
33818 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
33819 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
33821 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
33822 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
33823 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
33824 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
33825 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
33826 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
33827 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
33829 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
33830 ("vec_scatter_element<V_HW_4:mode>_DI")
33831 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
33832 ("vec_fpint<mode>", "vflls")
33833 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
33834 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
33835 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
33836 ("*vec_cmphe<mode>_cc"): ... these.
33838 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
33839 mode constant instead of magic value.
33841 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33843 * config/s390/s390.c (s390_expand_vec_compare): Support other
33844 vector floating point modes than just V2DF.
33845 (s390_expand_vcond): Likewise.
33846 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
33847 (s390_cannot_change_mode_class): Prevent mode changes between TF
33848 and V1TF in vector registers.
33849 * config/s390/s390.md (DF, SF): New mode attributes.
33850 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
33851 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
33852 SFmode support for VRs.
33853 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
33855 (VFT, VF_HW): New mode iterators.
33856 (vw, sdx): New mode attributes.
33857 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
33858 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
33859 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
33860 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
33861 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
33862 also the new vector floating point modes. Renaming to ...
33864 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
33865 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
33866 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
33867 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
33868 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
33869 ("vec_unordered<mode>"): ... these.
33871 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
33872 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
33873 ("*vec_extendv2df"): New insn definitions.
33875 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33877 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
33878 ("mulditi3_2", "*muldi3_sign"): New patterns.
33879 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
33880 rename the pattern definition.
33882 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33884 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
33886 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
33888 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33890 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
33891 instruction if possible.
33892 * config/s390/vector.md (vec_halfnumelts): New mode
33894 ("*vec_vllezlf<mode>"): New pattern.
33896 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33898 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
33899 ("popcountv4si2", "popcountv2di2"): Rename to ...
33900 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
33901 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
33903 ("popcount<mode>2_vxe"): New pattern.
33905 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33907 * common/config/s390/s390-common.c (processor_flags_table): Add
33909 * config.gcc: Add arch12.
33910 * config/s390/driver-native.c (s390_host_detect_local_cpu):
33911 Default to arch12 for unknown CPU model numbers.
33912 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
33913 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
33914 PROCESSOR_max sanity check.
33915 * config/s390/s390-opts.h (enum processor_type): Add
33917 * config/s390/s390.c (processor_table): Add arch12.
33918 (s390_expand_builtin): Add check for B_VXE flag.
33919 (s390_issue_rate): Add PROCESSOR_ARCH12.
33920 (s390_get_sched_attrmask): Likewise.
33921 (s390_get_unit_mask): Likewise.
33922 (s390_sched_score): Enable z13 scheduling for arch12.
33923 (s390_sched_reorder): Likewise.
33924 (s390_sched_variable_issue): Likewise.
33925 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
33927 (s390_tune_attr): Use z13 scheduling also for arch12.
33928 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
33929 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
33930 (TARGET_VXE_P): New macros.
33931 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
33932 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
33933 * config/s390/s390.opt: Add arch12 as processor_type.
33935 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33937 * config/s390/s390.md
33938 ("fixuns_truncdddi2", "fixuns_trunctddi2")
33939 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
33940 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
33942 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
33943 Rename expanders to ...
33945 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
33946 ("fixuns_truncdddi2_emu"): ... these.
33948 ("fixuns_trunc<mode>si2_emu"): New expander.
33950 ("*fixuns_truncdfdi2_z13"): Rename to ...
33951 ("*fixuns_truncdfdi2_vx"): ... this.
33953 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33955 * config/s390/2964.md: Remove the single element vector compare
33956 instructions which are no longer used.
33957 * config/s390/s390.c (s390_select_ccmode): Remove handling of
33959 (s390_canonicalize_comparison): Remove handling of DFmode
33961 (s390_expand_vec_compare_scalar): Remove function.
33962 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
33963 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
33965 ("*cmp<mode>_ccs"): Add wfcdb instruction.
33967 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33969 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
33971 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
33972 will anyway by matched by mov<mode>_64dfp.
33974 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33976 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
33977 vlef/vstef. Add missing operand to vleif.
33979 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33981 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
33982 pair for all vector types with 64 bit elements.
33983 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
33984 * config/s390/vector.md (V_HW_64): ... here.
33985 (V_128_NOSINGLE): New mode iterator.
33986 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
33987 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
33988 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
33989 ("*vec_load_pairv2di"): Change to ...
33990 ("*vec_load_pair<mode>"): ... this one.
33992 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33994 * config/s390/constraints.md: Add comments.
33995 (jKK): Reject element sizes > 8 bytes.
33996 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
33998 * config/s390/s390.md: Add the s_operand checks formerly in
33999 s390_split_ok_p to various splitters where they are still
34001 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
34002 for 128 bit vectors. Plus two splitters.
34004 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34006 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
34009 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34012 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
34013 error if the boundary argument is not constant.
34015 2017-03-24 Jakub Jelinek <jakub@redhat.com>
34017 PR rtl-optimization/80112
34018 * loop-doloop.c (doloop_condition_get): Don't check condition
34019 if cmp isn't SET with IF_THEN_ELSE src.
34021 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34023 PR tree-optimization/80158
34024 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
34025 replacing a candidate statement, also replace it for the
34026 candidate's alternate interpretation.
34027 (replace_rhs_if_not_dup): Likewise.
34028 (replace_one_candidate): Likewise.
34030 2017-03-24 Richard Biener <rguenther@suse.de>
34032 PR tree-optimization/80167
34033 * graphite-isl-ast-to-gimple.c
34034 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
34036 (translate_isl_ast_to_gimple::get_rename): Likewise.
34038 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
34040 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
34041 handling of certain combinations of target options, including the
34042 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
34043 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
34045 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34048 * config/arm/arm.md (*load_multiple): Add reload_completed to
34049 matching condition.
34051 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34052 Richard Biener <rguenth@suse.de>
34054 PR tree-optimization/79908
34055 PR tree-optimization/80136
34056 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
34057 been cast away, gimplify_and_add suffices.
34059 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
34061 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
34063 2017-03-23 Richard Biener <rguenther@suse.de>
34065 PR tree-optimization/80032
34066 * gimplify.c (gimple_push_cleanup): Forced unconditional
34067 cleanups still have to go to the conditional_cleanups
34070 2017-03-22 Jakub Jelinek <jakub@redhat.com>
34072 PR tree-optimization/80072
34073 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
34075 (next_operand_entry_id): Change type to unsigned int.
34076 (sort_by_operand_rank): Make sure to return the right return value
34077 even if unsigned fields are bigger than INT_MAX.
34078 (struct oecount): Change cnt and id type to unsigned int.
34079 (oecount_hasher::equal): Formatting fix.
34080 (oecount_cmp): Make sure to return the right return value
34081 even if unsigned fields are bigger than INT_MAX.
34082 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
34085 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
34086 TREE_READONLY on result if writing it more than once.
34089 * doc/invoke.texi (-fsanitize=thread): Document that with
34090 -fnon-call-exceptions atomics are not able to throw
34094 * tsan.c: Include tree-eh.h.
34095 (instrument_builtin_call): Call maybe_clean_eh_stmt or
34096 maybe_clean_or_replace_eh_stmt where needed.
34097 (instrument_memory_accesses): Add cfg_changed argument.
34098 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
34099 if it returned true.
34100 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
34102 PR rtl-optimization/63191
34103 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
34104 wrapper function, moved the whole old content into ...
34105 (ix86_delegitimize_address_1): ... this. New inline function.
34106 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
34107 true as last argument instead of ix86_delegitimize_address.
34109 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
34111 * config/aarch64/aarch64.c (generic_branch_cost): Copy
34112 cortexa57_branch_cost.
34114 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
34116 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
34118 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34121 * doc/md.texi (Constraints): Document wA constraint.
34122 * config/rs6000/constraints.md (wA): New.
34123 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
34124 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
34125 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
34126 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
34128 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
34131 * gimplify.c (is_oacc_declared): New function.
34132 (oacc_default_clause): Use it to set default flags for acc declared
34133 variables inside parallel regions.
34134 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
34135 declared variables.
34136 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
34137 declare attribute to any decl as necessary.
34139 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
34142 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
34143 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
34144 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
34145 (arm_arch_lpae): This.
34146 * config/arm/arm.c (arm_arch7ve): Rename into ...
34147 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
34148 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
34151 2017-03-22 Martin Liska <mliska@suse.cz>
34154 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
34155 error message instead of an ICE.
34157 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34159 * doc/extend.texi (6.11 Additional Floating Types): Revise.
34161 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
34163 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
34165 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
34168 2017-03-21 Martin Sebor <msebor@redhat.com>
34170 * doc/extend.texi: Use "cannot" instead of "can't."
34171 * doc/hostconfig.texi: Same.
34172 * doc/install.texi: Same.
34173 * doc/invoke.texi: Same.
34174 * doc/loop.texi: Same.
34175 * doc/md.texi: Same.
34176 * doc/objc.texi: Same.
34177 * doc/rtl.texi: Same.
34178 * doc/tm.texi: Same.
34179 * doc/tm.texi.in: Same.
34180 * doc/trouble.texi: Same.
34182 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
34185 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
34186 (collect_checksum_attributes): Set it.
34187 (die_checksum_ordered): Use it.
34189 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34191 PR tree-optimization/79908
34192 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
34193 change: For a VA_ARG whose LHS has been cast away, use
34194 force_gimple_operand to construct the side effects.
34196 2017-03-21 David Malcolm <dmalcolm@redhat.com>
34198 PR translation/80001
34199 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
34200 more amenable to translation.
34201 (oacc_loop_auto_partitions): Likewise.
34203 2017-03-21 Marek Polacek <polacek@redhat.com>
34204 Martin Sebor <msebor@redhat.com>
34206 PR tree-optimization/80109
34207 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
34208 on INTEGRAL_TYPE_P.
34210 2017-03-21 Jakub Jelinek <jakub@redhat.com>
34211 Segher Boessenkool <segher@kernel.crashing.org>
34214 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
34215 check reg_used_between_p between insn and one of succ or succ2
34216 depending on if succ is artificial insn not inserted into insn
34219 2017-03-21 Martin Liska <mliska@suse.cz>
34221 PR gcov-profile/80081
34222 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
34223 * doc/gcc.texi: Include gcov-dump stuff.
34224 * doc/gcov-dump.texi: New file.
34226 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
34228 PR rtl-optimization/79150
34229 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
34230 conditional jump, if the jump is the last insn of the loop.
34232 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34233 Richard Biener <rguenth@suse.de>
34235 PR tree-optimization/79908
34236 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
34237 been cast away, use force_gimple_operand to construct the side
34240 2017-03-21 Martin Liska <mliska@suse.cz>
34242 PR libfortran/79956
34243 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
34246 2017-03-21 Brad Spengler <spender@grsecurity.net>
34249 * plugin.c (htab_hash_plugin): New function.
34250 (add_new_plugin): Use it and adjust.
34251 (parse_plugin_arg_opt): Adjust.
34252 (init_one_plugin): Likewise.
34254 2017-03-21 Richard Biener <rguenther@suse.de>
34256 PR tree-optimization/80032
34257 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
34258 if set force the cleanup to happen unconditionally.
34259 (gimplify_target_expr): Push inserted clobbers with force_uncond
34260 to avoid them being removed by control-dependent DCE.
34262 2017-03-21 Richard Biener <rguenther@suse.de>
34264 PR tree-optimization/80122
34265 * tree-inline.c (copy_bb): Do not expans va-arg packs or
34266 va_arg_pack_len when the inlined call stmt requires pack
34268 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
34270 2017-03-21 Jakub Jelinek <jakub@redhat.com>
34273 * tsan.c (instrument_builtin_call): If the memory model argument
34274 is not a constant, assume it is valid.
34277 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
34280 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
34282 PR rtl-optimization/79910
34283 * combine.c (can_combine_p): Do not allow combining an I0 or I1
34284 if its dest is used by an insn before I2 (other than the combined
34285 insns themselves, which are properly handled already).
34287 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
34290 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
34292 * combine.c (record_used_regs): New static function.
34293 (try_combine): Handle situations where there is an additional
34294 instruction between I2 and I3 which needs to have a LOG_LINK
34298 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
34300 * combine.c (try_combine): Delete redundant i1 test. Call
34301 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
34303 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34306 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
34307 alternatives 13/14.
34309 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34311 PR tree-optimization/80054
34312 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
34313 the optimization if a PHI or any of its arguments is not dominated
34314 by the candidate's basis. Use gphi* rather than gimple* as
34316 (replace_profitable_candidates): Clean up a gimple* variable that
34317 should be a gphi* variable.
34319 2017-03-20 Martin Sebor <msebor@redhat.com>
34322 * doc/extend.texi (attribute constructor): Document present limitation.
34324 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
34327 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
34328 __POWER9_VECTOR__ #ifdef control, change template definition to
34329 use Power9-specific built-in function.
34330 (vec_any_eq): Likewise.
34331 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
34332 to control outcomes from this test.
34333 (vector_ae_<mode>p): For VEC_F modes, likewise.
34335 2017-03-20 Ian Lance Taylor <iant@google.com>
34337 * config/i386/i386.c (ix86_function_regparm): Save an extra
34338 register for -fsplit-stack with DECL_STATIC_CHAIN.
34340 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
34343 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
34344 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
34346 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
34348 * config/riscv/riscv.c (riscv_print_operand): Use "fence
34350 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
34353 2017-03-20 Marek Polacek <polacek@redhat.com>
34356 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
34358 2017-03-20 Richard Biener <rguenther@suse.de>
34360 PR tree-optimization/80113
34361 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
34362 allocate extra SSA name for PHI def.
34363 (add_close_phis_to_outer_loops): Likewise.
34364 (add_close_phis_to_merge_points): Likewise.
34365 (copy_loop_close_phi_args): Likewise.
34366 (copy_cond_phi_nodes): Likewise.
34368 2017-03-20 Martin Liska <mliska@suse.cz>
34370 PR middle-end/79753
34371 * tree-chkp.c (chkp_build_returned_bound): Do not build
34372 returned bounds for a LHS that's not a BOUNDED_P type.
34374 2017-03-20 Martin Liska <mliska@suse.cz>
34378 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
34379 COMPLEX_CST and VECTOR_CST.
34381 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34384 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
34385 target operand. A new splitter adds the clobber statement in case
34386 the target operand is dead anyway.
34388 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
34390 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
34391 to age-old versions of binutils and glibc.
34393 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
34395 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
34397 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
34399 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
34401 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
34403 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
34404 requirement for binutils 2.13.
34406 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
34408 * combine.c (try_combine): Delete redundant i1 test. Call
34409 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
34411 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
34413 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
34414 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
34416 <riscv64-*-elf>: Re-arrange section
34417 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
34418 <riscv32-*-linux>: Likewise.
34419 <riscv64-*-elf>: Likewise
34420 <riscv64-*-linux>: Likewise.
34422 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
34425 * aarch64.opt(verbose-cost-dump): Fix typo.
34427 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
34430 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
34431 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
34433 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
34435 * reload.c (find_reloads): When reloading a nonoffsettable address,
34436 use RELOAD_OTHER for it and its address reloads.
34438 PR rtl-optimization/79910
34439 * combine.c (record_used_regs): New static function.
34440 (try_combine): Handle situations where there is an additional
34441 instruction between I2 and I3 which needs to have a LOG_LINK
34444 2017-03-17 Jeff Law <law@redhat.com>
34446 PR tree-optimization/71437
34447 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
34448 conditional in the hash table first.
34449 (vrp_dom_walker::before_dom_children): Extract condition from
34450 ASSERT_EXPR. Record condition, its inverion and any implied
34451 conditions as well.
34453 2017-03-17 Marek Polacek <polacek@redhat.com>
34454 Markus Trippelsdorf <markus@trippelsdorf.de>
34456 PR tree-optimization/80079
34457 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
34460 2017-03-17 Richard Biener <rguenther@suse.de>
34462 PR middle-end/80075
34463 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
34464 Properly verify the LHS before the RHS possibly claims to be
34466 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
34469 2017-03-17 Martin Jambor <mjambor@suse.cz>
34471 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
34472 (List of -O2 options): Likewise.
34473 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
34476 2017-03-17 Tom de Vries <tom@codesourcery.com>
34478 * gcov-dump.c (print_usage): Print bug_report_url.
34480 2017-03-17 Richard Biener <rguenther@suse.de>
34482 PR middle-end/80050
34483 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
34484 (parser::peek): Likewise.
34486 2017-03-17 Richard Biener <rguenther@suse.de>
34488 PR tree-optimization/80048
34489 * sese.c (free_sese_info): Properly release rename_map and
34490 copied_bb_map elements.
34492 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
34494 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
34495 Add linked-list forward and backlinks. Insert on
34496 construction, remove on destruction.
34497 (class pass_store_merging): Add m_stores_head field.
34498 (pass_store_merging::terminate_and_process_all_chains):
34499 Iterate over m_stores_head list.
34500 (pass_store_merging::terminate_all_aliasing_chains):
34502 (pass_store_merging::execute): Check for debug stmts first.
34503 Push new chains onto the m_stores_head stack.
34505 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
34508 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
34509 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
34510 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
34512 2017-03-16 Jeff Law <law@redhat.com>
34514 PR tree-optimization/71437
34515 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
34516 member function. Implementation moved into after_dom_children
34517 member function and into the threader's thread_outgoing_edges
34519 (dom_opt_dom_walker::after_dom_children): Simplify by moving
34520 some code into new thread_outgoing_edges.
34521 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
34522 definition. Simplify marker handling (do it here). Assume we always
34523 have the available expression and the const/copies tables.
34524 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
34526 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
34527 * tree-vrp.c (equiv_stack): No longer file scoped.
34528 (vrp_dom_walker): New class.
34529 (vrp_dom_walker::before_dom_children): New member function.
34530 (vrp_dom_walker::after_dom_children): Likewise.
34531 (identify_jump_threads): Setup domwalker. Use it rather than
34532 walking edges in a random order by hand. Simplify setup/finalization.
34533 (finalize_jump_threads): Remove.
34534 (vrp_finalize): Do not call identify_jump_threads here.
34535 (execute_vrp): Do it here instead and call thread_through_all_blocks
34538 PR tree-optimization/71437
34539 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
34541 (simplify_stmt_for_jump_threading): Add basic_block argument. All
34543 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
34544 (dom_opt_dom_walker::thread_across_edge): Remove
34545 handle_dominating_asserts argument. All callers changed.
34546 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
34547 changes. Remove calls to lhs_of_dominating_assert. Other
34548 uses of handle_dominating_asserts turn into unconditional code
34549 (simplify_control_stmt_condition_1): Likewise.
34550 (simplify_control_stmt_condition): Likewise.
34551 (thread_through_normal_block, thread_across_edge): Likewise.
34552 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
34553 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
34554 object if it is not an SSA_NAME.
34555 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
34556 before calling into the VRP specific simplifiers.
34557 (identify_jump_threads): Remove handle_dominating_asserts
34560 2017-03-16 Jakub Jelinek <jakub@redhat.com>
34563 * tree-diagnostic.c (default_tree_printer): No longer static.
34564 * tree-diagnostic.h (default_tree_printer): New prototype.
34566 2017-03-16 Tamar Christina <tamar.christina@arm.com>
34568 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
34569 Change ins into fmov.
34571 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34573 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
34574 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
34575 Use h_con constraint for operand 1.
34576 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
34577 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
34579 2017-03-15 Jeff Law <law@redhat.com>
34581 PR tree-optimization/71437
34582 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
34583 (record_temporary_equivalences): Use it.
34585 PR tree-optimization/71437
34586 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
34587 tree-ssa-scopedtables.
34588 (lookup_avail_expr, build_and_record_new_cond): Likewise.
34589 (record_conditions, record_cond, vuse_eq): Likewise.
34590 (record_edge_info): Adjust to API tweak of record_conditions.
34591 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
34592 (record_temporary_equivalences, optimize_stmt): Likewise.
34593 (eliminate_redundant_computations): Likewise.
34594 (record_equivalences_from_stmt): Likewise.
34595 * tree-ssa-scopedtables.c: Include options.h and params.h.
34596 (vuse_eq): New function, moved from tree-ssa-dom.c
34597 (build_and_record_new_cond): Likewise.
34598 (record_conditions): Likewise. Accept vector of conditions rather
34599 than edge_equivalence structure for first argument.
34600 for the first argument.
34601 (avail_exprs_stack::lookup_avail_expr): New member function, moved
34602 from tree-ssa-dom.c.
34603 (avail_exprs_stack::record_cond): Likewise.
34604 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
34605 from tree-ssa-dom.c.
34606 (avail_exprs_stack): Add new member functions lookup_avail_expr
34608 (record_conditions): Declare.
34610 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
34613 * lra-constraints.c (process_alt_operands): Increase reject for
34614 reloading an input/output operand.
34616 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
34619 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
34620 insns to convert from signed/unsigned char/short to IEEE 128-bit
34622 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
34624 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
34627 * config/i386/i386.c (ix86_vector_duplicate_value): Create
34628 subreg of inner mode for values already in registers.
34630 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
34632 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
34633 iteration reg is used after the loop.
34635 2017-03-14 Martin Sebor <msebor@redhat.com>
34637 PR tree-optimization/79800
34638 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
34639 precision in negative-positive range.
34640 (format_floating): Call non-const overload with adjusted precision.
34642 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
34645 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
34648 2017-03-14 Martin Sebor <msebor@redhat.com>
34650 PR middle-end/80020
34651 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
34652 * builtins.def (aligned_alloc): Use it.
34655 * Makefile.in (GTFILES): Add calls.c.
34656 * calls.c: Include "gt-calls.h".
34658 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
34660 PR rtl-optimization/79728
34661 * regs.h (struct target_regs): New field
34662 x_contains_allocatable_regs_of_mode.
34663 (contains_allocatable_regs_of_mode): New macro.
34664 * reginfo.c (init_reg_sets_1): Initialize it, and change
34665 contains_reg_of_mode so it includes global regs as well.
34666 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
34667 rather than contains_regs_of_mode.
34669 2017-03-14 Martin Liska <mliska@suse.cz>
34671 * doc/invoke.texi: Document options that can't be combined with
34672 -fcheck-pointer-bounds.
34674 2017-03-14 Martin Liska <mliska@suse.cz>
34676 PR middle-end/79831
34677 * doc/invoke.texi (-Wchkp): Document the option.
34679 2017-03-14 Martin Liska <mliska@suse.cz>
34681 * Makefile.in: Install gcov-dump.
34683 2017-03-14 Martin Liska <mliska@suse.cz>
34685 * multiple_target.c (expand_target_clones): Bail out for
34686 an invalid attribute.
34688 2017-03-14 Richard Biener <rguenther@suse.de>
34690 * alias.c (struct alias_set_entry): Pack properly.
34691 * cfgloop.h (struct loop): Likewise.
34692 * cse.c (struct set): Likewise.
34693 * ipa-utils.c (struct searchc_env): Likewise.
34694 * loop-invariant.c (struct invariant): Likewise.
34695 * lra-remat.c (struct cand): Likewise.
34696 * recog.c (struct change_t): Likewise.
34697 * rtl.h (struct address_info): Likewise.
34698 * symbol-summary.h (function_summary): Likewise.
34699 * tree-loop-distribution.c (struct partition): Likewise.
34700 * tree-object-size.c (struct object_size_info): Likewise.
34701 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
34702 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
34703 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
34704 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
34705 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
34706 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
34707 (struct _stmt_vec_info): Likewise.
34709 2017-03-14 Martin Liska <mliska@suse.cz>
34712 * multiple_target.c (create_dispatcher_calls): Check that
34713 a target can create a function dispatcher.
34715 2017-03-14 Martin Liska <mliska@suse.cz>
34718 * multiple_target.c (expand_target_clones): Drop local.local
34719 flag for default implementation.
34721 2017-03-14 Richard Biener <rguenther@suse.de>
34723 PR tree-optimization/80030
34724 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
34726 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
34728 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
34729 gcc_fallthrough() instead of __attribute__((fallthrough));
34731 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
34733 * doc/gcc.texi: Remove "up" link to (DIR).
34734 * doc/gccint.texi: Ditto.
34736 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
34738 * doc/install.texi (Specific) <avr>: Remove reference to
34741 2017-03-13 Jeff Law <law@redhat.com>
34743 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
34744 attribute rather than comments.
34746 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
34747 match_scratch operand is highest.
34749 2017-03-13 Martin Liska <mliska@suse.cz>
34751 PR middle-end/78339
34752 * ipa-pure-const.c (warn_function_noreturn): If the declarations
34753 is a CHKP clone, use original declaration.
34755 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
34757 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
34758 (arc_conditional_register_usage): Use a different allocation order
34759 when optimizing for size.
34760 * common/config/arc/arc-common.c (arc_option_optimization_table):
34761 Section anchors default on when optimizing for size.
34763 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
34765 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
34767 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
34769 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
34770 * config/arc/arc.md (cpu_facility): Add cd variant.
34771 (*movqi_insn): Add code density variant.
34772 (*movhi_insn): Likewise.
34773 (*movqi_insn): Likewise.
34774 (*addsi3_mixed): Likewise.
34775 (subsi3_insn): Likewise.
34777 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
34779 * config/arc/arc.md (movsi_cond_exec): Update constraint.
34781 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
34783 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
34784 expressions with MINUS and UNARY ops.
34786 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34789 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
34791 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
34792 between vec_select and vector argument.
34793 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
34794 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
34795 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
34796 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
34797 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
34798 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
34800 2017-03-13 Richard Biener <rguenther@suse.de>
34803 * params.def (vect-max-peeling-for-alignment): Fix typo.
34805 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
34807 * doc/install.texi (Specific) <mips-*-*>: Remove description of
34808 issue that only occurred with binutils below 2.18.
34810 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
34812 * doc/install.texi (Specific) <cris-axis-elf>: No longer
34813 refer to binutils 2.11/2.12 minimum.
34815 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
34817 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
34818 ftp.kernel.org and simplify binutils requirement.
34820 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
34822 * doc/invoke.texi (Warning Options): Fix spelling of link-time
34824 (Optimize Options): Ditto. Also remove redundancy.
34826 2017-03-10 David Malcolm <dmalcolm@redhat.com>
34828 PR translation/79848
34829 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
34831 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
34832 to G_ to avoid double translation.
34834 2017-03-10 David Malcolm <dmalcolm@redhat.com>
34836 PR translation/79923
34837 * auto-profile.c (get_combined_location): Convert leading
34838 character of diagnostics to lower case and remove trailing period.
34839 (read_profile): Likewise for various diagnostics.
34840 * config/arm/arm.c (arm_option_override): Remove trailing period
34841 from various diagnostics.
34842 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
34843 (msp430_expand_delay_cycles): Likewise.
34845 2017-03-10 David Malcolm <dmalcolm@redhat.com>
34848 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
34849 full command-line argument, rather than just "str".
34850 (aarch64_validate_march): Likewise.
34851 (aarch64_validate_mtune): Likewise.
34853 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
34855 PR rtl-optimization/78911
34856 * lra-assigns.c (must_not_spill_p): New function.
34857 (spill_for): Use it.
34859 2017-03-10 Jakub Jelinek <jakub@redhat.com>
34861 PR tree-optimization/79981
34862 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
34863 ATOMIC_COMPARE_EXCHANGE ifn result.
34864 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
34865 IFN_ATOMIC_COMPARE_EXCHANGE.
34867 2017-03-10 David Malcolm <dmalcolm@redhat.com>
34870 * opts.c (parse_sanitizer_options): Add missing question mark to
34871 "did you mean" message.
34873 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34875 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
34877 (VMULEUH_UNS): Likewise.
34878 (VMULOUB_UNS): Likewise.
34879 (VMULOUH_UNS): Likewise.
34880 * config/rs6000/rs6000.c (builtin_function_type): Remove
34881 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
34883 2017-03-10 David Malcolm <dmalcolm@redhat.com>
34886 * read-rtl-function.c (function_reader::read_rtx_operand): Update
34887 x with result of extra_parsing_for_operand_code_0.
34888 (function_reader::extra_parsing_for_operand_code_0): Convert
34889 return type from void to rtx, returning x. When reading
34890 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
34891 larger size containing struct block_symbol.
34893 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
34895 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
34896 -mfloat128-hardware without -m64.
34898 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
34901 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
34902 entries to the case statement that marks unsigned arguments to
34903 overloaded functions.
34905 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
34907 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
34908 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
34910 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
34913 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
34914 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
34916 2017-03-10 Martin Liska <mliska@suse.cz>
34920 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
34921 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
34924 2017-03-10 Olivier Hainque <hainque@adacore.com>
34926 * tree-switch-conversion (array_value_type): Start by resetting
34927 candidate type to it's main variant.
34929 2017-03-10 Jakub Jelinek <jakub@redhat.com>
34931 PR rtl-optimization/79909
34932 * combine.c (try_combine): Use simplify_replace_rtx on individual
34933 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
34934 of the whole CALL_INSN_FUNCTION_USAGE.
34936 PR tree-optimization/79972
34937 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
34938 get_range_info on SSA_NAMEs. Formatting fixes.
34940 2017-03-10 Richard Biener <rguenther@suse.de>
34941 Jakub Jelinek <jakub@redhat.com>
34943 PR tree-optimization/77975
34944 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
34945 edge to be constant.
34946 (get_val_for): For constant x return it. Formatting fix.
34947 (loop_niter_by_eval): Avoid pointless looping if the next iteration
34948 would use the same bases as the current one.
34950 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34952 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
34953 instead of vec_select for V1TImode.
34954 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
34956 (VSX_LE_128): Add V1TI to this mode iterator.
34957 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
34958 (*vsx_le_perm_store_<mode>): Likewise.
34959 (pre-reload splitter for VSX stores): Likewise.
34960 (post-reload splitter for VSX stores): Likewise.
34961 (*vsx_xxpermdi2_le_<mode>): Likewise.
34962 (*vsx_lxvd2x2_le_<mode>): Likewise.
34963 (*vsx_stxvd2x2_le_<mode>): Likewise.
34965 2017-03-09 Michael Eager <eager@eagercon.com>
34967 Correct failures with --enable-checking=yes,rtl.
34969 * config/microblaze/microblaze.c (microblaze_expand_shift):
34970 Replace GET_CODE test with CONST_INT_P and INTVAL test with
34971 test for const0_rtx.
34972 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
34973 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
34975 2017-03-09 Richard Biener <rguenther@suse.de>
34977 PR tree-optimization/79977
34978 * graphite-scop-detection.c (scop_detection::merge_sese):
34979 Handle the case of extra exits to blocks dominating the entry.
34981 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
34983 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
34986 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
34988 PR rtl-optimization/79949
34989 * lra-constraints.c (process_alt_operands): Check memory when
34990 trying to predict a cycle. Print about the overall increase.
34992 2017-03-09 Richard Biener <rguenther@suse.de>
34994 PR middle-end/79971
34995 * gimple-expr.c (useless_type_conversion_p): Preserve
34996 TYPE_SATURATING for fixed-point types.
34998 2017-03-09 Richard Biener <rguenther@suse.de>
35001 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
35002 alignment of BLKmode params.
35004 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35007 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
35008 (VALL_NO_V2Q): Likewise.
35010 * config/aarch64/aarch64-simd.md
35011 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
35013 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
35014 VALL_NO_V2Q mode iterator.
35015 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
35017 2017-03-09 Martin Liska <mliska@suse.cz>
35019 PR tree-optimization/79631
35020 * tree-chkp-opt.c (chkp_is_constant_addr): Call
35021 tree_int_cst_sign_bit just for INTEGER constants.
35023 2017-03-09 Martin Liska <mliska@suse.cz>
35027 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
35030 2017-03-09 Marek Polacek <polacek@redhat.com>
35033 * tree.c (inchash::add_expr): Handle TREE_VEC.
35035 2017-03-09 Martin Liska <mliska@suse.cz>
35038 (chkp_narrow_size_and_offset): New function.
35039 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
35040 (void chkp_parse_bit_field_ref): New function.
35041 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
35042 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
35044 2017-03-09 Martin Liska <mliska@suse.cz>
35047 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
35048 (chkp_find_bounds_1): Remove gcc_unreachable.
35050 2017-03-09 Jakub Jelinek <jakub@redhat.com>
35053 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
35054 BUILT_IN_SYNC*, determine the access type from the size suffix and
35055 always build a MEM_REF with that type. Handle forgotten
35056 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
35059 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
35060 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
35061 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
35062 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
35063 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
35064 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
35065 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
35066 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
35067 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
35068 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
35069 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
35070 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
35071 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
35072 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
35073 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
35074 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
35075 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
35076 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
35077 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
35078 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
35079 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
35080 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
35081 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
35082 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
35083 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
35084 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
35085 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
35086 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
35087 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
35088 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
35089 definitions outside of __OPTIMIZE__ guarded section.
35092 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
35093 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
35094 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
35095 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
35098 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35100 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
35101 ("vfenez<mode>"): Add missing constraints.
35103 2017-03-08 Martin Sebor <msebor@redhat.com>
35106 * config/nds32/nds32.c (nds32_option_override):
35107 Fix misspelled diagnostic.
35109 2017-03-08 Jakub Jelinek <jakub@redhat.com>
35112 * gimplify.c (gimplify_omp_for): Replace index var in outer
35113 taskloop statement with an artificial variable and add
35114 OMP_CLAUSE_PRIVATE clause for it.
35116 2017-03-08 Richard Biener <rguenther@suse.de>
35118 PR tree-optimization/79955
35119 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
35120 for accesses that are completely outside of the variable.
35122 2017-03-08 Andrew Haley <aph@redhat.com>
35124 PR tree-optimization/79943
35125 * tree-ssa-loop-split.c (compute_new_first_bound): When
35126 calculating the new upper bound, (END-BEG) should be added, not
35129 2017-03-08 Jakub Jelinek <jakub@redhat.com>
35131 * config/avr/avr.md (setmemhi): Make sure match_dup
35132 operand number comes before match_scratch.
35134 2017-03-08 Richard Biener <rguenther@suse.de>
35136 PR tree-optimization/79920
35137 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
35138 with ncopies == 1 to ...
35139 (vect_transform_slp_perm_load): ... here. Properly compute
35140 all element loads by iterating VF times over the group. Do
35141 not handle ncopies (computed in a broken way) in
35142 vect_create_mask_and_perm.
35144 2017-03-08 Jakub Jelinek <jakub@redhat.com>
35147 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
35148 is a uniform vector, use uniform_vector_p return value instead of
35149 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
35151 2017-03-07 Marek Polacek <polacek@redhat.com>
35153 PR middle-end/79809
35154 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
35155 (alloca_call_type): Likewise.
35157 2017-03-07 Martin Liska <mliska@suse.cz>
35159 * gcov.c (process_args): Put comment to correct location.
35161 2017-03-07 Martin Liska <mliska@suse.cz>
35163 PR middle-end/68270
35164 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
35165 Use array_at_struct_end_p instead of DECL_CHAIN (field).
35166 (chkp_narrow_bounds_for_field): Likewise.
35167 (chkp_parse_array_and_component_ref): Pass one more argument to
35170 2017-03-07 Richard Biener <rguenther@suse.de>
35172 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
35175 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
35177 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
35178 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
35180 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35183 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
35184 to end of description.
35185 (PARAM_MAX_STORES_TO_MERGE): Likewise.
35187 2017-03-07 Jakub Jelinek <jakub@redhat.com>
35189 PR rtl-optimization/79901
35190 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
35192 (*avx512f_<code><mode>3<mask_name>): ... this.
35193 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
35194 iterator instead of VI8_AVX2_AVX512BW.
35196 PR rtl-optimization/79901
35197 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
35198 min/max expander, expand it using expand_vec_cond_expr.
35201 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
35204 2017-03-06 Jakub Jelinek <jakub@redhat.com>
35207 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
35208 to void * for PCH reasons.
35209 * dwarf2out.c (output_loc_operands, output_die): Cast
35210 v.val_vec.array to unsigned char *.
35212 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
35215 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
35218 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
35220 PR rtl-optimization/79571
35221 * lra-constraints.c (process_alt_operands): Calculate static
35222 reject and subtract it from overall when only addresses will be
35225 2017-03-06 Julia Koval <julia.koval@intel.com>
35228 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
35229 incoming stack boundary to 128 for 64-bit targets.
35231 2017-03-06 Richard Biener <rguenther@suse.de>
35233 PR tree-optimization/79894
35234 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
35235 to NULL after folding it.
35237 2017-03-06 Richard Biener <rguenther@suse.de>
35239 PR tree-optimization/79824
35240 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
35241 check disabling peeling for gaps.
35243 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
35245 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
35246 attributes): Document gettimeofday.
35248 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
35250 * config/s390/s390.c (s390_option_override_internal): Set
35251 PARAM_MIN_VECT_LOOP_BOUND
35253 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
35255 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
35256 * config/s390/s390.md: Likewise.
35258 2017-03-06 Jakub Jelinek <jakub@redhat.com>
35261 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
35262 (<avx2_avx512>_perm<mode>): Rename to ...
35263 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
35265 (<avx512>_perm<mode>_mask): Rename to ...
35266 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
35268 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
35269 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
35270 instead of VI8F_256_512.
35271 (avx512f_perm<mode>): New define_expand.
35272 (avx512f_perm<mode>_mask): Likewise.
35273 (avx512f_perm<mode>_1<mask_name>): New define_insn.
35274 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
35276 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
35278 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
35279 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
35281 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
35283 2017-03-06 Martin Liska <mliska@suse.cz>
35286 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
35287 when having a SSA NAME w/o VAR_DECL assigned to it.
35289 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
35291 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
35292 msa_dpsub_<su>_d): Fix MODE for vec_select.
35294 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
35296 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
35298 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
35300 2017-03-06 Richard Biener <rguenther@suse.de>
35302 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
35303 * plugin.c (register_plugin_info): Likewise.
35304 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
35306 2017-03-05 Jakub Jelinek <jakub@redhat.com>
35308 * config/i386/sse.md (sse_storehps, sse_storelps,
35309 avx_<castmode><avxsizesuffix>_<castmode>,
35310 avx512f_<castmode><avxsizesuffix>_<castmode>,
35311 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
35312 in condition that at least one operand is not a MEM.
35314 2017-03-03 Jakub Jelinek <jakub@redhat.com>
35316 PR middle-end/79805
35317 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
35318 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
35320 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
35321 gimple_call_nothrow_p flag based on whether original builtin can throw.
35322 If it can, emit following stmts on the fallthrough edge.
35323 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
35324 don't create new bb if inserting just debug stmts on the edge, try to
35325 insert them on the fallthru bb or just reset debug stmts.
35327 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
35330 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
35331 restore recog_data (including the operand rtxes inside it) around
35332 the call to get_insn_template.
35334 2017-03-03 Martin Sebor <msebor@redhat.com>
35336 PR tree-optimization/79699
35337 * context.c (context::~context): Free MPFR caches to avoid
35338 a memory leak on program exit.
35340 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35342 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
35343 Use wide_int::ulow () instead of .elt (0).
35345 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
35347 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
35348 (*pushxf): Limit oF constraint to 32bit targets and add oC
35349 constraint for 64bit targets.
35350 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
35351 (*pushdf): Change rmF constraint to rmC.
35353 2017-03-03 Martin Liska <mliska@suse.cz>
35355 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
35356 Remove unused variable.
35358 2017-03-03 Jakub Jelinek <jakub@redhat.com>
35361 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
35362 is a memory operand, increase num_memory.
35363 (ix86_expand_args_builtin): Likewise.
35365 2017-03-03 Jan Hubicka <jh@suse.cz>
35368 * ipa-devirt.c (maybe_record_node): Properly handle
35369 __cxa_pure_virtual visibility.
35371 2017-03-03 Martin Liska <mliska@suse.cz>
35373 PR tree-optimization/79803
35374 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
35376 (pass_loop_prefetch::execute): Disabled optimization if an
35377 assumption about L1 cache size is not met.
35379 2017-03-03 Martin Liska <mliska@suse.cz>
35381 PR rtl-optimization/79574
35382 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
35383 (hash_scan_set): Likewise.
35384 (dump_hash_table): Likewise.
35385 (hoist_code): Likewise.
35387 2017-03-03 Richard Biener <rguenther@suse.de>
35389 * fixed-value.c (fixed_from_string): Restore use of elt (1)
35390 in place of uhigh ().
35391 (fixed_convert_from_real): Likewise.
35393 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
35396 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
35398 2017-03-03 Richard Biener <rguenther@suse.de>
35400 PR middle-end/79818
35401 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
35402 TYPE_OVERFLOW_UNDEFINED check.
35404 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35406 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
35408 (vector_ae_<mode>_p): Likewise.
35409 (vector_nez_<mode>_p): Likewise.
35410 (vector_ne_v2di_p): Likewise.
35411 (vector_ae_v2di_p): Likewise.
35412 (vector_ne_<mode>_p): Likewise.
35413 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
35415 (vsx_tsqrt<mode>2_fe): Likewise.
35417 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
35420 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
35422 2017-03-02 Jakub Jelinek <jakub@redhat.com>
35424 PR rtl-optimization/79780
35425 * cprop.c (one_cprop_pass): When second and further conditional trap
35426 in a single basic block is turned into an unconditional trap, turn it
35427 into a deleted note to avoid RTL verification failures.
35429 2017-03-02 Richard Biener <rguenther@suse.de>
35431 * fold-const.c (const_binop): Use ulow () instead of elt (0).
35433 2017-03-02 Richard Biener <rguenther@suse.de>
35435 PR tree-optimization/79345
35437 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
35438 param and abort the walk, returning -1 if it is hit.
35439 (walk_aliased_vdefs): Take a limit param and pass it on.
35440 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
35441 defaulting to 0 and return a signed int.
35442 * tree-ssa-uninit.c (struct check_defs_data): New struct.
35443 (check_defs): New helper.
35444 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
35445 about uninitialized memory.
35446 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
35447 bogus uninitialized warning.
35448 (fixed_convert_from_real): Likewise.
35450 2017-03-02 Bin Cheng <bin.cheng@arm.com>
35452 PR tree-optimization/66768
35453 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
35454 iv_use if base object can't be determined.
35456 2017-03-02 Jakub Jelinek <jakub@redhat.com>
35458 PR tree-optimization/79345
35459 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
35460 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
35461 (get_pattern_stats): Initialize it.
35462 * genemit.c (gen_expand): Verify match_scratch numbers come after
35463 match_operand/match_dup numbers.
35464 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
35465 match_scratch numbers.
35466 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
35468 * config/s390/s390.md (trunctdsd2): Likewise.
35470 2017-03-02 Richard Biener <rguenther@suse.de>
35472 * wide-int.h (wide_int_storage::operator=): Implement in terms
35475 2017-03-02 Richard Biener <rguenther@suse.de>
35477 PR tree-optimization/79777
35478 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
35479 the to insert expression to sth existing.
35481 2017-03-01 Martin Sebor <msebor@redhat.com>
35483 PR middle-end/79692
35484 * gimple-ssa-sprintf.c
35485 (directive::known_width_and_precision): New function.
35486 (format_integer): Use it.
35487 (get_mpfr_format_length): Consider the full range of precision
35488 when computing %g output with the # flag. Set the likely byte
35489 count to 3 rather than 1 when precision is indeterminate.
35490 (format_floating): Correct the lower bound of precision.
35492 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35494 * doc/invoke.texi: Document default code model for 64-bit Linux.
35496 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35499 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
35500 udiv rather than div since input pattern is unsigned.
35502 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
35504 * config/i386/i386.c (print_reg): Warn for values of
35505 unsupported size in integer register.
35507 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
35510 * config/rs6000/predicates.md (current_file_function_operand): Do
35511 not allow self calls to be local if the function is replaceable.
35513 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
35516 * config/rs6000/altivec.h (vec_ctz and others): Change the
35517 preprocessor macro that controls conditional compilation from
35518 _ARCH_PWR9 to __POWER9_VECTOR__.
35519 (vec_all_ne): Change parameterization of __altivec_scalar_pred
35520 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
35521 control (instead of _ARCH_PWR9 control) so that template
35522 definition uses power9-specific function.
35523 (vec_any_eq): Likewise.
35524 (vec_all_ne): Change macro definition to use a power9-specific
35525 expansion under #ifdef __POWER9_VECTOR__ control (instead of
35526 _ARCH_PWR9 control).
35527 (vec_any_eq) Likewise.
35528 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
35529 expansion for CMPNEF to remove support for xvcmpnesp instruction.
35530 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
35531 support for xvcmpnedp instruction.
35532 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
35533 macro expansion so that Power9 implementation of vec_all_ne does
35534 not use the AltiVec predicate framework.
35535 (VCMPNEH_P): Likewise.
35536 (VCMPNEW_P): Likewise.
35537 (VCMPNED_P): Likewise.
35538 (VCMPNEFP_P): Likewise.
35539 (VCMPNEDP_P): Likewise.
35540 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
35541 implementation of vec_any_eq to not use AltiVec predicate
35543 (VCMPAEH_P): Likewise.
35544 (VCMPAEW_P): Likewise.
35545 (VCMPAED_P): Likewise.
35546 (VCMPAEFP_P): Likewise.
35547 (VCMPAEDP_P): Likewise.
35548 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
35549 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
35550 not use the AltiVec predicate framework.
35551 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
35552 of vec_any_eq to not use AltiVec predicate framework.
35553 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
35554 support for predefined __POWER9_VECTOR__ macro to indicate that
35555 Power9 instruction selection is enabled.
35556 (altivec_overloaded_builtins): Remove extraneous
35557 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
35558 function argument types RS6000_BTI_bool_V16QI and
35559 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
35560 entry for overloaded function argument types RS6000_BTI_bool_V4SI
35561 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
35562 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
35563 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
35564 Power9 for implementations of vec_cmpne. Change the signature for
35565 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
35566 (representing vec_all_ne) to remove the previously described first
35567 argument of type RS6000_BTI_INTSI, as this was an artifact of
35568 reliance on the AltiVec predicate framework, which is no longer
35569 used in the implementation of these functions. Add
35570 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
35571 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
35572 since, unlike the AltiVec predicate framework implementation, we
35573 do not share function descriptors between vec_alle and vec_anyeq.
35574 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
35575 set of modes that receive special treatment even when
35576 TARGET_P9_VECTOR is true. The special treatment emits code that
35577 does not depend on Power9 instructions.
35578 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
35579 define_expand to not rely on AltiVec predicate framework.
35580 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
35582 (vector_ne_v2di_p): Change this define_expand to not rely on
35583 AltiVec predicate framework.
35584 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
35586 (vector_ne_<mode>_p): Change this define_expand to not rely on
35587 AltiVec predicate framework.
35588 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
35590 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
35591 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
35592 define_insn pattern.
35593 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
35594 define_insn pattern because the xvcmpne<VSs>. instruction is not
35596 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
35597 instruction is not supported.
35599 2017-03-01 Jakub Jelinek <jakub@redhat.com>
35601 * config/nvptx/nvptx.c: Include intl.h.
35603 2017-03-01 Martin Jambor <mjambor@suse.cz>
35606 * ipa-prop.h (ipa_bits): Removed field known.
35607 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
35608 to pointers. Adjusted their comments to warn about their sharing.
35609 (ipcp_transformation_summary): Change bits to a vector of pointers.
35610 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
35611 (ipa_get_ipa_bits_for_value): Declare.
35612 * tree-vrp.h (value_range): Mark as GTY((for_user)).
35613 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
35614 (ipa_bits_hash_table): Likewise.
35615 (ipa_vr_ggc_hash_traits): Likewise.
35616 (ipa_vr_hash_table): Likewise.
35617 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
35618 being pointers and vr_known being removed.
35619 (ipa_set_jf_unknown): Likewise.
35620 (ipa_get_ipa_bits_for_value): New function.
35621 (ipa_set_jfunc_bits): Likewise.
35622 (ipa_get_value_range): New overloaded functions.
35623 (ipa_set_jfunc_vr): Likewise.
35624 (ipa_compute_jump_functions_for_edge): Use the above functions to
35625 construct bits and vr parts of jump functions.
35626 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
35627 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
35629 (ipcp_grow_transformations_if_necessary): Also allocate
35630 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
35632 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
35633 them. Fix too long lines.
35634 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
35635 vr_known being removed.
35636 (ipa_read_jump_function): Use new setter functions to construct bits
35637 and vr parts of jump functions or set them to NULL.
35638 (write_ipcp_transformation_info): Adjust for bits being pointers.
35639 (read_ipcp_transformation_info): Likewise.
35640 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
35642 Include gt-ipa-prop.h.
35643 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
35645 (ipcp_store_bits_results): Likewise.
35646 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
35647 Do not write to existing jump functions but use a temporary instead.
35649 2017-03-01 Jakub Jelinek <jakub@redhat.com>
35652 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
35653 attempt to use its first operand as BIT_FIELD_REF base.
35655 2017-03-01 Richard Biener <rguenther@suse.de>
35657 PR middle-end/79721
35658 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
35659 interpolating formula in wrapping arithmetic.
35660 (chrec_apply): Convert chrec_evaluate return value to wanted type.
35662 2017-03-01 Jakub Jelinek <jakub@redhat.com>
35664 PR tree-optimization/79734
35665 * tree-vect-generic.c (expand_vector_condition): Optimize
35666 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
35667 Handle VEC_COND_EXPR where comparison has different inner width from
35668 type's inner width.
35670 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
35672 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
35673 markup, and similar issues. Remove @opindex entries for things
35674 that aren't options. Add missing -mmpy-option entries.
35676 2017-02-28 Jakub Jelinek <jakub@redhat.com>
35678 PR tree-optimization/79737
35679 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
35680 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
35681 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
35682 instead of byte_size. Formatting fix.
35683 (shift_bytes_in_array_right): Formatting fix.
35685 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
35688 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
35689 condition on optimize for the leaf function test.
35691 2017-02-28 Martin Liska <mliska@suse.cz>
35694 * read-rtl-function.c (function_reader::handle_unknown_directive):
35695 Bail out when one uses -flto.
35697 2017-02-28 Martin Liska <mliska@suse.cz>
35699 * common.opt: Replace space with tabular for options of <number>
35701 * config/i386/i386.opt: Show <number> value for
35702 -mlarge-data-threshold.
35703 * opts.c (print_filtered_help): Do not display number in hexadecimal
35706 2017-02-28 Martin Liska <mliska@suse.cz>
35708 * common.opt: Fix --help=option -Q for options which are of
35711 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
35713 * config/i386/i386.c (print_reg): Error out for values
35714 of 8-bit size in invalid integer register.
35716 2017-02-28 Martin Sebor <msebor@redhat.com>
35718 PR tree-optimization/79691
35719 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
35721 2017-02-28 Jakub Jelinek <jakub@redhat.com>
35724 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
35725 gcc_unreachable with output_operand_lossage.
35727 2017-02-28 Richard Biener <rguenther@suse.de>
35729 PR tree-optimization/79740
35730 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
35732 (visit_nary_op): Insert the nary into the hashtable if we
35733 pattern-matched sth.
35734 * tree-ssa-pre.c (eliminate_insert): Robustify.
35736 2017-02-28 Richard Biener <rguenther@suse.de>
35738 PR middle-end/79731
35739 * fold-const.c (decode_field_reference): Reject out-of-bound
35742 2017-02-28 Jakub Jelinek <jakub@redhat.com>
35744 * config/i386/i386.c: Include intl.h.
35745 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
35746 instead of just cond ? "..." : "...".
35747 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
35748 * coverage.c (read_counts_file): Likewise.
35749 * omp-offload.c: Include intl.h.
35750 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
35751 of just cond ? "..." : "...".
35752 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
35753 of just cond ? "..." : "...".
35755 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
35758 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
35760 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
35761 'tune for' CPU name.
35762 * config/arm/arm-cpu-data.h: Regenerated.
35764 2017-02-28 Richard Biener <rguenther@suse.de>
35766 PR tree-optimization/79732
35767 * tree-inline.c (expand_call_inline): Do not shadow var.
35769 2017-02-28 Richard Biener <rguenther@suse.de>
35771 PR tree-optimization/79723
35772 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
35773 address-space properly.
35775 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
35777 * doc/optinfo.texi (Optimization groups): Fix option used for
35779 * doc/invoke.texi (-fopt-info): Document "omp".
35780 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
35781 (OPTGROUP_ALL): Add OPTGROUP_OMP.
35782 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
35783 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
35784 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
35786 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
35788 * dumpfile.c (optgroup_options): Instead of "openmp", associate
35789 OPTGROUP_OMP with "omp".
35791 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
35794 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
35795 for arithmetic shift of unsigned V2DI.
35797 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
35799 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
35800 arc/linux.h headers.
35801 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
35802 (LINK_SPEC): Likewise.
35803 (ARC_TLS_EXTRA_START_SPEC): Likewise.
35804 (EXTRA_SPECS): Likewise.
35805 (STARTFILE_SPEC): Likewise.
35806 (ENDFILE_SPEC): Likewise.
35807 (LIB_SPEC): Likewise.
35808 (TARGET_SDATA_DEFAULT): Likewise.
35809 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
35810 (MULTILIB_DEFAULTS): Likewise.
35811 (DWARF2_UNWIND_INFO): Likewise.
35812 * config/arc/big.h: New file.
35813 * config/arc/elf.h: Likewise.
35814 * config/arc/linux.h: Likewise.
35815 * config/arc/t-uClibc: Remove.
35817 2017-02-27 Bin Cheng <bin.cheng@arm.com>
35819 PR tree-optimization/77536
35820 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
35821 (tree_transform_and_unroll_loop): Use above function to compute the
35822 estimated niter of unrolled loop and use it when scaling profile.
35823 Also use count info rather than frequency if it's non-zero.
35824 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
35825 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
35826 (vect_transform_loop): Call above function.
35828 2017-02-27 Richard Biener <rguenther@suse.de>
35830 PR tree-optimization/45397
35831 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
35832 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
35833 (visit_nary_op): Add pattern matching for CSEing sign-changed
35834 or truncated operations with wider ones.
35836 2017-02-27 Richard Biener <rguenther@suse.de>
35838 PR tree-optimization/79690
35839 * tree-vect-stmts.c (vectorizable_store): Use vector type
35840 built from the DR with address-space.
35842 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
35844 * doc/invoke.texi (Optimize Options): Refine the description
35845 of asan-use-after-return.
35847 2017-02-25 Alan Modra <amodra@gmail.com>
35849 PR rtl-optimization/79584
35850 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
35851 base, not ad->base_term, the reg within base. Remove assertion
35852 that ad->base == ad->base_term. Replace gen_int_mode using
35853 bogus mode with const0_rtx.
35855 2017-02-25 Jakub Jelinek <jakub@redhat.com>
35857 PR middle-end/79396
35858 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
35859 FMA_EXPR like tcc_binary or tcc_unary.
35861 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
35864 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
35866 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
35867 (output_loc_operands): Handle DW_OP_call_ref and
35868 DW_OP_GNU_variable_value.
35869 (struct variable_value_struct): New type.
35870 (struct variable_value_hasher): Likewise.
35871 (variable_value_hash): New variable.
35872 (string_types): Remove.
35873 (copy_loc_descr): New function.
35874 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
35875 (prepend_loc_descr_to_each): New function.
35876 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
35877 instead of add_loc_descr_to_each if the first argument is single
35878 location list and the second has multiple.
35879 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
35880 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
35881 when looking for variable value which doesn't have other location info.
35882 (loc_list_from_tree): Formatting fix.
35883 (gen_array_type_die): Simplify DW_AT_string_length handling.
35884 (adjust_string_types): Remove.
35885 (gen_subprogram_die): Don't call adjust_string_types nor test/set
35886 string_types. Call resolve_variable_values.
35887 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
35888 (resolve_addr_in_expr): Likewise. Add A argument.
35889 (copy_deref_exprloc): Remove deref argument. Adjust for the
35890 original expression being DW_OP_GNU_variable_value with optionally
35891 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
35892 optionally after it.
35893 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
35894 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
35895 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
35896 (variable_value_hasher::hash, variable_value_hasher::equal): New
35898 (resolve_variable_value_in_expr, resolve_variable_value,
35899 resolve_variable_values, note_variable_value_in_expr,
35900 note_variable_value): New functions.
35901 (dwarf2out_early_finish): Call note_variable_value on all toplevel
35904 2017-02-24 Jakub Jelinek <jakub@redhat.com>
35907 * opts.h (handle_generated_option): Add GENERATED_P argument.
35908 * opts-common.c (handle_option): Adjust function comment.
35909 (handle_generated_option): Add GENERATED_P argument, pass it to
35911 (control_warning_option): Pass false to handle_generated_option
35913 * opts.c (maybe_default_option): Pass true to handle_generated_option
35915 * optc-gen.awk: Likewise.
35917 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
35919 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
35920 a REG, look at the REG it is a SUBREG of.
35921 (splitter for cmpeqsi_t): Ditto.
35923 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
35925 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
35926 the special USEs with the pattern of the insn, not the insn itself.
35928 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
35931 * doc/invoke.texi: Document -mload-store-pairs.
35933 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
35934 Sandra Loosemore <sandra@codesourcery.com>
35936 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
35937 argument isn't a CONST_INT.
35938 (nios2_alternate_compare_const): Assert op is a CONST_INT.
35939 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
35940 (nios2_validate_compare): Bypass alternate compare logic if *op2
35941 is not a CONST_INT.
35942 (ldstwm_operation_p): Return false if first_base is not a REG or
35943 if first_offset is not a CONST_INT.
35945 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
35947 * config/cris/cris.md: Use correct operand in a define_peephole2.
35949 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
35951 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
35953 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
35955 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
35956 this_insn if it is an INSN or JUMP_INSN.
35957 (force_offsettable): Look at base, not at addr.
35958 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
35959 on things that aren't necessarily CONST_INTs.
35961 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
35963 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
35964 -mfpmath=sse is the default also for x86-32 targets with SSE2
35965 instruction set when @option{-ffast-math} is enabled
35967 2017-02-24 Jeff Law <law@redhat.com>
35969 PR rtl-optimizatoin/79286
35970 * ira.c (update_equiv_regs): Drop may_trap_p exception to
35973 2017-02-24 Richard Biener <rguenther@suse.de>
35975 PR tree-optimization/79389
35976 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
35979 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
35981 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
35982 function comment to reflect reality.
35983 (loop_exits_before_overflow): Fix typo in function description.
35985 2017-02-24 Richard Biener <rguenther@suse.de>
35987 PR tree-optimization/79389
35988 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
35989 properly that a threading opportunity exists. Detect conditional
35990 copy/constant propagation opportunities.
35992 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
35994 * config/visium/visium.md (type): Add trap.
35995 (b): New mode attribute.
35996 (*btst): Rename into...
35997 (*btst<mode>): ...this and adjust.
35998 (*cbranchsi4_btst_insn): Rename into...
35999 (*cbranch<mode>4_btst_insn): ...this and adjust.
36000 (trap): New define_insn.
36002 2017-02-23 Jakub Jelinek <jakub@redhat.com>
36004 PR tree-optimization/79389
36005 * ifcvt.c (struct noce_if_info): Add rev_cond field.
36006 (noce_reversed_cond_code): New function.
36007 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
36008 reversed_comparison_code. Formatting fix.
36009 (noce_try_store_flag): Test rev_cond != NULL in addition to
36010 reversed_comparison_code.
36011 (noce_try_store_flag_constants): Likewise.
36012 (noce_try_store_flag_mask): Likewise.
36013 (noce_try_addcc): Use rev_cond if non-NULL instead of
36014 reversed_comparison_code.
36015 (noce_try_cmove_arith): Likewise. Formatting fixes.
36016 (noce_try_minmax, noce_try_abs): Clear rev_cond.
36017 (noce_find_if_block): Initialize rev_cond.
36018 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
36019 instead of false as last argument never attempt to reverse it
36022 2017-02-23 Bin Cheng <bin.cheng@arm.com>
36024 PR tree-optimization/79663
36025 * tree-predcom.c (combine_chains): Process refs in reverse order
36026 only for ZERO length chains, and add explaining comment.
36028 2017-02-23 Jeff Law <law@redhat.com>
36030 PR tree-optimization/79578
36031 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
36032 in call to operand_equal_p.
36034 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
36037 * config/i386/cpuid.h: Fix another undefined behavior.
36039 2017-02-23 Richard Biener <rguenther@suse.de>
36041 PR tree-optimization/79683
36042 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
36043 vector types for data-refs.
36045 2017-02-23 Martin Liska <mliska@suse.cz>
36047 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
36049 2017-02-23 Jakub Jelinek <jakub@redhat.com>
36051 PR middle-end/79665
36052 * internal-fn.c (get_range_pos_neg): Moved to ...
36053 * tree.c (get_range_pos_neg): ... here. No longer static.
36054 * tree.h (get_range_pos_neg): New prototype.
36055 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
36056 are known to be in between 0 and signed maximum inclusive, try to
36057 expand both unsigned and signed divmod and use the cheaper one from
36060 2017-02-22 Jeff Law <law@redhat.com>
36062 PR tree-optimization/79578
36063 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
36064 to compare base operands.
36066 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
36069 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
36070 gpc_reg_operand instead of fpr_reg_operand.
36072 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
36074 * config/mips/mips.c (mips_return_in_memory): Force FP
36075 vector types to be returned in memory for o32 ABI.
36077 2017-02-22 Jakub Jelinek <jakub@redhat.com>
36079 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
36080 instead of DW_TAG_member for static data member declarations and don't
36081 set no_linkage_name for static inline data members.
36082 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
36085 2017-02-22 Martin Liska <mliska@suse.cz>
36087 * doc/invoke.texi: Replace inequality signs with square brackets
36090 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
36093 * lra-constraints.c (simplify_operand_subreg): Handle
36094 WORD_REGISTER_OPERATIONS targets.
36096 2017-02-22 Jakub Jelinek <jakub@redhat.com>
36099 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
36100 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
36101 elimination by swapping fld*.
36103 2017-02-22 Richard Biener <rguenther@suse.de>
36105 PR tree-optimization/79673
36106 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
36107 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
36108 irrelevant address-space qualifiers and avoiding a
36109 ADDR_SPACE_CONVERT_EXPR from fold_convert.
36111 2017-02-22 Richard Biener <rguenther@suse.de>
36113 PR tree-optimization/79666
36114 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
36115 to not symbolically negate if that may introduce undefined
36118 2017-02-22 Martin Liska <mliska@suse.cz>
36121 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
36122 * data-streamer-out.c (streamer_write_gcov_count_stream):
36124 * value-prof.c (stream_out_histogram_value): Make assert more
36125 precise based on type of counter.
36127 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
36130 * config/i386/i386.md (standard_x87sse_constant_load splitter):
36131 Use nonimmediate_operand instead of memory_operand for operand 1.
36132 (float-extend standard_x87sse_constant_load splitter): Ditto.
36134 2017-02-21 Jeff Law <law@redhat.com>
36136 PR tree-optimization/79621
36137 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
36138 blocks with edges to themselves.
36140 2017-02-21 Jakub Jelinek <jakub@redhat.com>
36143 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
36144 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
36145 Use gimple_call_builtin_p.
36148 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
36149 on temporarily removed DEBUG_INSNs.
36151 PR tree-optimization/79649
36152 * tree-loop-distribution.c (classify_partition): Give up on
36153 non-generic address space loads/stores.
36155 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
36157 * doc/loop.texi (Loop manipulation): Remove nonexistent
36158 tree_ssa_loop_version from the documentation.
36159 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
36161 2017-02-21 Jakub Jelinek <jakub@redhat.com>
36164 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
36165 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
36166 * config/rs6000/rs6000.c: Include except.h.
36167 (rs6000_expand_split_stack_prologue): Call
36168 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
36170 2017-02-21 Martin Jambor <mjambor@suse.cz>
36173 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
36174 have been analyzed.
36176 2017-02-21 Martin Jambor <mjambor@suse.cz>
36178 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
36179 for backward compatibility only.
36180 * doc/invoke.texi (Option Summary): Remove all references to
36181 -fipa-cp-alignment.
36183 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
36187 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36189 * lra-constraints.c (curr_insn_transform): Handle
36190 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
36192 2017-02-21 Martin Liska <mliska@suse.cz>
36194 * config/i386/i386.opt: Replace -masm-dialect with -masm.
36196 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
36198 PR translation/79638
36199 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
36201 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
36204 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
36205 (arm_function_ok_for_sibcall): Return false for an indirect call by
36206 descriptor if all the argument registers are used.
36207 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
36208 alignment of the function.
36210 2017-02-21 Jakub Jelinek <jakub@redhat.com>
36212 PR tree-optimization/61441
36213 * simplify-rtx.c (simplify_const_unary_operation): For
36214 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
36215 the sNaN unmodified.
36217 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
36219 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
36220 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
36221 instead of SYSTEM_HEADER_DIR.
36223 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
36224 Martin Liška <mliska@suse.cz>
36226 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
36227 Fix typos and grammar, use active voice, and clarify.
36229 2017-02-20 Marek Polacek <polacek@redhat.com>
36231 PR middle-end/79537
36232 * gimplify.c (gimplify_expr): Handle unused *&&L;.
36235 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
36237 2017-02-20 Jakub Jelinek <jakub@redhat.com>
36240 * config/i386/i386.c (ix86_expand_builtin): Handle
36241 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
36242 ix86_builtins_isa[fcode].isa as a requirement of those
36243 flags and any other flag in the bitmask.
36244 (ix86_init_mmx_sse_builtins): Use 0 instead of
36245 ~OPTION_MASK_ISA_64BIT as mask.
36246 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
36247 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
36248 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
36249 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
36251 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36254 * lra-constraints.c (split_reg): Check requested split mode
36255 is supported by the register.
36257 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36259 * lra-constraints.c (simplify_operand_subreg): Remove early
36262 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36265 * lra-constraints.c (curr_insn_transform): Tighten condition
36266 for converting SUBREG reloads from OP_OUT to OP_INOUT.
36268 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36271 * lra-constraints.c (curr_insn_transform): Handle
36272 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
36274 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
36277 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
36279 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
36281 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
36284 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
36287 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
36289 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
36292 2017-02-18 Jakub Jelinek <jakub@redhat.com>
36294 * final.c (last_columnnum, override_columnnum): New variables.
36295 (final_start_function): Set last_columnnum, pass it to begin_prologue
36296 hook and pass 0 to dwarf2out_begin_prologue.
36297 (final_scan_insn): Update override_columnnum. Pass last_columnnum
36298 to source_line debug hook.
36299 (notice_source_line): Compute last_columnnum and for debug_column_info
36300 return true on column changes.
36301 * debug.h (struct gcc_debug_hooks): Add column argument to
36302 source_line and begin_prologue hooks.
36303 (debug_nothing_int_charstar_int_bool): Remove prototype.
36304 (debug_nothing_int_int_charstar,
36305 debug_nothing_int_int_charstar_int_bool): New prototypes.
36306 (dwarf2out_begin_prologue): Add column argument.
36307 * debug.c (do_nothing_debug_hooks): Adjust source_line and
36308 begin_prologue hooks.
36309 (debug_nothing_int_charstar_int_bool): Remove.
36310 (debug_nothing_int_int_charstar,
36311 debug_nothing_int_int_charstar_int_bool): New functions.
36312 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
36313 through to dwarf2out_source_line.
36314 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
36315 (dwarf2out_source_line): Add column argument, emit it if requested.
36316 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
36318 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
36319 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
36320 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
36321 through to dwarf2out_begin_prologue.
36322 (vmsdbgout_source_line): Add column argument, pass it through to
36323 dwarf2out_source_line.
36324 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
36325 dbxout_source_line caller.
36326 (dbxout_source_line): Add column argument.
36328 * common.opt (gno-column-info, gcolumn-info): New options.
36329 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
36330 (check_die): Also test for multiple DW_AT_decl_column attributes.
36331 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
36332 DW_AT_decl_column if requested.
36333 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
36335 (gen_variable_die): Likewise.
36336 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
36337 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
36340 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
36341 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
36342 (ix86_handle_option): Handle OPT_m3dnowa.
36343 * doc/invoke.texi (-m3dnowa): Document.
36344 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
36345 -m3dnowa instead of -m3dnow -march=athlon.
36348 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
36349 instead of gcc_assert for K, r and R code checks. Formatting fixes.
36351 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36354 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
36355 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
36356 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
36357 generator for vsx_xxpermdi_<mode>_be.
36358 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
36359 force big-endian semantics.
36360 (vsx_xxpermdi_<mode>_be): New define_expand with same
36361 implementation as previous version of vsx_xxpermdi_<mode>.
36363 2017-02-17 Jakub Jelinek <jakub@redhat.com>
36365 PR tree-optimization/79327
36366 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
36367 variable, its initialization and use.
36369 2017-02-17 Julia Koval <julia.koval@intel.com>
36371 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
36372 (OPTION_MASK_ISA_PKU_UNSET): New.
36373 (ix86_handle_option): Handle -mrdpid.
36374 * config/i386/cpuid.h (bit_RDPID): New.
36375 * config/i386/driver-i386.c (host_detect_local_cpu):
36376 Detect RDPID feature.
36377 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
36378 * config/i386/i386-c.c (ix86_target_macros_internal):
36380 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
36381 (ix86_valid_target_attribute_inner_p): Add "rdpid".
36382 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
36383 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
36384 * config/i386/i386.md (define_insn "rdpid"): New.
36385 * config/i386/i386.opt Add -mrdpid.
36386 * config/i386/immintrin.h (_rdpid_u32): New.
36388 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
36390 PR rtl-optimization/79541
36391 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
36392 instead of transforming it into USE.
36394 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
36396 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
36397 If HONOR_SNANS (SFmode) force the input to a register.
36398 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
36399 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
36400 an frsp or similar insn.
36402 2017-02-17 Martin Liska <mliska@suse.cz>
36404 PR rtl-optimization/79577
36405 * params.def (selsched-max-sched-times): Increase minimum to 1.
36407 2017-02-17 Martin Liska <mliska@suse.cz>
36409 PR rtl-optimization/79574
36410 * gcse.c (want_to_gcse_p): Prevent integer overflow.
36412 2017-02-17 Martin Liska <mliska@suse.cz>
36414 PR tree-optimization/79529
36415 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
36416 ssa_defined_default_def_p to handle cases which are implicitly
36418 * tree-ssa.c (ssa_defined_default_def_p): New function.
36419 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
36420 which are implicitly defined.
36421 * tree-ssa.h (ssa_defined_default_def_p): Declare.
36423 2017-02-17 Richard Biener <rguenther@suse.de>
36425 PR middle-end/79576
36426 * params.def (max-ssa-name-query-depth): Limit to 10.
36428 2017-02-17 Richard Biener <rguenther@suse.de>
36430 PR tree-optimization/79552
36431 * tree-ssa-structalias.c (visit_loadstore): Properly verify
36434 2017-02-17 Richard Biener <rguenther@suse.de>
36437 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
36439 2017-02-17 Marek Polacek <polacek@redhat.com>
36441 PR middle-end/79536
36442 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
36443 (fold_negate_expr): New wrapper.
36445 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
36447 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
36448 Correct terminology and de-emphasize pre-standard behavior.
36450 2017-02-16 Alan Modra <amodra@gmail.com>
36452 PR rtl-optimization/79286
36453 * ira.c (def_dominates_uses): New function.
36454 (update_equiv_regs): Don't create an equivalence for insns that
36455 may trap where the register def does not dominate the use.
36457 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
36459 PR rtl-optimization/78127
36460 * lra.c (lra): Call lra_eliminate before finish the loop after
36463 2017-02-16 Richard Biener <rguenther@suse.de>
36465 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
36467 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
36468 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
36469 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
36470 (isl_val_int_from_wi): New function.
36471 (extract_affine_gmp): Rename to ...
36472 (extract_affine_wi): ... this, take a widest_int.
36473 (extract_affine_int): Just wrap extract_affine_wi.
36474 (add_param_constraints): Use isl_val_int_from_wi.
36475 (add_loop_constraints): Likewise, and extract_affine_wi.
36477 2017-02-15 Jeff Law <law@redhat.com>
36479 PR middle-end/79521
36480 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
36481 ira_init_register_move_cost_if_necessary.
36483 2017-02-15 Martin Sebor <msebor@redhat.com>
36485 PR middle-end/32003
36486 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
36487 removed in a prior commit.
36489 2017-02-15 Bin Cheng <bin.cheng@arm.com>
36491 PR tree-optimization/79347
36492 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
36493 counters during peeling.
36495 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
36497 * Makefile.in (site.exp): Remove "set ISLVER".
36499 2017-02-15 Jakub Jelinek <jakub@redhat.com>
36502 * real.c (real_from_integer): Call real_convert even for decimal.
36504 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
36507 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
36509 2017-02-14 Andrew Pinski <apinski@cavium.com>
36511 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
36512 cores and change the partno/implementer to be correct.
36513 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
36514 the 'B" as the implementer.
36515 * config/aarch64/aarch64-tune.md: Regenerate.
36517 2017-02-14 Carl Love <cel@us.ibm.com>
36519 * config/rs6000/rs6000.c: Add case statement entry to make the
36520 xvcvuxdsp built-in argument unsigned.
36521 * config/rs6000/vsx.md: Fix the source and return operand types so they
36522 match the instruction definitions from the ISA document. Fix typo
36523 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
36526 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
36529 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
36530 member early_clobber_alts.
36531 * lra-lives.c (reg_early_clobber_p): New.
36532 (process_bb_lives): Use it.
36533 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
36534 (debug_operand_data): Initialize early_clobber_alts.
36535 (setup_operand_alternative): Set up early_clobber_alts.
36536 (collect_non_operand_hard_regs): Ditto. Pass early clobber
36537 alternatives to new_insn_reg.
36538 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
36540 (lra_update_insn_regno_info): Pass the new arg.
36542 2017-02-14 Jakub Jelinek <jakub@redhat.com>
36544 PR middle-end/79505
36545 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
36546 (new_oacc_loop_raw): Don't clear already cleared fields.
36549 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
36550 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
36551 _mm512_prefetch_i64gather_ps): New inline functions and macros.
36553 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
36556 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
36558 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
36561 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
36562 the extra instruction to the right place to store 128-bit constant
36565 2017-02-14 Martin Sebor <msebor@redhat.com>
36567 PR middle-end/79448
36568 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
36569 warning for strings of unknown length.
36571 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
36573 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
36575 2017-02-14 Jeff Law <law@redhat.com>
36578 * ira-costs.c (scan_one_insn): Initialize register move costs
36579 for pseudos seen in USE/CLOBBER insns.
36581 PR tree-optimization/79095
36582 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
36583 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
36584 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
36585 if the operands are known to be not equal, then the resulting range
36587 (intersect_ranges): If the new range is ~[0,0] and the old range is
36588 wide, then prefer ~[0,0].
36589 * tree-vrp.c (overflow_comparison_p_1): New function.
36590 (overflow_comparison_p): New function.
36591 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
36592 if NAME is used in an overflow test.
36593 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
36594 overflow check that can be expressed as an equality test, then adjust
36595 ops to be that equality test.
36597 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36599 * config/s390/s390-builtin-types.def: Remove flags argument.
36600 * config/s390/s390.c (s390_init_builtins): Likewise.
36602 2017-02-14 Martin Liska <mliska@suse.cz>
36604 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
36605 vector. Fix trailing white spaces.
36607 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
36609 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
36612 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36614 PR rtl-optimization/68664
36615 * config/arm/arm.c (arm_sched_can_speculate_insn):
36616 New function. Declare prototype.
36617 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
36619 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36621 PR rtl-optimization/68664
36622 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
36624 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
36626 2017-02-14 Amit Pawar <amit.pawar@amd.com>
36628 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
36629 max skip bytes for function, loop and jump.
36631 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
36633 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
36634 ABS_EXPR for gimple dump.
36636 2017-02-14 Jakub Jelinek <jakub@redhat.com>
36639 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
36641 PR tree-optimization/79408
36642 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
36643 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
36644 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
36645 also if rhs1 is INTEGER_CST.
36647 2017-02-14 Richard Biener <rguenther@suse.de>
36649 PR middle-end/79432
36650 * tree-into-ssa.c (insert_phi_nodes): When the function can
36651 have abnormal edges rewrite SSA names with broken use-def
36652 dominance out of SSA and register them for PHI insertion.
36654 2017-02-13 Martin Sebor <msebor@redhat.com>
36656 PR middle-end/79496
36657 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
36658 clearing info.nowrite flag when snprintf size argument is a range.
36660 2017-02-13 Jakub Jelinek <jakub@redhat.com>
36662 * cprop.c (cprop_jump): Add missing space in string literal.
36663 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
36664 (get_constraint_for_component_ref): Likewise.
36665 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
36666 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
36667 * lra-constraints.c (process_alt_operands): Likewise.
36668 * ipa-inline.c (inline_small_functions): Likewise.
36669 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
36670 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
36671 * trans-mem.c (diagnose_tm_1_op): Likewise.
36672 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
36673 (grid_parallel_clauses_gridifiable): Likewise.
36675 * config/nvptx/mkoffload.c (process): Add space in between
36678 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
36679 "MOD4_SSE_REGS" and "ALL_REGS".
36681 * spellcheck.c (test_data): Add , in between "foo" and "food".
36683 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36686 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
36687 boundary crossing check and subsequent code generation agree.
36689 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36691 * config/aarch64/aarch64.c (has_memory_op): Delete.
36692 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
36695 2017-02-13 Jakub Jelinek <jakub@redhat.com>
36697 PR rtl-optimization/79388
36698 PR rtl-optimization/79450
36699 * combine.c (distribute_notes): When removing TEM_INSN for which
36700 corresponding dest has last value recorded, invalidate that last
36703 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36705 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
36706 of explicit '@'. Add missing assembly comment marker on branch costs
36709 2017-02-13 Nathan Sidwell <nathan@acm.org>
36711 * gengtype-lex.l (<in_struct>): Add '/'.
36713 2017-02-13 Martin Liska <mliska@suse.cz>
36716 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
36718 2017-02-13 Richard Biener <rguenther@suse.de>
36720 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
36722 * configure: Re-generate.
36723 * config.in: Likewise.
36724 * graphite-dependences.c: Simplify as if
36725 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
36726 * graphite-isl-ast-to-gimple.c: Likewise.
36727 * graphite-optimize-isl.c: Likewise.
36728 * graphite-poly.c: Likewise.
36729 * graphite-sese-to-poly.c: Likewise.
36730 * graphite.h: Likewise.
36731 * toplev.c: Include isl/version.h and use isl_version () for
36732 printing the ISL version.
36733 * doc/install.texi: Update ISL requirement.
36735 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
36737 * doc/standards.texi (Standards): Update reference to
36740 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
36742 * doc/extend.texi (Named Address Spaces): sourceware.org now
36744 * doc/install.texi (Binaries): Ditto.
36747 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
36749 * doc/cpp.texi: Replace "stringify"/"stringification" with C
36750 standard terminology "stringize"/"stringizing" throughout.
36751 * doc/cppinternals.texi: Likewise.
36753 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
36755 * doc/extend.texi: Fix some spelling mistakes and typos.
36756 * doc/invoke.texi: Likewise.
36758 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
36761 * params.def (inline-min-speedup) Change from 10 to 8.
36763 2017-02-11 Jakub Jelinek <jakub@redhat.com>
36765 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
36768 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
36771 * ipa-inline-analysis.c (get_minimal_bb): New function.
36772 (record_modified): Use it.
36773 (remap_edge_change_prob): Handle also ancestor functions.
36775 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
36777 * doc/contrib.texi (Contributors): Remove broken link into
36778 the Mauve CVS repository.
36780 2017-02-11 Jakub Jelinek <jakub@redhat.com>
36782 PR middle-end/79454
36783 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
36784 result computation whenever lhs doesn't have vector mode, not
36785 just when it has BLKmode.
36787 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
36789 * doc/makefile.texi (profiledbootstrap): Refer to the
36790 installation instructions only in textual form.
36792 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36795 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
36797 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
36799 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
36800 (Specific): Update mingw-w64 reference.
36802 (Specific): Remove broken link to Renesas RX processor.
36804 2017-02-10 Richard Biener <rguenther@suse.de>
36806 * toplev.c (process_options): Do not mention obsolete graphite
36807 options when printing sorry message about missing graphite support.
36808 Mention -floop-nest-optimize.
36810 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
36812 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
36813 (vtst_p16): Likewise.
36814 (vtstq_p8): Likewise.
36815 (vtstq_p16): Likewise.
36817 (vtstq_p64): Likewise.
36818 * config/arm/arm_neon.h (vgetq_lane_p64): New.
36819 (vset_lane_p64): New.
36820 (vsetq_lane_p64): New.
36822 2017-02-10 Jakub Jelinek <jakub@redhat.com>
36824 PR tree-optimization/79411
36825 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
36826 stmt operands are SSA_NAMEs used in abnormal phis.
36827 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
36830 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
36833 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
36836 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
36838 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
36840 2017-02-09 Jakub Jelinek <jakub@redhat.com>
36842 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
36846 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
36847 not arbitrary TREE_CONSTANT.
36850 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
36851 "omp declare target link" attribute unless is_global_var.
36852 * omp-offload.c (find_link_var_op): Likewise.
36854 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
36855 Chung-Lin Tang <cltang@codesourcery.com>
36857 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
36859 (gimplify_adjust_omp_clauses): Don't delete TILE.
36860 (gimplify_omp_for): Deal with TILE.
36861 * internal-fn.c (expand_GOACC_TILE): New function.
36862 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
36864 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
36865 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
36867 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
36868 avoid DIV for outermost collapse var.
36869 (expand_oacc_for): Insert tile element loop as needed. Adjust.
36870 Remove out of date comments, fix whitespace.
36871 * omp-general.c (omp_extract_for_data): Deal with tiling.
36872 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
36873 adjust OLF_DIM_BASE value.
36874 (struct omp_for_data): Add tiling field.
36875 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
36876 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
36877 for auto loops. Remove default auto determining, moved to
36878 oacc_loop_fixed_partitions.
36879 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
36880 stmts, add e_mask field.
36881 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
36882 (oacc_thread_numbers): Use oacc_dim_call.
36883 (oacc_xform_tile): New.
36884 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
36885 (finish_oacc_loop): Adjust for ifns vector.
36886 (oacc_loop_discover_walk): Append loop abstraction sites to list,
36887 add case for GOACC_TILE fns.
36888 (oacc_loop_xform_loop): Delete.
36889 (oacc_loop_process): Iterate over call list directly, and add
36890 handling for GOACC_TILE fns.
36891 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
36893 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
36894 vector partitioning to outer loops. Assign 2 partitions to loops
36895 when available. Add TILE handling.
36896 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
36897 (execite_oacc_device_lower): Process GOACC_TILE fns,
36898 ignore unknown specs.
36899 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
36900 * tree.c (omp_clause_num_ops): Adjust TILE ops.
36901 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
36903 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
36905 * configure.ac (ACX_BUGURL): Update.
36906 * configure: Regenerate.
36908 2017-02-09 Richard Biener <rguenther@suse.de>
36910 PR tree-optimization/69823
36911 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
36912 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
36914 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
36916 * config/arc/arc-c.def: Add __NPS400__ definition.
36917 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
36918 (TARGET_NPS400): Define.
36920 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
36922 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
36924 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
36925 pointer, arch_info.
36926 (arc_cpu_types): Fill the arch_info field with a pointer into the
36927 arc_arch_types table.
36928 (arc_selected_cpu): Declare.
36929 * config/arc/arc.c (arc_selected_cpu): Make global.
36930 (arc_selected_arch): Delete.
36931 (arc_base_cpu): Delete.
36932 (arc_override_options): Remove references to deleted variables,
36933 update access to arch information.
36934 (ARC_OPT): Update access to arch information.
36935 (ARC_OPTX): Likewise.
36936 * config/arc/arc.h (arc_base_cpu): Remove declaration.
36937 (TARGET_ARC600): Update access to arch information.
36938 (TARGET_ARC601): Likewise.
36939 (TARGET_ARC700): Likewise.
36940 (TARGET_EM): Likewise.
36941 (TARGET_HS): Likewise.
36942 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
36945 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
36948 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
36949 condition/operands for integer GE/LE/GEU/LEU operations.
36951 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
36953 PR translation/79397
36954 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
36957 2017-02-08 Martin Jambor <mjambor@suse.cz>
36960 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
36961 whether allocation happened.
36962 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
36963 nothing was allocated.
36965 2017-02-08 Jakub Jelinek <jakub@redhat.com>
36967 PR tree-optimization/79408
36968 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
36969 constant, but SSA_NAME with a known integer range, use the minimum
36970 of that range instead of op1 to determine if modulo can be replaced
36971 with its first operand.
36973 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36975 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
36977 2017-02-08 Richard Biener <rguenther@suse.de>
36979 PR tree-optimization/71824
36980 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
36981 Check all loops contained in the merged region.
36983 2017-02-07 Andrew Pinski <apinski@cavium.com>
36985 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
36987 2017-02-07 Andrew Pinski <apinski@cavium.com>
36989 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
36990 (thunderxt88): Likewise.
36991 (thunderxt81): Disable LSE and change v8.1 to v8.
36992 (thunderxt83): Likewise.
36994 2017-02-07 Jakub Jelinek <jakub@redhat.com>
36995 Richard Biener <rguenther@suse.de>
36997 PR middle-end/79399
36998 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
36999 type from int to size_t.
37000 * ira-costs.c (struct_costs_size): Change type from int to size_t.
37002 2017-02-07 Jakub Jelinek <jakub@redhat.com>
37004 PR rtl-optimization/79386
37005 * cprop.c (bypass_conditional_jumps): Initialize
37006 bypass_last_basic_block already before splitting bbs after
37007 unconditional traps...
37008 (bypass_conditional_jumps): ... rather than here.
37011 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
37012 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
37013 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
37014 fix -masm=intel patterns.
37016 2017-02-07 Richard Biener <rguenther@suse.de>
37018 PR tree-optimization/79256
37019 PR middle-end/79278
37020 * builtins.c (get_object_alignment_2): Use min_align_of_type
37021 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
37022 and ADJUST_FIELD_ALIGN.
37024 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
37026 * doc/tm.texi: Regenerate.
37027 * stor-layout.c (layout_decl): Adjust.
37028 (update_alignment_for_field): Likewise.
37029 (place_field): Likewise.
37030 (min_align_of_type): Likewise.
37031 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
37032 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
37033 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
37034 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
37035 * config/frv/frv.c (frv_adjust_field_align): Likewise.
37036 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
37037 * config/i386/i386.c (x86_field_alignment): Likewise.
37038 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
37039 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
37040 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
37041 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
37042 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
37043 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
37047 2017-01-30 Richard Biener <rguenther@suse.de>
37049 PR tree-optimization/79256
37050 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
37051 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
37054 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
37056 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
37057 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
37058 builtins to SImode and emit a zero-extend, if necessary.
37060 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
37062 * docs/invoke.texi (RISC-V Options): Alphabetize.
37064 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
37066 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
37069 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
37071 * config/riscv/riscv.c: New file.
37072 * common/config/riscv/riscv-common.c: Likewise.
37073 * config.gcc: Likewise.
37074 * config/riscv/constraints.md: Likewise.
37075 * config/riscv/elf.h: Likewise.
37076 * config/riscv/generic.md: Likewise.
37077 * config/riscv/linux.h: Likewise.
37078 * config/riscv/multilib-generator: Likewise.
37079 * config/riscv/peephole.md: Likewise.
37080 * config/riscv/pic.md: Likewise.
37081 * config/riscv/predicates.md: Likewise.
37082 * config/riscv/riscv-builtins.c: Likewise.
37083 * config/riscv/riscv-c.c: Likewise.
37084 * config/riscv/riscv-ftypes.def: Likewise.
37085 * config/riscv/riscv-modes.def: Likewise.
37086 * config/riscv/riscv-opts.h: Likewise.
37087 * config/riscv/riscv-protos.h: Likewise.
37088 * config/riscv/riscv.h: Likewise.
37089 * config/riscv/riscv.md: Likewise.
37090 * config/riscv/riscv.opt: Likewise.
37091 * config/riscv/sync.md: Likewise.
37092 * config/riscv/t-elf-multilib: Likewise.
37093 * config/riscv/t-linux: Likewise.
37094 * config/riscv/t-linux-multilib: Likewise.
37095 * config/riscv/t-riscv: Likewise.
37096 * configure.ac: Likewise.
37097 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
37098 Waterman as RISC-V maintainers.
37099 * doc/install.texi: Add RISC-V entries.
37100 * doc/invoke.texi: Add RISC-V options section.
37101 * doc/md.texi: Add RISC-V constraints section.
37102 * configure: Regenerated.
37104 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
37107 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
37108 false values to be constant vectors with all 0 or all 1 bits set.
37109 (vcondu<mode><mode>): Likewise.
37110 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
37112 (fpmask_comparison_operator): Update comment.
37113 (vecint_comparison_operator): New predicate.
37114 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
37115 vector conditionals when the true and false values are constant
37116 vectors with all 0 bits or all 1 bits set.
37118 2017-02-06 Martin Sebor <msebor@redhat.com>
37120 PR tree-optimization/79376
37121 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
37123 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
37125 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
37126 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
37127 to simplify split condition.
37129 2017-02-06 Jakub Jelinek <jakub@redhat.com>
37131 * omp-expand.c (oxpand_omp_atomic_fetch_op,
37132 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
37135 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
37137 PR rtl-optimization/68664
37138 * target.def (can_speculate_insn): New hook.
37139 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
37140 * doc/tm.texi: Regenerate.
37141 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
37142 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
37143 (rs6000_sched_can_speculate_insn): New function.
37145 2017-02-06 Jakub Jelinek <jakub@redhat.com>
37147 PR tree-optimization/79284
37148 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
37149 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
37150 vectorizable_mask_load_store, vectorizable_operation,
37151 vect_is_simple_cond, get_same_sized_vectype): Use it instead
37152 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
37153 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
37154 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
37155 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
37156 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
37157 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
37158 is_gimple_assign (stmt). Replace another such test with
37159 is_gimple_assign (stmt).
37161 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
37164 * config/avr/avr.c (rtl-iter.h): Include it.
37165 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
37166 (avr_legitimate_combined_insn): ...and implementation.
37168 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
37170 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
37171 * config/s390/s390.c (s390_const_operand_ok)
37172 (s390_canonicalize_comparison, s390_extract_part)
37173 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
37174 (s390_contiguous_bitmask_p, s390_rtx_costs)
37175 (legitimize_pic_address): Likewise.
37176 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
37177 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
37178 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
37179 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
37180 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
37182 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
37184 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
37185 REGNO($0) == REGNO($1).
37187 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37189 * config/s390/linux.h(SIZE_TYPE): Add comment.
37191 2017-02-06 Julian Brown <julian@codesourcery.com>
37192 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
37193 Virendra Pathak <virendra.pathak@broadcom.com>
37195 * config/aarch64/aarch64-cores.def: Change the scheduler
37197 * config/aarch64/aarch64.md: Include thunderx2t99.md.
37198 * config/aarch64/thunderx2t99.md: New file.
37200 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
37202 * doc/standards.texi (Go Language): Update link to language
37205 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
37207 * tree-eh.c (lower_resx): Sanitize profile.
37208 (cleanup_empty_eh_move_lp): Likewise.
37210 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
37213 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
37215 * cfgloopmanip.h (loop_version): Update prototype.
37216 * modulo-sched.c (sms_schedule): Update call of loop_version.
37217 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
37218 * tree-parloops.c (gen_parallel_loop): Likewise.
37219 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
37220 * tree-ssa-loop-split.c (split_loop): Likewise.
37221 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
37222 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
37224 2017-02-05 Martin Liska <mliska@suse.cz>
37227 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
37229 (print_operand_address): Initialize a struct to zero.
37231 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
37233 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
37234 garbage collector only in textual form.
37236 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
37238 * doc/extend.texi (x86 specific memory model extensions for
37239 transactional memory): Simplify a phrase.
37241 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
37244 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
37245 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
37246 (atomic_storedi_1): Likewise.
37248 2017-02-04 Jakub Jelinek <jakub@redhat.com>
37250 PR tree-optimization/79338
37251 * tree-parloops.c (gather_scalar_reductions): Don't call
37252 vect_analyze_loop_form for loop->inner before destroying loop's
37255 2017-02-03 Martin Sebor <msebor@redhat.com>
37257 PR tree-optimization/79327
37258 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
37259 when precision has resulted in leading zeros.
37260 (format_integer): Adjust the likely counter to assume an unknown
37261 argument that may be zero is non-zero.
37263 2017-02-03 Jason Merrill <jason@redhat.com>
37266 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
37267 avoid copying non-taken branch.
37269 2017-02-03 Jakub Jelinek <jakub@redhat.com>
37271 PR tree-optimization/79340
37272 * tree-vect-loop.c (vectorizable_reduction): Release
37273 vec_defs elements after safe_splicing them into other vectors.
37276 PR tree-optimization/79327
37277 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
37278 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
37280 (format_integer): Use wide_int_to_tree instead of build_int_cst
37281 + to_?hwi. If argmin is NULL, just set argmin and argmax to
37282 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
37283 of shortest and longest sequence.
37285 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
37287 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
37288 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
37290 2017-02-03 Walter Lee <walt@tilera.com>
37293 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
37294 after initial stackframe link reg save.
37295 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
37297 2017-02-03 Jakub Jelinek <jakub@redhat.com>
37300 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
37301 wu for stxssp alternative.
37303 2017-02-03 Martin Sebor <msebor@redhat.com>
37305 PR tree-optimization/79352
37306 * gimple-fold.c (get_range_strlen): Add argument.
37307 (get_range_strlen): Change return type to bool.
37308 (get_maxval_strlen): Pass in a dummy argument.
37309 * gimple-fold.h (get_range_strlen): Change return type to bool.
37310 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
37311 * tree.h (array_at_struct_end_p): Add argument.
37312 * tree.c (array_at_struct_end_p): Handle it.
37314 2017-02-03 Martin Liska <mliska@suse.cz>
37317 * multiple_target.c (create_dispatcher_calls): Redirect edge
37318 from a caller of a dispatcher.
37319 (expand_target_clones): Make the clones local.
37320 (ipa_target_clone): Do both target clones and resolvers.
37321 (ipa_dispatcher_calls): Remove the pass.
37322 (pass_dispatcher_calls::gate): Likewise.
37323 (make_pass_dispatcher_calls): Likewise.
37324 * passes.def (pass_target_clone): Put as very first IPA early
37327 2017-02-03 Martin Liska <mliska@suse.cz>
37329 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
37330 in case of a function with ifunc attribute.
37332 2017-02-03 Martin Liska <mliska@suse.cz>
37334 * cgraph.c (cgraph_node::dump): Dump function version info.
37335 * symtab.c (symtab_node::dump_base): Add missing new line.
37337 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
37339 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
37340 (ifcombine_ifandif): Use it.
37342 2017-02-03 Martin Liska <mliska@suse.cz>
37344 * doc/invoke.texi: Document default value for
37345 use-after-scope-direct-emission-threshold.
37347 2017-02-03 Martin Liska <mliska@suse.cz>
37349 PR tree-optimization/79339
37350 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
37351 (format_floating): Likewise.
37353 2017-02-03 Martin Liska <mliska@suse.cz>
37356 * ipa-prop.c (ipa_node_params_t::insert): Remove current
37358 (ipa_node_params_t::remove): Likewise.
37359 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
37360 initialization from removed ipa_node_params_t::insert.
37361 (ipa_node_params::~ipa_node_params): Move from removed
37362 ipa_node_params_t::release.
37363 * symbol-summary.h (symbol_summary::m_released): New member.
37364 Do not release a summary twice. Do not allow to call finalizer
37365 for types of a summary that live in GGC memory.
37367 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
37369 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
37372 2017-02-02 Martin Sebor <msebor@redhat.com>
37374 PR middle-end/79275
37375 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
37376 (format_string): Tighten up the range of output for non-constant
37377 strings and correct the expected range for wide non-constant strings.
37379 2017-02-02 Martin Sebor <msebor@redhat.com>
37381 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
37383 PR middle-end/32003
37384 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
37386 (-fdump-tree-@var): Add to index and document how to come up
37387 with pass-specific option and dump file names.
37388 (-fdump-passes): Clarify where to look for output.
37390 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
37392 PR middle-end/77445
37393 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
37394 statistics of the analyzed path; allow threading for speed when
37395 any of BBs along the path are optimized for speed.
37397 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
37399 PR middle-end/78468
37400 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
37401 settings of the virtual registers.
37404 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
37406 * explow.c (get_dynamic_stack_size): Take known alignment of stack
37407 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
37410 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37412 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
37413 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
37415 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37417 * config/s390/s390.md: Add missing comments with the expanded
37419 * config/s390/vector.md: Likewise.
37420 * config/s390/vx-builtins.md: Likewise.
37422 2017-02-02 Jakub Jelinek <jakub@redhat.com>
37425 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
37426 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
37427 conditions on a single line.
37429 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37431 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
37432 __S390_VX__ to __VX__.
37434 2017-02-01 Andrew Pinski <apinski@cavium.com>
37436 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
37437 stmt_info to record_stmt_cost.
37438 (vect_get_known_peeling_cost): Pass stmt_info if known to
37440 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
37441 cpu_vector_cost field into
37442 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
37443 field into vec_int_stmt_cost and vec_fp_stmt_cost.
37444 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
37445 splitting of scalar_stmt_cost and vec_stmt_cost.
37446 (thunderx_vector_cost): Likewise.
37447 (cortexa57_vector_cost): LIkewise.
37448 (exynosm1_vector_cost): Likewise.
37449 (xgene1_vector_cost): Likewise.
37450 (thunderx2t99_vector_cost): Improve after the splitting of the two
37452 (aarch64_builtin_vectorization_cost): Update for the splitting of
37453 scalar_stmt_cost and vec_stmt_cost.
37455 2017-02-01 Torvald Riegel <triegel@redhat.com>
37456 Richard Henderson <rth@redhat.com>
37458 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
37459 conditional on existance of a fast atomic load.
37460 * optabs-query.c (can_atomic_load_p): New function.
37461 * optabs-query.h (can_atomic_load_p): Declare it.
37462 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
37463 no fast atomic load is available for the particular size of access.
37464 (expand_atomic_compare_and_swap): Likewise.
37465 (expand_atomic_load): Likewise.
37466 (expand_atomic_store): Likewise.
37467 (expand_atomic_fetch_op): Likewise.
37468 * testsuite/lib/target-supports.exp
37469 (check_effective_target_sync_int_128): Remove x86 because it provides
37470 no fast atomic load.
37471 (check_effective_target_sync_int_128_runtime): Likewise.
37473 2017-02-01 Richard Biener <rguenther@suse.de>
37475 * graphite.c: Include tree-vectorizer.h for find_loop_location.
37476 (graphite_transform_loops): Provide opt-info for optimized nests.
37477 * tree-parloop.c (parallelize_loops): Provide opt-info for
37478 parallelized loops.
37480 2017-02-01 Richard Biener <rguenther@suse.de>
37482 PR middle-end/79315
37483 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
37484 was not set before.
37486 2017-02-01 Richard Biener <rguenther@suse.de>
37488 PR tree-optimization/71824
37489 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
37490 Verify the loops are valid in the merged SESE region.
37491 (scop_detection::can_represent_loop_1): Check analyzing the
37492 evolution of the number of iterations in the region succeeds.
37494 2017-01-31 Ian Lance Taylor <iant@golang.org>
37496 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
37497 REG_ARGS_SIZE note to 32-bit push insns and call insn.
37499 2017-01-31 David Malcolm <dmalcolm@redhat.com>
37501 PR preprocessor/79210
37502 * input.c (get_substring_ranges_for_loc): Replace line_width
37503 assertion with error-handling.
37505 2017-01-31 Richard Biener <rguenther@suse.de>
37507 PR tree-optimization/77318
37508 * graphite-sese-to-poly.c (extract_affine): Fix assert.
37509 (create_pw_aff_from_tree): Take loop parameter.
37510 (add_condition_to_pbb): Pass loop of the condition to
37511 create_pw_aff_from_tree.
37513 2017-01-31 Jakub Jelinek <jakub@redhat.com>
37515 * config/s390/s390.c (s390_asan_shadow_offset): New function.
37516 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
37518 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
37522 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
37524 (convert_int_to_float128): Likewise.
37525 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
37526 (convert_int_to_float128): Likewise.
37527 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
37528 (UNSPEC_IEEE128_CONVERT): Likewise.
37529 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
37530 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
37531 Use local variables for IBM extended format.
37532 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
37533 (fix_trunc<mode>si2_fprs): Likewise.
37534 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
37535 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
37536 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
37537 to know that we can now have integers of all sizes in vector
37539 (fix<uns>_<mode>di2_hw): Likewise.
37540 (float<uns>_<mode>si2_hw): Likewise.
37541 (fix_<mode>si2_hw): Likewise.
37542 (fixuns_<mode>si2_hw): Likewise.
37543 (float<uns>_<mode>di2_hw): Likewise.
37544 (float_<mode>di2_hw): Likewise.
37545 (float_<mode>si2_hw): Likewise.
37546 (floatuns_<mode>di2_hw): Likewise.
37547 (floatuns_<mode>si2_hw): Likewise.
37548 (xscvqp<su>wz_<mode>): Delete, no longer used.
37549 (xscvqp<su>dz_<mode>): Likewise.
37550 (xscv<su>dqp_<mode>): Likewise.
37551 (ieee128_mfvsrd_64bit): Likewise.
37552 (ieee128_mfvsrd_32bit): Likewise.
37553 (ieee128_mfvsrwz): Likewise.
37554 (ieee128_mtvsrw): Likewise.
37555 (ieee128_mtvsrd_64bit): Likewise.
37556 (ieee128_mtvsrd_32bit): Likewise.
37558 2017-01-31 Martin Liska <mliska@suse.cz>
37561 * ipa-prop.c (ipa_free_all_node_params): Call release method
37562 instead of ~sumbol_summary to not to trigger double times
37565 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
37567 PR tree-optimization/71691
37568 * bitmap.h (class auto_bitmap): New.
37569 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
37570 is_maybe_undefined instead of ssa_undefined_value_p.
37572 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37574 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
37575 __S390_ARCH_LEVEL__ to __ARCH__.
37577 2017-01-31 Jakub Jelinek <jakub@redhat.com>
37579 PR tree-optimization/79267
37580 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
37581 if should_remove_lhs_p is true.
37583 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
37586 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
37587 (add_alignment_attribute): New.
37588 (base_type_die): Add alignment attribute.
37589 (subrange_type_die): Likewise.
37590 (modified_type_die): Likewise.
37591 (gen_array_type_die): Likewise.
37592 (gen_descr_array_type_die: Likewise.
37593 (gen_enumeration_type_die): Likewise.
37594 (gen_subprogram_die): Likewise.
37595 (gen_variable_die): Likewise.
37596 (gen_field_die): Likewise.
37597 (gen_ptr_to_mbr_type_die): Likewise.
37598 (gen_struct_or_union_type_die): Likewise.
37599 (gen_subroutine_type_die): Likewise.
37600 (gen_typedef_die): Likewise.
37601 (base_type_cmp): Compare alignment attribute.
37603 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37606 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
37607 (setb_unsigned) New pattern for setb with CCUNS.
37608 * config/rs6000/rs6000.c (expand_block_compare): Use a different
37609 subfc./subfe sequence to avoid overflow problems. Generate a
37610 shorter sequence with cmpld/setb for power9.
37611 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
37612 for generating subfc. instruction.
37613 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
37614 now uses this instruction.
37616 2017-01-30 Ian Lance Taylor <iant@google.com>
37619 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
37620 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
37622 2017-01-30 Martin Sebor <msebor@redhat.com>
37624 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
37625 Move constant to the right of a relational operator.
37626 (get_mpfr_format_length, format_character, format_string): Ditto.
37627 (should_warn_p, maybe_warn): Same.
37629 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
37631 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
37634 * asan.c (get_translation_unit_decl): Remove function.
37635 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
37637 2017-01-30 Martin Liska <mliska@suse.cz>
37639 PR gcov-profile/79259
37640 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
37641 -fprofile-generate.
37643 2017-01-30 Martin Liska <mliska@suse.cz>
37646 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
37647 Initialize variables with NULL value.
37649 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
37652 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
37654 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
37656 2017-01-30 Richard Biener <rguenther@suse.de>
37658 PR tree-optimization/79276
37659 * tree-vrp.c (process_assert_insertions): Properly adjust common
37660 when removing a duplicate.
37662 2017-01-30 Richard Biener <rguenther@suse.de>
37664 PR tree-optimization/79256
37665 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
37666 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
37668 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
37670 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
37673 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
37674 ("*r<noxa>sbg_<mode>_sll_bitmask")
37675 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
37676 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
37677 Use contiguous_bitmask_nowrap_operand.
37679 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37682 * config/rs6000/altivec.h (vec_xl): Revise #define.
37683 (vec_xst): Likewise.
37685 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
37687 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
37689 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
37691 PR rtl-optimization/79194
37692 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
37693 traps before call to bypass_conditional_jumps.
37695 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
37697 PR tree-optimization/71374
37698 * lra-constraints.c (check_conflict_input_operands): New.
37699 (match_reload): Use it.
37701 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
37704 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
37705 account to calculate conflict_set.
37707 2017-01-27 Bin Cheng <bin.cheng@arm.com>
37709 PR rtl-optimization/78559
37710 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
37711 other_insn in combine.
37713 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
37715 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
37716 uint16_type_node for BT_UINT16.
37718 2017-01-27 David Malcolm <dmalcolm@redhat.com>
37720 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
37721 "RTL Tests" to menu.
37722 (GIMPLE Tests): New node.
37723 (RTL Tests): New node.
37725 2017-01-27 Richard Biener <rguenther@suse.de>
37727 PR tree-optimization/79245
37728 * tree-loop-distribution.c (distribute_loop): Apply cost
37729 modeling also to detected patterns.
37731 2017-01-27 Richard Biener <rguenther@suse.de>
37733 PR tree-optimization/71433
37734 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
37735 (compare_assert_loc): New function.
37736 (process_assert_insertions): Sort and optimize assert locations
37737 to remove duplicates and push down identical assertions on
37738 edges to their destination block.
37740 2017-01-27 Richard Biener <rguenther@suse.de>
37742 PR tree-optimization/79244
37743 * tree-vrp.c (remove_range_assertions): Forcefully propagate
37744 out SSA names even if abnormal.
37746 2017-01-27 Jakub Jelinek <jakub@redhat.com>
37748 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
37749 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
37750 instead of MPFR_RNDN.
37752 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
37755 * arm.c (arm_option_override): Don't call build_target_option_node
37756 until after doing all option overrides.
37757 (arm_valid_target_attribute_tree): Likewise.
37759 2017-01-27 Martin Liska <mliska@suse.cz>
37761 * doc/invoke.texi (-fprofile-arcs): Document profiling support
37762 for {cd}tors and C++ {cd}tors.
37764 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
37766 * config/s390/s390.md ("*setmem_long_and")
37767 ("*setmem_long_and_31z"): Use zero_extend instead of and.
37769 2017-01-26 Martin Sebor <msebor@redhat.com>
37771 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
37774 2017-01-26 Martin Sebor <msebor@redhat.com>
37776 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
37777 HAVE_DFmode before using XFmode or DFmode.
37778 (parse_directive): Avoid using the z length modifier to avoid
37779 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
37781 PR middle-end/78703
37782 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
37783 to accept adjustment as an array.
37784 (get_int_range): New function.
37785 (struct directive): Make width and prec arrays.
37786 (directive::set_width, directive::set_precision): Call get_int_range.
37787 (format_integer, format_floating): Handle width and precision ranges.
37788 (format_string, parse_directive): Same.
37790 2017-01-26 Jakub Jelinek <jakub@redhat.com>
37793 * dwarf2out.c (generate_skeleton_bottom_up): For children with
37794 comdat_type_p set, just clone them, but keep the children in the
37798 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
37799 which have direct callers with -fvar-tracking-assignments enabled
37801 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
37802 inside of type units.
37804 2017-01-26 Martin Sebor <msebor@redhat.com>
37806 PR middle-end/78703
37807 * gimple-ssa-sprintf.c (struct result_range): Add likely and
37809 (struct format_result): Replace number_chars, number_chars_min,
37810 and number_chars_max with a single member of struct result_range.
37812 (format_result::operator+=): Adjust.
37813 (struct fmtresult): Remove bounded. Handle likely and unlikely
37815 (fmtresult::adjust_for_width_or_precision): New function.
37816 (fmtresult:type_max_digits): New function.
37817 (bytes_remaining): Handle likely and unlikely counters.
37818 (min_bytes_remaining): Remove.
37819 (format_percent): Simplify.
37820 (format_integer, format_floating): Set likely and unlikely counters.
37821 (get_string_length, format_character, format_string): Same.
37822 (format_plain, should_warn_p): New function.
37823 (maybe_warn): Call should_warn_p. Update diagnostic messages
37824 and handle those for all directives, including plain strings.
37825 (format_directive): Handle likely and unlikely counters.
37826 Remove unnecessary quoting from diagnostics. Add an informational
37828 (add_bytes): Remove.
37829 (pass_sprintf_length::compute_format_length): Simplify.
37830 (try_substitute_return_value): Handle likely and unlikely counters.
37832 2017-01-26 Carl Love <cel@us.ibm.com>
37834 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
37835 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
37837 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
37840 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
37841 endianess for subregs into account.
37842 * lra-constraints.c (lra_constraints): Do risky transformations
37843 always on the first iteration.
37844 * lra-lives.c (check_pseudos_live_through_calls): Add arg
37845 last_call_used_reg_set.
37846 (process_bb_lives): Define and use last_call_used_reg_set.
37847 * lra.c (lra): Always continue after lra_constraints on the first
37850 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
37852 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
37854 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
37856 2017-01-26 Jakub Jelinek <jakub@redhat.com>
37858 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
37859 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
37860 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
37861 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
37862 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
37863 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
37864 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
37865 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
37866 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
37868 2017-01-26 Marek Polacek <polacek@redhat.com>
37871 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
37872 for the third operand.
37874 2017-01-26 Jakub Jelinek <jakub@redhat.com>
37876 PR middle-end/79236
37877 * omp-low.c (struct omp_context): Add simt_stmt field.
37878 (scan_omp_for): Return omp_context *.
37879 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
37880 context to the _simt_ SIMD stmt.
37881 (lower_omp_for): For combined SIMD with sibling _simt_
37882 SIMD, make sure to use the same decls in _looptemp_
37883 clauses as in the sibling.
37885 2017-01-26 David Sherwood <david.sherwood@arm.com>
37887 PR middle-end/79212
37888 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
37891 2017-01-26 Jakub Jelinek <jakub@redhat.com>
37894 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
37895 emit fld b; fld a; if possible.
37897 * brig-builtins.def: Update copyright years.
37898 * config/arm/arm_acle_builtins.def: Update copyright years.
37900 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
37903 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
37904 constraint instead of o for the stxsd instruction.
37906 2017-01-25 Carl Love <cel@us.ibm.com>
37908 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
37909 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
37911 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
37913 * doc/invoke.texi (C++ Dialect Options): Fix typo.
37915 2017-01-25 Richard Biener <rguenther@suse.de>
37917 PR tree-optimization/69264
37918 * target.def (vector_alignment_reachable): Improve documentation.
37919 * doc/tm.texi: Regenerate.
37920 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
37922 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
37923 earlier changes with respect to TYPE_USER_ALIGN.
37924 (vector_alignment_reachable_p): Likewise. Improve dumping.
37926 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37929 * config/arm/arm.md (xordi3): Force constant operand into a register
37932 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37934 * doc/invoke.texi (-fstore-merging): Correct default optimization
37935 levels at which it is enabled.
37936 (-O): Move -fstore-merging from list to...
37939 2017-01-25 Richard Biener <rguenther@suse.de>
37942 * omp-expand.c: Include debug.h.
37943 (expand_omp_taskreg): Make sure to generate early debug before
37944 outlining anything from a function.
37945 (expand_omp_target): Likewise.
37946 (grid_expand_target_grid_body): Likewise.
37948 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
37951 * asan.c (get_translation_unit_decl): New function.
37952 (asan_add_global): Extract modules file name from globals
37953 TRANSLATION_UNIT_DECL name.
37955 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
37958 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
37959 for long calls with APCS frame and VFP.
37961 2017-01-24 David Malcolm <dmalcolm@redhat.com>
37963 * cfg.c (original_copy_tables_initialized_p): New function.
37964 * cfg.h (original_copy_tables_initialized_p): New decl.
37965 * cfgrtl.c (relink_block_chain): Guard the call to
37966 free_original_copy_tables with a call to
37967 original_copy_tables_initialized_p.
37968 * cgraph.h (symtab_node::native_rtl_p): New decl.
37969 * cgraphunit.c (symtab_node::native_rtl_p): New function.
37970 (symtab_node::needed_p): Don't assert for early assembly output
37971 for __RTL functions.
37972 (cgraph_node::finalize_function): Set "force_output" for __RTL
37974 (cgraph_node::analyze): Bail out early for __RTL functions.
37975 (analyze_functions): Update assertion to support __RTL functions.
37976 (cgraph_node::expand): Bail out early for __RTL functions.
37977 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
37979 * function.h (struct function): Update comment for field
37981 * gimple-expr.c: Include "tree-pass.h".
37982 (gimple_has_body_p): Return false for __RTL functions.
37983 * Makefile.in (OBJS): Add run-rtl-passes.o.
37984 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
37986 (gcc::pass_manager::get_clean_slate): New accessor.
37987 * passes.c: Include "insn-addr.h".
37988 (should_skip_pass_p): Add logging. Update logic for running
37989 "expand" to be compatible with both __GIMPLE and __RTL. Guard
37990 property-provider override so it is only done for gimple passes.
37992 (skip_pass): New function.
37993 (execute_one_pass): Call skip_pass when skipping passes.
37994 * read-md.c (md_reader::read_char): Support filtering
37995 the input to a subset of line numbers.
37996 (md_reader::md_reader): Initialize fields
37997 m_first_line and m_last_line.
37998 (md_reader::read_file_fragment): New function.
37999 * read-md.h (md_reader::read_file_fragment): New decl.
38000 (md_reader::m_first_line): New field.
38001 (md_reader::m_last_line): New field.
38002 * read-rtl-function.c (function_reader::create_function): Only
38003 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
38004 curr_properties. Set DECL_INITIAL to a dummy block.
38005 (read_rtl_function_body_from_file_range): New function.
38006 * read-rtl-function.h (read_rtl_function_body_from_file_range):
38008 * run-rtl-passes.c: New file.
38009 * run-rtl-passes.h: New file.
38011 2017-01-24 Jeff Law <law@redhat.com>
38013 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
38016 2017-01-24 Bin Cheng <bin.cheng@arm.com>
38018 PR tree-optimization/79159
38019 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
38020 (record_nonwrapping_iv): Improve boundary using above function if no
38021 value range information.
38023 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
38024 Martin Jambor <mjambor@suse.cz>
38026 * brig-builtins.def: New file.
38027 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
38028 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
38029 (DEF_HSAIL_SAT_BUILTIN): Likewise.
38030 (DEF_HSAIL_INTR_BUILTIN): Likewise.
38031 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
38032 * builtin-types.def (BT_INT8): New.
38033 (BT_INT16): Likewise.
38034 (BT_UINT8): Likewise.
38035 (BT_UINT16): Likewise.
38036 (BT_FN_ULONG): Likewise.
38037 (BT_FN_UINT_INT): Likewise.
38038 (BT_FN_UINT_ULONG): Likewise.
38039 (BT_FN_UINT_LONG): Likewise.
38040 (BT_FN_UINT_PTR): Likewise.
38041 (BT_FN_ULONG_PTR): Likewise.
38042 (BT_FN_INT8_FLOAT): Likewise.
38043 (BT_FN_INT16_FLOAT): Likewise.
38044 (BT_FN_UINT32_FLOAT): Likewise.
38045 (BT_FN_UINT16_FLOAT): Likewise.
38046 (BT_FN_UINT8_FLOAT): Likewise.
38047 (BT_FN_UINT64_FLOAT): Likewise.
38048 (BT_FN_UINT16_UINT32): Likewise.
38049 (BT_FN_UINT32_UINT16): Likewise.
38050 (BT_FN_UINT16_UINT16_UINT16): Likewise.
38051 (BT_FN_INT_PTR_INT): Likewise.
38052 (BT_FN_UINT_PTR_UINT): Likewise.
38053 (BT_FN_LONG_PTR_LONG): Likewise.
38054 (BT_FN_ULONG_PTR_ULONG): Likewise.
38055 (BT_FN_VOID_UINT64_UINT64): Likewise.
38056 (BT_FN_UINT8_UINT8_UINT8): Likewise.
38057 (BT_FN_INT8_INT8_INT8): Likewise.
38058 (BT_FN_INT16_INT16_INT16): Likewise.
38059 (BT_FN_INT_INT_INT): Likewise.
38060 (BT_FN_UINT_FLOAT_UINT): Likewise.
38061 (BT_FN_FLOAT_UINT_UINT): Likewise.
38062 (BT_FN_ULONG_UINT_UINT): Likewise.
38063 (BT_FN_ULONG_UINT_PTR): Likewise.
38064 (BT_FN_ULONG_ULONG_ULONG): Likewise.
38065 (BT_FN_UINT_UINT_UINT): Likewise.
38066 (BT_FN_VOID_UINT_PTR): Likewise.
38067 (BT_FN_UINT_UINT_PTR: Likewise.
38068 (BT_FN_UINT32_UINT64_PTR): Likewise.
38069 (BT_FN_INT_INT_UINT_UINT): Likewise.
38070 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
38071 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
38072 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
38073 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
38074 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
38075 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
38076 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
38077 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
38078 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
38079 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
38080 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
38081 * doc/frontends.texi: List BRIG FE.
38082 * doc/install.texi (Testing): Add BRIG tesring requirements.
38083 * doc/invoke.texi (Overall Options): Mention BRIG.
38084 * doc/standards.texi (Standards): Doucment BRIG HSA version.
38086 2017-01-24 Richard Biener <rguenther@suse.de>
38088 PR translation/79208
38089 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
38091 2017-01-24 Martin Jambor <mjambor@suse.cz>
38094 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
38095 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
38096 and known_contexts.
38098 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
38100 PR middle-end/79123
38101 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
38102 casts from signed to unsigned really don't have a range.
38104 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
38106 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
38107 GMP_RNDx for compatiblity.
38109 2017-01-24 Martin Liska <mliska@suse.cz>
38112 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
38113 that would prevent us to call alloca with -1 as argument.
38115 2017-01-24 Jakub Jelinek <jakub@redhat.com>
38117 * dwarf2out.c (output_compilation_unit_header, output_file_names):
38118 Avoid -Wformat-security warning.
38120 2017-01-23 Andrew Pinski <apinski@cavium.com>
38122 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
38125 2017-01-23 Martin Sebor <msebor@redhat.com>
38127 PR middle-end/78703
38128 * gimple-ssa-sprintf.c (warn_level): New global.
38129 (format_integer): Use it here and throughout the rest of the file.
38130 Use the same switch to compute sign as base.
38131 (maybe_warn): New function.
38132 (format_directive): Factor out warnings into maybe_warn.
38133 Add debugging output. Use warn_level.
38134 (add_bytes): Use warn_level.
38135 (pass_sprintf_length::compute_format_length): Add debugging output.
38136 (try_substitute_return_value): Same.
38137 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
38139 PR middle-end/78703
38140 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
38141 (struct fmtresult, format_integer, format_floating): Adjust.
38142 (fmtresult::fmtresult): Set max correctly in two argument ctor.
38143 (get_string_length, format_string,format_directive): Same.
38144 (pass_sprintf_length::compute_format_length): Same.
38145 (try_substitute_return_value): Simplify slightly.
38147 PR middle-end/78703
38148 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
38149 (fmtresult::operator+=): Outlined.
38150 (struct fmtresult): Add ctors.
38151 (struct conversion_spec): Rename...
38152 (struct directive): ...to this. Add and remove data members.
38153 (directive::set_width, directive::set_precision): New functions.
38154 (format_percent): Use fmtresult ctor.
38155 (get_width_and_precision): Remove.
38156 (format_integer): Make naming changes. Avoid computing width and
38158 (format_floating): Same. Adjust indentation.
38159 (format_character, format_none): New functions.
38160 (format_string): Moved character handling to format_character.
38161 (format_directive): Remove arguments, change return type.
38162 (parse_directive): New function.
38163 (pass_sprintf_length::compute_format_length): Move directive
38164 parsing to parse_directive.
38166 2017-01-23 Jakub Jelinek <jakub@redhat.com>
38168 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
38169 (assign_assembler_name_if_needed): ... this.
38170 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
38171 (assign_assembler_name_if_needed): ... this.
38172 (free_lang_data_in_cgraph): Adjust callers.
38173 * cgraphunit.c (cgraph_node::analyze): Likewise.
38174 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
38177 2017-01-23 Richard Biener <rguenther@suse.de>
38179 PR tree-optimization/79088
38180 PR tree-optimization/79188
38181 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
38182 resetting loop bounds after last path deletion. Reset loop
38183 bounds of the target loop, make code match the comments.
38184 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
38185 Make sure loops need no fixups.
38187 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
38189 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
38190 exponent support with double type for first argument.
38191 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
38192 type returned by __builtin_vec_extract_sig,
38193 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
38194 functions from "vector int" to "vector unsigned int" or from
38195 "vector long long int" to "vector unsigned long long int".
38196 Changed type returned by __builtin_vec_extract_exp,
38197 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
38198 functions from "vector int" to "vector unsigned int" or from
38199 "vector long long int" to "vector unsigned long long int".
38200 Changed return type of __builtin_vec_test_data_class,
38201 __builtin_vec_test_data_class_sp, and
38202 __builtin_vec_test_data_class_dp from "vector int" to
38203 "vector bool int" or from "vector long long int" to "vector bool
38204 long long int" and changed second argument type from "unsigned
38205 int" to "int". Added new overloaded function forms "vector float
38206 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
38207 "vector float __builtin_vec_insert_exp_sp (vector float, vector
38208 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
38209 double, vector unsigned long long int)" and "vector double
38210 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
38211 long int)". Changed return type of
38212 __builtin_scalar_test_data_class and
38213 __builtin_scalar_test_data_class_sp and
38214 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
38215 int" and changed second argument from "unsigned int" to "int".
38216 Changed type returned by __builtin_scalar_test_neg,
38217 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
38218 from "int" to "bool int". Added new overloaded function form
38219 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
38220 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
38221 exponent double-precision with floating point first argument.
38222 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
38223 documentation of scalar_test_data_class, scalar_test_neg,
38224 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
38225 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
38226 vec_test_data_class built-in functions to reflect refinements in
38227 their type signatures.
38229 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
38231 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
38233 (aarch64_elf_asm_destructor): Likewise.
38235 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
38237 PR rtl-optimization/78634
38238 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
38239 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
38240 * ifcvt.c (noce_try_cmove): Add missing cost check.
38242 PR rtl-optimization/71724
38243 * combine.c (if_then_else_cond): Look for situations where it is
38244 beneficial to undo the work of one of the recursive calls.
38246 2017-01-23 Bin Cheng <bin.cheng@arm.com>
38248 PR tree-optimization/70754
38249 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
38250 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
38251 combined stmt before it if not NULL.
38252 (combine_chains): Process refs reversely and compute dominance point
38255 2017-01-23 Martin Liska <mliska@suse.cz>
38257 PR tree-optimization/79196
38258 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
38259 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
38261 (strlen_optimize_stmt): Call the renamed function.
38263 2017-01-23 Michael Matz <matz@suse.de>
38265 PR tree-optimization/78384
38266 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
38268 2017-01-23 Richard Biener <rguenther@suse.de>
38270 PR tree-optimization/79186
38271 * tree-vrp.c (register_new_assert_for): Make sure we've seen
38272 both incoming edges before moving an assert.
38274 2017-01-23 Martin Jambor <mjambor@suse.cz>
38276 * ipa-prop.c (load_from_param_1): Removed.
38277 (load_from_unmodified_param): Bits from load_from_param_1 put back
38279 (load_from_param): Removed.
38280 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
38281 with stmt. Reverted back to use of load_from_unmodified_param.
38283 2017-01-23 Martin Jambor <mjambor@suse.cz>
38286 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
38287 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
38288 field a pointer to garbage collected vector, mark lattices and
38289 ipcp_orig_node with GTY((skip)).
38290 (ipa_get_param_count): Adjust to descriptors being a pointer.
38291 (ipa_get_param): Likewise.
38292 (ipa_get_type): Likewise.
38293 (ipa_get_param_move_cost): Likewise.
38294 (ipa_set_param_used): Likewise.
38295 (ipa_get_controlled_uses): Likewise.
38296 (ipa_set_controlled_uses): Likewise.
38297 (ipa_is_param_used): Likewise.
38298 (ipa_node_params_t): Move into garbage collector. New methods insert
38300 (ipa_node_params_sum): Annotate wth GTY(()).
38301 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
38303 (ipa_load_from_parm_agg): Adjust declaration.
38304 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
38305 * ipa-profile.c (ipa_profile): Likewise.
38306 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
38307 (ipa_populate_param_decls): Make descriptors parameter garbage
38309 (ipa_dump_param): Adjust to descriptors being a pointer.
38310 (ipa_alloc_node_params): Likewise.
38311 (ipa_initialize_node_params): Likewise.
38312 (load_from_param_1): Make descriptors parameter garbage collected.
38313 (load_from_unmodified_param): Likewise.
38314 (load_from_param): Likewise.
38315 (ipa_load_from_parm_agg): Likewise.
38316 (ipa_node_params::~ipa_node_params): Removed.
38317 (ipa_free_all_node_params): Remove call to delete operator.
38318 (ipa_node_params_t::insert): New.
38319 (ipa_node_params_t::remove): Likewise.
38320 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
38321 copy known_csts and known_contexts vectors.
38322 (ipa_read_node_info): Adjust to descriptors being a pointer.
38323 (ipcp_modif_dom_walker): Make m_descriptors field garbage
38325 (ipcp_transform_function): Make descriptors variable garbage
38328 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
38330 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
38331 * config/i386/avx512dqintrin.h: Ditto.
38332 * config/i386/avx512fintrin.h: Ditto.
38333 * config/i386/i386.c: Handle new builtins.
38334 * config/i386/i386-builtin.def: Add new builtins.
38335 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
38336 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
38338 2017-01-23 Jakub Jelinek <jakub@redhat.com>
38339 Martin Liska <mliska@suse.cz>
38341 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
38342 * asan.c (asan_expand_poison_ifn): Support stores and use
38343 appropriate ASAN report function.
38344 * internal-fn.c (expand_ASAN_POISON_USE): New function.
38345 * internal-fn.def (ASAN_POISON_USE): Declare.
38346 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
38347 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
38348 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
38349 ASAN_POISON calls w/o LHS.
38350 * tree-ssa.c (execute_update_addresses_taken): Create clobber
38351 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
38352 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
38353 * gimplify.c (asan_poison_variables): Add attribute
38354 use_after_scope_memory to variables that really needs to live
38356 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
38357 having the attribute.
38359 2017-01-23 Martin Liska <mliska@suse.cz>
38361 * asan.c (create_asan_shadow_var): New function.
38362 (asan_expand_poison_ifn): Likewise.
38363 * asan.h (asan_expand_poison_ifn): New declaration.
38364 * internal-fn.c (expand_ASAN_POISON): Likewise.
38365 * internal-fn.def (ASAN_POISON): New builtin.
38366 * sanopt.c (pass_sanopt::execute): Expand
38367 asan_expand_poison_ifn.
38368 * tree-inline.c (copy_decl_for_dup_finish): Make function
38370 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
38371 * tree-ssa.c (is_asan_mark_p): New function.
38372 (execute_update_addresses_taken): Rewrite local variables
38373 (identified just by use-after-scope as addressable) into SSA.
38375 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
38377 * doc/install.texi (Specific): opensource.apple.com uses https
38378 now. Remove trailing slash.
38380 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
38382 * README.Portability: Remove note on an Irix compatibility issue.
38384 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
38386 * gcov.c (INCLUDE_ALGORITHM): Define.
38387 (INCLUDE_VECTOR): Define.
38388 No longer include <vector> and <algorithm> directly.
38390 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
38392 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
38394 * doc/invoke.texi (Code Gen Options): Ditto.
38396 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
38399 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
38401 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
38403 rtl-optimization/79125
38404 * cprop.c (local_cprop_pass): Handle cases where we make an
38405 unconditional trap.
38407 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
38411 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
38412 read from, for big endian.
38414 2017-01-20 Jiong Wang <jiong.wang@arm.com>
38416 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
38417 register pauth builtins for LP64 only.
38419 2017-01-20 Marek Polacek <polacek@redhat.com>
38422 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
38425 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
38427 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
38429 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
38430 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
38431 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
38433 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38436 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
38437 in big-endian mode when they are not a single duplicated value.
38439 2017-01-20 Richard Biener <rguenther@suse.de>
38441 * BASE-VER: Bump to 7.0.1.
38443 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
38445 * omp-low.c (omplow_simd_context): New struct. Use it...
38446 (lower_rec_simd_input_clauses): ...here and...
38447 (lower_rec_input_clauses): ...here to hold common data. Adjust all
38448 references to idx, lane, max_vf, is_simt.
38450 2017-01-20 Graham Markall <graham.markall@embecosm.com>
38452 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
38455 2017-01-20 Martin Jambor <mjambor@suse.cz>
38457 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
38458 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
38460 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
38461 (GTFILES): Rename hsa.c to hsa-common.c.
38462 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
38463 * hsa-dump.c: Likewise.
38464 * hsa-gen.c: Likewise.
38465 * hsa-regalloc.c: Likewise.
38466 * ipa-hsa.c: Likewise.
38467 * omp-expand.c: Likewise.
38468 * omp-low.c: Likewise.
38469 * toplev.c: Likewise.
38471 2017-01-20 Marek Polacek <polacek@redhat.com>
38474 * doc/invoke.texi: Document -Wduplicated-branches.
38475 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
38476 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
38477 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
38478 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
38479 return 0 only when not OEP_LEXICOGRAPHIC.
38480 (fold_build_cleanup_point_expr): Use the expression
38481 location when building CLEANUP_POINT_EXPR.
38482 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
38483 * tree.c (add_expr): Handle error_mark_node.
38485 2017-01-20 Martin Liska <mliska@suse.cz>
38488 * tree-profile.c (init_ic_make_global_vars): Do not call
38490 (gimple_init_gcov_profiler): Likewise.
38492 2017-01-20 Martin Liska <mliska@suse.cz>
38495 * cgraph.h (maybe_create_reference): Remove argument and
38497 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
38499 * ipa-cp.c (create_specialized_node): Likewise.
38500 * symtab.c (symtab_node::maybe_create_reference): Handle
38501 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
38503 2017-01-20 Martin Liska <mliska@suse.cz>
38505 * read-rtl-function.c (function_reader::create_function): Use
38506 build_decl instread of build_decl_stat.
38508 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
38510 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
38511 * config/i386/avx512dqintrin.h: Ditto.
38512 * config/i386/avx512fintrin.h: Ditto.
38513 * config/i386/i386-builtin-types.def: Add new types.
38514 * config/i386/i386.c: Handle new types.
38515 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
38516 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
38517 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
38518 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
38519 (__builtin_ia32_kshiftridi): New.
38520 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
38522 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
38526 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
38527 define to rs6000_init_stack_protect_guard.
38528 (rs6000_init_stack_protect_guard): New function.
38530 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
38531 Yunqiang Su <yunqiang.su@imgtec.com>
38533 * config.gcc (supported_defaults): Add madd4.
38534 (with_madd4): Add validation.
38535 (all_defaults): Add madd4.
38536 * config/mips/mips.opt (mmadd4): New option.
38537 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
38539 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
38541 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
38542 (ISA_HAS_FUSED_MADD4): Likewise.
38543 * doc/invoke.texi (-mmadd4): Document the new option.
38544 * doc/install.texi (--with-madd4): Document the new option.
38546 2017-01-19 Jiong Wang <jiong.wang@arm.com>
38548 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
38549 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
38550 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
38551 (aarch64_init_pauth_hint_builtins): New.
38552 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
38553 (aarch64_expand_builtin): Expand new builtins.
38555 2017-01-19 Jiong Wang <jiong.wang@arm.com>
38557 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
38558 * combine-stack-adj.c (no_unhandled_cfa): Handle
38559 REG_CFA_TOGGLE_RA_MANGLE.
38560 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
38561 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
38562 info for return address signing.
38563 (aarch64_expand_epilogue): Likewise.
38565 2017-01-19 Jiong Wang <jiong.wang@arm.com>
38567 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
38568 * config/aarch64/aarch64-protos.h
38569 (aarch64_return_address_signing_enabled): New declaration.
38570 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
38572 (aarch64_expand_prologue): Sign return address before it's pushed onto
38574 (aarch64_expand_epilogue): Authenticate return address fetched from
38576 (aarch64_override_options): Sanity check for ILP32 and ISA level.
38577 (aarch64_attributes): New function attributes for "sign-return-address".
38578 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
38579 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
38580 ("*do_return"): Generate combined instructions according to key index.
38581 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
38582 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
38584 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
38585 * config/aarch64/aarch64.opt (msign-return-address=): New.
38586 * doc/extend.texi (AArch64 Function Attributes): Documents
38587 "sign-return-address=".
38588 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
38590 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
38592 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
38593 overall option summary.
38595 2017-01-19 Jiong Wang <jiong.wang@arm.com>
38597 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
38598 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
38599 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
38600 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
38602 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
38604 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
38605 -mpower9-minmax by default for -mcpu=power9.
38606 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
38607 128-bit floating point.
38609 2017-01-20 Alan Modra <amodra@gmail.com>
38611 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
38612 optimizing for size.
38614 2017-01-20 Alan Modra <amodra@gmail.com>
38617 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
38618 for strcmp and strncmp from corresponding builtin decl.
38620 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
38622 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
38623 instead of i386/rtems-64.h.
38624 * config/i386/rtems-64.h: Remove.
38626 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
38630 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
38632 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
38634 2017-01-19 Tamar Christina <tamar.christina@arm.com>
38636 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
38637 Change int to HOST_WIDE_INT.
38638 * config/aarch64/aarch64-protos.h
38639 (aarch64_simd_gen_const_vector_dup): Likewise.
38640 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
38642 2017-01-19 David Malcolm <dmalcolm@redhat.com>
38644 * langhooks-def.h (lhd_type_for_size): New decl.
38645 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
38646 * langhooks.c (lhd_type_for_size): New function, taken from
38649 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
38651 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
38652 define_bypass for CR latency.
38653 (power9-cracked-alu): Update bypass latency and remove power9-branch.
38654 (power9-alu2): Add define_bypass for CR latency.
38656 (power9-mul): Update insn latency.
38657 (power9-mul-compare): Update insn latency, bypass latency and remove
38660 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38662 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
38664 * config/aarch64/aarch64.md
38665 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
38666 aarch64_nopcrelative_literal_loads.
38667 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
38669 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
38671 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
38672 TARGET_LOONGSON_3A.
38673 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
38675 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
38678 * config.gcc (supported_defaults): Add lxc1-sxc1.
38679 (with_lxc1_sxc1): Add validation.
38680 (all_defaults): Add lxc1-sxc1.
38681 * config/mips/mips.opt (mlxc1-sxc1): New option.
38682 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
38684 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
38685 __mips_no_lxc1_sxc1.
38686 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
38687 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
38688 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
38690 2017-01-19 Richard Biener <rguenther@suse.de>
38692 PR tree-optimization/72488
38693 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
38694 sure to restore SSA info.
38695 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
38697 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
38699 PR rtl-optimization/79121
38700 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
38701 of the inner type when shifting an extended value.
38703 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
38706 * symtab.c (symtab_node::equal_address_to): Fix comparing of
38707 interposable aliases.
38709 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
38712 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
38713 Use the evmergelohi instruction.
38714 (mov_si<mode>_e500_subreg4_2_le): Likewise.
38715 (mov_sitf_e500_subreg8_2_be): Likewise.
38716 (mov_sitf_e500_subreg12_2_le): Likewise.
38717 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
38718 (mov_si<mode>_e500_subreg4_2_be): Likewise.
38719 (mov_sitf_e500_subreg8_2_le): Likewise.
38720 (mov_sitf_e500_subreg12_2_be): Likewise.
38722 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38724 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
38725 attribute from vecsimple to vecperm.
38726 (altivec_vbpermq2): Likewise.
38728 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38731 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
38733 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38734 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
38735 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
38736 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
38737 case where N arg is SIZE_MAX.
38738 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
38739 (cmpstrsi): Add pattern.
38741 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
38743 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
38744 __builtin_vec_revb builtins.
38745 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
38746 built-in functions to support generation of the ISA 3.0 XXBR<x>
38747 vector byte reverse instructions.
38748 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
38749 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
38750 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
38751 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
38752 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
38753 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
38754 (P9V_BUILTIN_VEC_REVB): Likewise.
38755 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
38756 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
38757 (p9_xxbrq_v16qi): Likewise.
38758 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
38759 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
38760 (p9_xxbrh_v8hi): Likewise.
38761 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
38762 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
38763 vec_revb built-in functions.
38765 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
38767 PR rtl-optimization/78952
38768 * config/i386/i386.md (any_extract): New code iterator.
38769 (*insvqi_2): Use any_extract for source operand.
38770 (*insvqi_3): Use any_shiftrt for source operand.
38772 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
38774 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
38776 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
38778 2017-01-18 Matthias Klose <doko@ubuntu.com>
38780 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
38782 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38784 * config/rs6000/altivec.h (vec_bperm): Change #define.
38785 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
38786 (altivec_vbpermq2): New define_insn.
38787 (altivec_vbpermd): Likewise.
38788 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
38789 function interface.
38790 (VBPERMD): Likewise.
38791 (VBPERM): New polymorphic function interface.
38792 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
38793 Add entries for P9V_BUILTIN_VEC_VBPERM.
38794 * doc/extend.texi: Add interfaces for vec_bperm.
38796 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38798 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
38799 first letter of error messages.
38800 (s390_resolve_overloaded_builtin): Likewise.
38801 * config/s390/s390.c (s390_expand_builtin): Likewise.
38802 (s390_invalid_arg_for_unprototyped_fn): Likewise.
38803 (s390_valid_target_attribute_inner_p): Likewise.
38804 * config/s390/s390.md ("tabort"): Likewise.
38806 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
38808 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
38809 (ISA_AVOID_DIV_HILO): New macro.
38810 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
38811 (ISA_HAS_DDIV): Likewise.
38813 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
38815 * doc/invoke.texi (fabi-version): Correct number of occurrences.
38817 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
38819 * doc/invoke.texi (fabi-version): Spelling fix.
38821 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
38824 * doc/invoke.texi (fabi-version): Mention mangling fix for
38827 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
38830 * doc/invoke.texi (fabi-version): Document discriminator mangling.
38832 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
38835 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
38836 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
38838 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
38840 (stack_protect_test): Ditto.
38841 * config/rs6000/rs6000.opt (mstack-protector-guard=,
38842 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
38844 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
38845 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
38846 -mstack-protector-guard-offset=.
38847 (RS/6000 and PowerPC Options): Ditto.
38849 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
38851 * config/i386/i386.h (MASK_CLASS_P): New define.
38852 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
38853 there are no registers from different register sets also when
38854 mask registers are used. Update function comment.
38855 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
38856 to (*k/*r) and (*k/*km) alternatives.
38858 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
38860 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
38861 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
38862 (EH_RETURN_HANDLER_RTX): New define.
38863 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
38864 Force frame pointer in EH return functions.
38865 (aarch64_expand_epilogue): Add barrier for eh_return.
38866 (aarch64_final_eh_return_addr): Remove.
38867 (aarch64_eh_return_handler_rtx): New function.
38868 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
38870 (aarch64_eh_return_handler_rtx): New prototype.
38872 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38874 * config/rs6000/altivec.h (vec_rlmi): New #define.
38875 (vec_vrlnm): Likewise.
38876 (vec_rlnm): Likewise.
38877 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
38878 (UNSPEC_VRLNM): Likewise.
38879 (VIlong): New mode iterator.
38880 (altivec_vrl<VI_char>mi): New define_insn.
38881 (altivec_vrl<VI_char>nm): Likewise.
38882 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
38884 (VRLDNM): Likewise.
38885 (RLNM): New polymorphic function entry.
38886 (VRLWMI): New monomorphic function entry.
38887 (VRLDMI): Likewise.
38888 (RLMI): New polymorphic function entry.
38889 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
38890 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
38891 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
38894 2017-01-17 Jakub Jelinek <jakub@redhat.com>
38897 * dwarf2out.c (field_byte_offset): Restore the
38898 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
38899 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
38900 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
38903 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
38906 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
38908 2017-01-17 Jakub Jelinek <jakub@redhat.com>
38911 * dwarf2out.c (add_data_member_location_attribute): For constant
38912 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
38913 instead of DW_AT_data_member_location, DW_AT_bit_offset and
38914 DW_AT_byte_size attributes.
38916 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
38918 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
38919 after forcing to constant memory when the code model is medium.
38921 2017-01-17 Julia Koval <julia.koval@intel.com>
38924 * config/i386/avx512fintrin.h
38925 (_mm512_i32gather_ps): Change __addr type to void const*.
38926 (_mm512_mask_i32gather_ps): Ditto.
38927 (_mm512_i32gather_pd): Ditto.
38928 (_mm512_mask_i32gather_pd): Ditto.
38929 (_mm512_i64gather_ps): Ditto.
38930 (_mm512_mask_i64gather_ps): Ditto.
38931 (_mm512_i64gather_pd): Ditto.
38932 (_mm512_mask_i64gather_pd): Ditto.
38933 (_mm512_i32gather_epi32): Ditto.
38934 (_mm512_mask_i32gather_epi32): Ditto.
38935 (_mm512_i32gather_epi64): Ditto.
38936 (_mm512_mask_i32gather_epi64): Ditto.
38937 (_mm512_i64gather_epi32): Ditto.
38938 (_mm512_mask_i64gather_epi32): Ditto.
38939 (_mm512_i64gather_epi64): Ditto.
38940 (_mm512_mask_i64gather_epi64): Ditto.
38941 (_mm512_i32scatter_ps): Change __addr type to void*.
38942 (_mm512_mask_i32scatter_ps): Ditto.
38943 (_mm512_i32scatter_pd): Ditto.
38944 (_mm512_mask_i32scatter_pd): Ditto.
38945 (_mm512_i64scatter_ps): Ditto.
38946 (_mm512_mask_i64scatter_ps): Ditto.
38947 (_mm512_i64scatter_pd): Ditto.
38948 (_mm512_mask_i64scatter_pd): Ditto.
38949 (_mm512_i32scatter_epi32): Ditto.
38950 (_mm512_mask_i32scatter_epi32): Ditto.
38951 (_mm512_i32scatter_epi64): Ditto.
38952 (_mm512_mask_i32scatter_epi64): Ditto.
38953 (_mm512_i64scatter_epi32): Ditto.
38954 (_mm512_mask_i64scatter_epi32): Ditto.
38955 (_mm512_i64scatter_epi64): Ditto.
38956 (_mm512_mask_i64scatter_epi64): Ditto.
38957 * config/i386/avx512pfintrin.h
38958 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
38959 (_mm512_mask_prefetch_i32gather_ps): Ditto.
38960 (_mm512_mask_prefetch_i64gather_pd): Ditto.
38961 (_mm512_mask_prefetch_i64gather_ps): Ditto.
38962 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
38963 (_mm512_prefetch_i32scatter_ps): Ditto.
38964 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
38965 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
38966 (_mm512_prefetch_i64scatter_pd): Ditto.
38967 (_mm512_prefetch_i64scatter_ps): Ditto.
38968 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
38969 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
38970 * config/i386/avx512vlintrin.h
38971 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
38972 (_mm_mmask_i32gather_ps): Ditto.
38973 (_mm256_mmask_i32gather_pd): Ditto.
38974 (_mm_mmask_i32gather_pd): Ditto.
38975 (_mm256_mmask_i64gather_ps): Ditto.
38976 (_mm_mmask_i64gather_ps): Ditto.
38977 (_mm256_mmask_i64gather_pd): Ditto.
38978 (_mm_mmask_i64gather_pd): Ditto.
38979 (_mm256_mmask_i32gather_epi32): Ditto.
38980 (_mm_mmask_i32gather_epi32): Ditto.
38981 (_mm256_mmask_i32gather_epi64): Ditto.
38982 (_mm_mmask_i32gather_epi64): Ditto.
38983 (_mm256_mmask_i64gather_epi32): Ditto.
38984 (_mm_mmask_i64gather_epi32): Ditto.
38985 (_mm256_mmask_i64gather_epi64): Ditto.
38986 (_mm_mmask_i64gather_epi64): Ditto.
38987 (_mm256_i32scatter_ps): Change __addr type to void*.
38988 (_mm256_mask_i32scatter_ps): Ditto.
38989 (_mm_i32scatter_ps): Ditto.
38990 (_mm_mask_i32scatter_ps): Ditto.
38991 (_mm256_i32scatter_pd): Ditto.
38992 (_mm256_mask_i32scatter_pd): Ditto.
38993 (_mm_i32scatter_pd): Ditto.
38994 (_mm_mask_i32scatter_pd): Ditto.
38995 (_mm256_i64scatter_ps): Ditto.
38996 (_mm256_mask_i64scatter_ps): Ditto.
38997 (_mm_i64scatter_ps): Ditto.
38998 (_mm_mask_i64scatter_ps): Ditto.
38999 (_mm256_i64scatter_pd): Ditto.
39000 (_mm256_mask_i64scatter_pd): Ditto.
39001 (_mm_i64scatter_pd): Ditto.
39002 (_mm_mask_i64scatter_pd): Ditto.
39003 (_mm256_i32scatter_epi32): Ditto.
39004 (_mm256_mask_i32scatter_epi32): Ditto.
39005 (_mm_i32scatter_epi32): Ditto.
39006 (_mm_mask_i32scatter_epi32): Ditto.
39007 (_mm256_i32scatter_epi64): Ditto.
39008 (_mm256_mask_i32scatter_epi64): Ditto.
39009 (_mm_i32scatter_epi64): Ditto.
39010 (_mm_mask_i32scatter_epi64): Ditto.
39011 (_mm256_i64scatter_epi32): Ditto.
39012 (_mm256_mask_i64scatter_epi32): Ditto.
39013 (_mm_i64scatter_epi32): Ditto.
39014 (_mm_mask_i64scatter_epi32): Ditto.
39015 (_mm256_i64scatter_epi64): Ditto.
39016 (_mm256_mask_i64scatter_epi64): Ditto.
39017 (_mm_i64scatter_epi64): Ditto.
39018 (_mm_mask_i64scatter_epi64): Ditto.
39019 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
39020 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
39021 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
39022 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
39023 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
39024 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
39025 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
39026 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
39027 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
39028 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
39029 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
39030 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
39031 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
39032 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
39033 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
39034 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
39035 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
39036 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
39037 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
39038 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
39039 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
39040 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
39041 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
39042 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
39043 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
39044 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
39045 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
39046 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
39047 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
39048 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
39049 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
39050 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
39051 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
39052 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
39053 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
39054 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
39055 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
39056 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
39057 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
39058 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
39059 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
39060 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
39061 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
39062 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
39063 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
39064 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
39065 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
39066 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
39067 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
39068 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
39069 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
39070 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
39071 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
39072 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
39073 definitions accordingly.
39075 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
39076 Kuan-Lin Chen <kuanlinchentw@gmail.com>
39079 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
39082 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
39085 * ira-conflicts.c (ira_build_conflicts): Update total conflict
39086 hard regs for inner regno.
39088 2017-01-17 Martin Liska <mliska@suse.cz>
39091 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
39092 assumption and add comment.
39094 2017-01-17 Nathan Sidwell <nathan@acm.org>
39096 * ipa-visibility.c (localize_node): New function, broken out of ...
39097 (function_and_variable_visibility): ... here. Call it.
39099 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
39101 PR middle-end/77445
39102 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
39103 correctly set frequency of oudgoing edge.
39104 (duplicate_thread_path): Fix profile updating.
39106 2017-01-17 Jakub Jelinek <jakub@redhat.com>
39109 * configure.ac: Add GCC_BASE_VER.
39110 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
39111 version from BASE-VER file.
39112 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
39113 (gcc.o): Depend on $(BASEVER).
39114 * common.opt (dumpfullversion): New option.
39115 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
39116 * doc/invoke.texi: Document -dumpfullversion.
39117 * doc/install.texi: Document --with-gcc-major-version-only.
39118 * configure: Regenerated.
39120 2017-01-17 Richard Biener <rguenther@suse.de>
39122 PR tree-optimization/71433
39123 * tree-vrp.c (register_new_assert_for): Merge same asserts
39124 on all incoming edges.
39125 (process_assert_insertions_for): Handle insertions at the
39128 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
39130 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
39131 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
39133 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
39136 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
39139 2017-01-17 Alan Modra <amodra@gmail.com>
39142 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
39143 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
39144 symbolic stack limit when pic.
39146 2017-01-16 Martin Sebor <msebor@redhat.com>
39148 PR tree-optimization/78608
39149 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
39151 2017-01-16 Jeff Law <law@redhat.com>
39154 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
39155 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
39156 for several include directories that may be relative to sysroot.
39157 * config/i386/x-mingw32 (gplus_includedir): Define.
39158 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
39159 (native_system_includedir): Likewise.
39160 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
39161 override if TARGET_SYSTEM_ROOT is defined.
39162 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
39164 PR tree-optimization/79090
39165 PR tree-optimization/33562
39166 PR tree-optimization/61912
39167 PR tree-optimization/77485
39168 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
39169 and computed trims into the dump file.
39171 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
39173 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
39175 2017-01-16 Jakub Jelinek <jakub@redhat.com>
39178 * gimplify.c (gimplify_init_constructor): If want_value and
39179 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
39183 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
39184 sequence. Formatting fixes.
39185 (doloop_optimize): Formatting fixes.
39188 * gcc.c (debug_level_greater_than_spec_func): New function.
39189 (static_spec_functions): Add debug-level-gt spec function.
39190 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
39192 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
39193 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
39194 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
39195 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
39196 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
39197 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
39199 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
39201 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
39202 QImode fixups to general and mask registers only.
39204 2017-01-16 Carl Love <cel@us.ibm.com>
39206 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
39207 for built-in functions
39208 vector signed char vec_nabs (vector signed char)
39209 vector signed short vec_nabs (vector signed short)
39210 vector signed int vec_nabs (vector signed int)
39211 vector signed long long vec_nabs (vector signed long long)
39212 vector float vec_nabs (vector float)
39213 vector double vec_nabs (vector double)
39214 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
39216 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
39217 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
39218 * doc/extend.texi: Update the documentation file for the new built-in
39221 2017-01-16 Martin Sebor <msebor@redhat.com>
39223 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
39226 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39228 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
39229 UNSPEC_VSX__XXSPLTD to require special splat handling.
39231 2017-01-16 David Malcolm <dmalcolm@redhat.com>
39234 * system.h: Poison strndup.
39236 2017-01-16 Alan Modra <amodra@gmail.com>
39239 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
39242 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
39244 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
39246 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
39248 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
39249 call recog here. Assert that INSN_CODE (insn) is non-negative.
39251 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
39254 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
39256 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
39257 in the currently scheduled RTL fragment.
39259 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
39261 PR rtl-optimization/78751
39262 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
39265 2017-01-14 Jeff Law <law@redhat.com>
39267 PR tree-optimization/79090
39268 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
39269 variable length stores.
39270 (compute_trims): Delete dead assignment to *trim_tail.
39271 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
39274 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
39276 PR rtl-optimization/78626
39277 PR rtl-optimization/78727
39278 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
39279 of a block, and split such blocks after everything else is finished.
39281 2017-01-14 Alan Modra <amodra@gmail.com>
39284 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
39285 target legitimate_combined_insn.
39286 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
39287 (rs6000_legitimate_combined_insn): New function.
39288 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
39290 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
39291 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
39292 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
39294 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
39296 * doc/frontends.texi (G++ and GCC): Remove references to Java.
39298 2017-01-13 Jeff Law <law@redhat.com>
39300 PR tree-optimization/33562
39301 PR tree-optimization/61912
39302 PR tree-optimization/77485
39303 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
39305 (delete_dead_assignment): Likewise.
39306 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
39307 statement to delete_dead_call and delete_dead_assignment.
39309 2017-01-13 David Malcolm <dmalcolm@redhat.com>
39312 * substring-locations.c (format_warning_va): Strengthen case 1 so
39313 that both endpoints of the substring must be within the format
39314 range for just the substring to be printed.
39316 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
39318 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
39319 * config/i386/i386.c (ix86_target_string): Add missing options
39320 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
39321 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
39322 flags_other and ix86_target_other to flags2_other. Display unknown
39324 (ix86_valid_target_attribute_inner_p): Add missing options and
39325 reorder options by implied ISAs, as in ix86_target_string.
39327 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
39329 * hash-table.h (hash_table::too_empty_p): New function.
39330 (hash_table::expand): Use it.
39331 (hash_table::traverse): Likewise.
39332 (hash_table::empty_slot): Use sizeof (value_type) instead of
39333 sizeof (PTR) to convert bytes to elements. Shrink the table
39334 if the current size is excessive for the current number of
39337 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
39339 * ira-costs.c (record_reg_classes): Break from the inner loop
39340 early once alt_fail is known to be true. Update outer loop
39341 handling accordingly.
39343 2017-01-13 Jeff Law <law@redhat.com>
39345 * tree-ssa-dse.c (decrement_count): New function.
39346 (increment_start_addr, maybe_trim_memstar_call): Likewise.
39347 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
39348 when we know the partially dead statement is a mem* function.
39350 PR tree-optimization/61912
39351 PR tree-optimization/77485
39352 * tree-ssa-dse.c: Include expr.h.
39353 (maybe_trim_constructor_store): New function.
39354 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
39356 PR tree-optimization/33562
39357 PR tree-optimization/61912
39358 PR tree-optimization/77485
39359 * doc/invoke.texi: Document new dse-max-object-size param.
39360 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
39361 * tree-ssa-dse.c: Include params.h.
39362 (dse_store_status): New enum.
39363 (initialize_ao_ref_for_dse): New, partially extracted from
39365 (valid_ao_ref_for_dse, normalize_ref): New.
39366 (setup_live_bytes_from_ref, compute_trims): Likewise.
39367 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
39368 (maybe_trim_partially_dead_store): Likewise.
39369 (maybe_trim_complex_store): Likewise.
39370 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
39371 Track what bytes live from the original store. Return tri-state
39372 for dead, partially dead or live.
39373 (dse_dom_walker): Add constructor, destructor and new private members.
39374 (delete_dead_call, delete_dead_assignment): New extracted from
39376 (dse_optimize_stmt): Make a member of dse_dom_walker.
39377 Use initialize_ao_ref_for_dse.
39379 PR tree-optimization/33562
39380 PR tree-optimization/61912
39381 PR tree-optimization/77485
39382 * sbitmap.h (bitmap_count_bits): Prototype.
39383 (bitmap_clear_range, bitmap_set_range): Likewise.
39384 * sbitmap.c (bitmap_clear_range): New function.
39385 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
39387 2017-01-13 Martin Liska <mliska@suse.cz>
39390 * function.c (set_cfun): Add new argument force.
39391 * function.h (set_cfun): Likewise.
39392 * ipa-inline-transform.c (inline_call): Use the function when
39393 strict alising from is dropped for function we inline to.
39395 2017-01-13 Richard Biener <rguenther@suse.de>
39397 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
39398 for dumping GIMPLE INTEGER_CSTs.
39400 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
39402 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
39403 to 201112L since C++17.
39405 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
39408 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
39409 if -fsanitize=kernel-address is present.
39411 2017-01-13 Richard Biener <rguenther@suse.de>
39413 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
39414 as _Literal ( type ) number in case usual suffixes do not
39415 preserve all information.
39417 2017-01-13 Richard Biener <rguenther@suse.de>
39419 PR tree-optimization/77283
39420 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
39421 and ssa-iterators.h.
39422 (is_feasible_trace): Implement a cost model based on joiner
39425 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
39428 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
39429 char or short to __float128/_Float128 directly.
39431 2017-01-12 Martin Sebor <msebor@redhat.com>
39433 to -Wformat-overflow.
39434 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
39435 (min_bytes_remaining): Same.
39436 (get_string_length): Same.
39437 (format_string): Same.
39438 (format_directive): Same.
39440 (pass_sprintf_length::handle_gimple_call): Same.
39442 2017-01-12 Jakub Jelinek <jakub@redhat.com>
39444 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
39445 info.nowrite calls with no lhs that can't throw. Return bool
39446 whether gsi_remove has been called or not.
39447 (pass_sprintf_length::handle_gimple_call): Return bool whether
39448 try_substitute_return_value called gsi_remove. Formatting fix.
39449 (pass_sprintf_length::execute): Don't use gsi_remove if
39450 handle_gimple_call returned true.
39453 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
39454 be removed due to side-effects, don't remove following barrier nor
39455 turn the successor edge into fallthru edge.
39457 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39460 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
39461 element-reversing loads and stores as not swappable.
39463 2017-01-12 Nathan Sidwell <nathan@acm.org>
39464 Nicolai Stange <nicstange@gmail.com>
39466 * combine.c (try_combine): Don't ignore result of overlap checking
39467 loop. Combine overlap & asm check into single loop.
39469 2017-01-12 Richard Biener <rguenther@suse.de>
39471 * tree-pretty-print.c (dump_generic_node): Provide -gimple
39472 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
39474 2017-01-12 Richard Biener <rguenther@suse.de>
39476 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
39477 and TS_TARGET_OPTION directly derive from TS_BASE.
39478 * tree-core.h (tree_optimization_option): Derive from tree_base.
39479 (tree_target_option): Likewise.
39481 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
39483 * config/i386/i386.c (memory_address_length): Increase len
39484 only when rip_relative_addr_p returns false.
39486 2017-01-11 Julia Koval <julia.koval@intel.com>
39488 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
39489 (OPTION_MASK_ISA_SGX_SET): New.
39490 (ix86_handle_option): Handle OPT_msgx.
39491 * config.gcc: Added sgxintrin.h.
39492 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
39493 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
39494 * config/i386/i386.c (ix86_target_string): Add -msgx.
39496 (ix86_option_override_internal): Handle new options.
39497 (ix86_valid_target_attribute_inner_p): Add sgx.
39498 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
39499 * config/i386/i386.opt: Add msgx.
39500 * config/i386/sgxintrin.h: New file.
39501 * config/i386/x86intrin.h: Add sgxintrin.h.
39503 2017-01-11 Jakub Jelinek <jakub@redhat.com>
39506 * fold-const.c (maybe_nonzero_address): Return 1 for function
39508 (tree_single_nonzero_warnv_p): Don't handle function local objects
39512 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
39515 2017-01-11 David Malcolm <dmalcolm@redhat.com>
39518 * opts.c: Include "spellcheck.h"
39519 (struct string_fragment): New struct.
39520 (struct edit_distance_traits<const string_fragment &>): New
39522 (get_closest_sanitizer_option): New function.
39523 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
39525 2017-01-11 Jakub Jelinek <jakub@redhat.com>
39527 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
39529 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
39530 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
39531 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
39532 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
39533 for initial die_offset if dwarf_split_debug_info.
39534 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
39535 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
39537 (output_skeleton_debug_sections): Formatting fix. Use
39538 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
39539 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
39541 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
39543 * config/arm/cortex-a53.md: Add bypasses for
39544 cortex_a53_r2f_cvt.
39545 (cortex_a53_r2f): Only use for transfers.
39546 (cortex_a53_f2r): Likewise.
39547 (cortex_a53_r2f_cvt): Add reservation for conversions.
39548 (cortex_a53_f2r_cvt): Likewise.
39550 2017-01-11 Tamar Christina <tamar.christina@arm.com>
39552 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
39553 to all inlined functions, change static to extern.
39555 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
39558 * config/arm/arm.c (legitimize_pic_address): Handle reference to
39560 (arm_assemble_integer): Likewise.
39562 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
39564 * config.gcc: Use new awk script to check CPU, FPU and architecture
39565 parameters for --with-... options.
39566 * config/arm/parsecpu.awk: New file
39567 * config/arm/arm-cpus.in: New file.
39568 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
39570 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
39572 * config/arm/t-arm: Update dependency rules.
39573 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
39574 of processing .def files.
39575 * config/arm/genopt.sh: Deleted.
39576 * config/arm/gentune.sh: Deleted.
39577 * config/arm/arm-cores.def: Deleted.
39578 * config/arm/arm-arches.def: Deleted.
39579 * config/arm/arm-fpus.def: Deleted.
39580 * config/arm/arm-tune.md: Regenerated.
39581 * config/arm/arm-tables.opt: Regenerated.
39582 * config/arm/arm-cpu.h: New generated file.
39583 * config/arm/arm-cpu-data.h: New generated file.
39584 * config/arm/arm-cpu-cdata.h: New generated file.
39586 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
39589 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
39591 (input_varpool_node): Unpack dynamically_initialized bit.
39593 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
39595 PR rtl-optimization/79032
39596 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
39597 the alignment of the adjusted memory reference against that of MODE,
39598 instead of the alignment of the original memory reference.
39600 2017-01-11 Martin Jambor <mjambor@suse.cz>
39602 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
39604 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
39605 decorated functions.
39607 2017-01-11 Richard Biener <rguenther@suse.de>
39609 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
39610 set range/nonnull info for PHI results. Do not set it on
39611 stmts marked for removal.
39613 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
39615 * expr.c (store_field): In the bitfield case, fetch the return value
39616 from the registers before applying a single big-endian adjustment.
39617 Always do a final load for a BLKmode value not larger than a word.
39619 2017-01-10 David Malcolm <dmalcolm@redhat.com>
39622 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
39623 that we correctly handle column numbers greater than
39624 LINE_MAP_MAX_COLUMN_NUMBER.
39626 2017-01-10 Martin Sebor <msebor@redhat.com>
39628 PR middle-end/78245
39629 * gimple-ssa-sprintf.c (get_destination_size): Call
39630 {init,fini}object_sizes.
39631 * tree-object-size.c (addr_object_size): Adjust.
39632 (pass_through_call): Adjust.
39633 (pass_object_sizes::execute): Adjust.
39634 * tree-object-size.h (fini_object_sizes): Declare.
39636 2017-01-10 Martin Sebor <msebor@redhat.com>
39638 PR tree-optimization/78775
39639 * builtins.c (get_size_range): Move...
39640 * calls.c: ...to here.
39641 (alloc_max_size): Accept zero argument.
39642 (operand_signed_p): Remove.
39643 (maybe_warn_alloc_args_overflow): Call get_size_range.
39644 * calls.h (get_size_range): Declare.
39646 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
39648 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
39649 from TI's devices.csv file as of September 2016.
39650 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
39652 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
39654 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
39655 * doc/invoke.texi: Likewise.
39656 * doc/md.texi: Likewise.
39657 * doc/objc.texi: Likewise.
39659 2017-01-10 Joshua Conner <joshconner@google.com>
39661 * config/arm/fuchsia-elf.h: New file.
39662 * config/fuchsia.h: New file.
39663 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
39664 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
39666 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
39668 2016-01-10 Richard Biener <rguenther@suse.de>
39670 PR tree-optimization/79034
39671 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
39672 Propagate out degenerate PHIs in the joiner.
39674 2017-01-10 Martin Liska <mliska@suse.cz>
39676 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
39677 (sort_congruence_classes_by_decl_uid): Likewise.
39678 (sort_congruence_class_groups_by_decl_uid): Likewise.
39679 (sem_item_optimizer::merge_classes): Sort class, groups in these
39680 classes and members in the groups by DECL_UID of declarations.
39681 This would make merge operations stable.
39683 2017-01-10 Martin Liska <mliska@suse.cz>
39685 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
39686 usage of m_classes_vec.
39687 (sem_item_optimizer::~sem_item_optimizer): Likewise.
39688 (sem_item_optimizer::get_group_by_hash): Likewise.
39689 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
39690 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
39691 (sem_item_optimizer::verify_classes): Likewise.
39692 (sem_item_optimizer::process_cong_reduction): Likewise.
39693 (sem_item_optimizer::dump_cong_classes): Likewise.
39694 (sem_item_optimizer::merge_classes): Likewise.
39695 * ipa-icf.h (congruence_class_hash): Rename from
39696 congruence_class_group_hash. Remove declaration of m_classes_vec.
39698 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
39700 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
39701 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
39702 * config.gcc: Add avx512vpopcntdqintrin.h.
39703 * config/i386/avx512vpopcntdqintrin.h: New.
39704 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
39705 * config/i386/i386-builtin-types.def: Add new types.
39706 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
39707 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
39708 __builtin_ia32_vpopcountq_v8di_mask): New.
39709 * config/i386/i386-c.c (ix86_target_macros_internal): Define
39710 __AVX512VPOPCNTDQ__.
39711 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
39712 (PTA_AVX512VPOPCNTDQ): Define.
39713 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
39714 TARGET_AVX512VPOPCNTDQ_P): Define.
39715 * config/i386/i386.opt: Add mavx512vpopcntdq.
39716 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
39717 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
39719 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
39721 PR middle-end/77484
39722 * predict.def (PRED_CALL): Set to 67.
39724 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
39726 * expr.c (store_field): In the bitfield case, if the value comes from
39727 a function call and is of an aggregate type returned in registers, do
39728 not modify the field mode; extract the value in all cases if the mode
39729 is BLKmode and the size is not larger than a word.
39731 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
39734 * config/i386/cpuid.h: Fix undefined behavior.
39736 2017-01-04 Jeff Law <law@redhat.com>
39738 PR tree-optimization/79007
39739 PR tree-optimization/67955
39740 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
39741 conservative for pt.null when flag_non_call_exceptions is on.
39743 2017-01-09 Jakub Jelinek <jakub@redhat.com>
39745 PR translation/79019
39746 PR translation/79020
39747 * params.def (PARAM_INLINE_MIN_SPEEDUP,
39748 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
39749 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
39751 * config/avr/avr.opt (maccumulate-args): Likewise.
39752 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
39753 * common.opt (freport-bug): Likewise.
39754 * cif-code.def (CIF_FINAL_ERROR): Likewise.
39755 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
39756 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
39757 translatable string.
39758 * config/i386/i386.c (function_value_32): Likewise.
39759 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
39760 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
39762 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
39763 * common/config/msp430/msp430-common.c (msp430_handle_option):
39765 * symtab.c (symtab_node::verify_base): Likewise.
39766 * opts.c (set_debug_level): Likewise.
39767 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
39768 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
39769 missing whitespace to translatable strings.
39770 * config/avr/avr.md (bswapsi2): Fix typo in comment.
39771 * config/sh/superh.h: Likewise.
39772 * config/i386/xopintrin.h: Likewise.
39773 * config/i386/znver1.md: Likewise.
39774 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
39775 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
39776 * double-int.h (struct double_int): Likewise.
39777 * double-int.c (div_and_round_double): Likewise.
39778 * wide-int.cc: Likewise.
39779 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
39780 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
39781 * cfgcleanup.c (crossjumps_occured): Renamed to ...
39782 (crossjumps_occurred): ... this.
39783 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
39786 PR tree-optimization/78899
39787 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
39788 returning bool return struct loop *, NULL for failure and the new
39790 (versionable_outer_loop_p): Don't version outer loop if it has
39791 dont_vectorized bit set.
39792 (tree_if_conversion): When versioning outer loop, ensure
39793 tree_if_conversion is performed also on the inner loop of the
39794 non-vectorizable outer loop copy.
39795 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
39796 LOOP_VECTORIZED in inner loop of the scalar outer loop and
39797 prevent vectorization of it.
39798 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
39799 the outer loop vectorization of the non-scalar version is attempted
39800 before vectorization of the inner loop in scalar version. If
39801 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
39802 vectorization of its inner loop.
39803 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
39804 has 2 inner loops, rename also on edges from bb whose single pred
39805 is outer_loop->header. Fix typo in function comment.
39807 2017-01-09 Martin Sebor <msebor@redhat.com>
39810 * asan.c (asan_emit_stack_protection): Increase local buffer size
39811 to avoid snprintf truncation warning.
39813 2017-01-09 Andrew Pinski <apinski@cavium.com>
39815 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
39816 to reference thunderx2t99 for the tuning structure
39817 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
39819 (thunderx2t99_extra_costs): This.
39820 * config/aarch64/aarch64-tune.md: Regenerate.
39821 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
39822 (vulcan_addrcost_table): This.
39823 (vulcan_regmove_cost): Rename to ...
39824 (thunderx2t99_regmove_cost): This.
39825 (vulcan_vector_cost): Rename to ...
39826 (thunderx2t99_vector_cost): this.
39827 (vulcan_branch_cost): Rename to ...
39828 (thunderx2t99_branch_cost): This.
39829 (vulcan_tunings): Rename to ...
39830 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
39831 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
39833 2017-01-09 Martin Jambor <mjambor@suse.cz>
39837 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
39838 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
39839 (propagate_vr_accross_jump_function): Use the above function for all
39840 value range computations for pass-through jump functions and type
39841 converasion from explicit value range values.
39842 (ipcp_propagate_stage): Do not attempt to deduce types of formal
39843 parameters from TYPE_ARG_TYPES.
39844 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
39845 (ipa_write_node_info): Stream type of the actual argument.
39846 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
39848 2017-01-09 Martin Liska <mliska@suse.cz>
39851 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
39852 (lookup_compiler): Do not show error message with have_E.
39854 2017-01-09 Jakub Jelinek <jakub@redhat.com>
39856 PR tree-optimization/78938
39857 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
39858 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
39859 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
39860 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
39863 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39865 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
39868 2017-01-09 Richard Biener <rguenther@suse.de>
39870 PR tree-optimization/78997
39871 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
39872 name condition properly.
39874 2017-01-09 Richard Biener <rguenther@suse.de>
39877 * dwarf2out.c (is_cxx): New overload with context.
39878 (is_naming_typedef_decl): Use it.
39880 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
39882 * invoke.texi (Option Summary): Correct spacing in option lists
39883 and add line breaks to fix over-long lines.
39885 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
39887 PR middle-end/17660
39889 * extend.texi (Common Variable Attributes): Add xref to GCC
39890 Internals manual to explain mode attribute keywords.
39892 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
39895 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
39896 and Preprocessor Options.
39897 (Options for Linking): Document -pthread here....
39898 (RS/6000 and PowerPC Options): ...not here.
39899 (Solaris 2 Options): ...or here.
39900 * doc/cppopts.texi: Document -pthread.
39902 2017-01-08 Martin Sebor <msebor@redhat.com>
39904 PR middle-end/77708
39905 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
39906 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
39907 New member functions.
39908 (format_directive): Used them.
39910 (pass_sprintf_length::handle_gimple_call): Same.
39911 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
39912 to avoid truncation for any argument.
39913 (extract_affine_mul): Same.
39914 * tree.c (get_file_function_name): Same.
39916 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
39918 PR middle-end/77484
39919 * predict.def (PRED_INDIR_CALL): Set to 86.
39921 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
39923 PR preprocessor/54124
39924 * doc/cppopts.texi: Reformat -d subtable to list the full name
39925 of the options. Add cross-reference to the docs for the general
39926 compiler -d options.
39927 * doc/invoke.texi (Developer Options): Add cross-reference to the
39928 preprocessor-specific -d option documentation.
39930 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
39932 PR preprocessor/13498
39933 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
39934 redudant material, and reflect new command-line options.
39935 (System Headers): Likewise.
39937 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
39939 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
39940 -isystem, and -idirafter. Copy-edit.
39941 * doc/cppopts.texi: Copy-edit. Remove contradiction about
39942 default for -ftrack-macro-expansion. Delete obsolete and
39943 badly-formatted implementation details about -fdebug-cpp output.
39944 * doc/cppwarnopts.texi: Copy-edit.
39946 2017-01-07 David Malcolm <dmalcolm@redhat.com>
39949 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
39950 that the transition from a max line width >= 1<<10 to narrower
39951 lines works correctly.
39953 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
39955 * doc/options.texi (PerFunction): New.
39956 * opt-functions.awk (switch_flags): Map both Optimization and
39957 PerFunction to CL_OPTIMIZATION.
39958 * opth-gen.awk: Test for PerFunction flag along with
39960 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
39961 it only when the latter is present. Skip those that don't in
39962 the hash function generator.
39963 * common.opt (fvar-tracking): Mark as PerFunction instead of
39965 (fvar-tracking-assignments): Likewise.
39966 (fvar-tracking-assignments-toggle): Likewise.
39967 (fvar-tracking-uninit): Likewise.
39969 2017-01-07 Jakub Jelinek <jakub@redhat.com>
39971 PR translation/79018
39972 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
39975 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
39978 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
39979 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
39980 TARGET_LONG_JUMP_TABLE_OFFSETS.
39981 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
39982 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
39983 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
39984 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
39985 * config/m68k/m68k.md (tablejump expander): Likewise.
39986 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
39987 TARGET_LONG_JUMP_TABLE_OFFSETS.
39988 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
39989 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
39991 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
39992 David Holsgrove <david.holsgrove@xilinx.com>
39994 * common/config/microblaze/microblaze-common.c
39995 (TARGET_EXCEPT_UNWIND_INFO): Remove.
39996 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
39998 * config/microblaze/microblaze.c (microblaze_must_save_register)
39999 (microblaze_expand_epilogue, microblaze_return_addr): Handle
40001 (microblaze_eh_return): New function.
40002 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
40003 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
40004 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
40005 * config/microblaze/microblaze.md (eh_return): New pattern.
40007 2017-01-06 Jakub Jelinek <jakub@redhat.com>
40009 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
40010 GCC_DIAGNOSTIC_STRINGIFY): Define.
40012 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
40014 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40016 * config/arm/arm.md (<mcrr>): New.
40018 * config/arm/arm.c (arm_arch5te): New.
40019 (arm_option_override): Set arm_arch5te.
40020 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
40022 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
40023 (arm_mcrr_qualifiers): ... this. New.
40024 (MRRC_QUALIFIERS): Define to...
40025 (arm_mrrc_qualifiers): ... this. New.
40026 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
40028 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
40029 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
40030 (MRRCI, mrrc, MRRC): New.
40031 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
40032 VUNSPEC_MRRC2): New.
40034 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40036 * config/arm/arm.md (<mcr>): New.
40038 * config/arm/arm.c (arm_coproc_builtin_available): Add
40039 support for mcr, mrc, mcr2 and mrc2.
40040 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
40041 (arm_mcr_qualifiers): ... this. New.
40042 (MRC_QUALIFIERS): Define to ...
40043 (arm_mrc_qualifiers): ... this. New.
40044 (MCR_QUALIFIERS): Define to ...
40045 (arm_mcr_qualifiers): ... this. New.
40046 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
40048 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
40049 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
40050 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
40051 VUNSPEC_MRC2): New.
40053 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40055 * config/arm/arm.md (*ldc): New.
40059 * config/arm/arm.c (arm_coproc_builtin_available): Add
40060 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
40061 (arm_coproc_ldc_stc_legitimate_address): New.
40062 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
40063 'qualifier_const_pointer'.
40064 (LDC_QUALIFIERS): Define to...
40065 (arm_ldc_qualifiers): ... this. New.
40066 (STC_QUALIFIERS): Define to...
40067 (arm_stc_qualifiers): ... this. New.
40068 * config/arm/arm-protos.h
40069 (arm_coproc_ldc_stc_legitimate_address): New.
40070 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
40071 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
40072 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
40073 stc2, stcl, stc2l): New.
40074 * config/arm/constraints.md (Uz): New.
40075 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
40076 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
40077 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
40078 VUNSPEC_STC2L): New.
40080 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40082 * config/arm/arm.md (<cdp>): New.
40083 * config/arm/arm.c (neon_const_bounds): Rename this ...
40084 (arm_const_bounds): ... this.
40085 (arm_coproc_builtin_available): New.
40086 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
40087 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
40088 (CDP_QUALIFIERS): Define to...
40089 (arm_cdp_qualifiers): ... this. New.
40091 (arm_expand_builtin_args): Add case for 6 arguments.
40092 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
40093 (arm_const_bounds): ... this.
40094 (arm_coproc_builtin_available): New.
40095 * config/arm/arm_acle.h (__arm_cdp): New.
40097 * config/arm/arm_acle_builtins.def (cdp): New.
40099 * config/arm/iterators.md (CDPI,CDP,cdp): New.
40100 * config/arm/neon.md: Rename all 'neon_const_bounds' to
40101 'arm_const_bounds'.
40102 * config/arm/types.md (coproc): New.
40103 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
40104 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
40105 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
40106 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
40108 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40110 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
40111 (UBINOP_QUALIFIERS): New.
40113 (acle_builtin_data): New. Change comment.
40114 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
40115 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
40116 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
40117 arm_acle_builtins.def.
40118 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
40119 (arm_init_acle_builtins): New.
40120 (CRC32_BUILTIN): Remove.
40121 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
40122 crc32cb, crc32ch and crc32cw.
40123 (arm_init_crc32_builtins): Remove.
40124 (arm_init_builtins): Use arm_init_acle_builtins rather
40125 than arm_init_crc32_builtins.
40126 (arm_expand_acle_builtin): New.
40127 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
40128 * config/arm/arm_acle_builtins.def: New.
40130 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40132 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
40133 (arm_builtin_datum): ... this.
40134 (arm_init_neon_builtin): Rename to ...
40135 (arm_init_builtin): ... this. Add a new parameters PREFIX
40136 and USE_SIG_IN_NAME.
40137 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
40138 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
40139 'arm_builtin_datum'.
40140 (arm_init_vfp_builtins): Likewise.
40141 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
40142 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
40143 (arm_expand_neon_args): Rename to ...
40144 (arm_expand_builtin_args): ... this. Rename builtin_arg
40145 enum values and differentiate between ARG_BUILTIN_MEMORY
40146 and ARG_BUILTIN_NEON_MEMORY.
40147 (arm_expand_neon_builtin_1): Rename to ...
40148 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
40149 values, arm_expand_builtin_args and add bool parameter NEON.
40150 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
40151 (arm_expand_vfp_builtin): Likewise.
40152 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
40154 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40156 PR middle-end/77484
40157 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
40158 * predict.c (tree_estimate_probability_bb): Reverse direction of
40159 polymorphic call predictor.
40161 2017-01-06 David Malcolm <dmalcolm@redhat.com>
40163 * passes.c (execute_one_pass): Split out pass-skipping logic into...
40164 (determine_pass_name_match): ...this new function and...
40165 (should_skip_pass_p): ...this new function.
40167 2017-01-06 Nathan Sidwell <nathan@acm.org>
40169 * ipa-visibility.c (function_and_variable_visibility): Reformat
40170 comments and long lines. Remove extrneous if.
40171 * symtab.c (symtab_node::make_decl_local): Fix code format.
40172 (symtab_node::set_section_for_node): Fix comment typo.
40174 2017-01-06 Martin Liska <mliska@suse.cz>
40177 * lra-constraints.c: Rename invariant to lra_invariant.
40178 * predict.c (set_even_probabilities): Initialize e to NULL.
40180 2017-01-05 Martin Sebor <msebor@redhat.com>
40182 PR tree-optimization/78910
40183 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
40184 (format_integer): Correct off-by-one error in the handling
40185 of precision with negative numbers in signed conversions..
40187 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
40189 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
40191 2017-01-05 Jakub Jelinek <jakub@redhat.com>
40193 PR tree-optimization/71016
40194 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
40195 factor_out_conditional_conversion. Formatting fix.
40196 (factor_out_conditional_conversion): Add cond_stmt argument.
40197 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
40198 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
40201 2017-01-05 David Malcolm <dmalcolm@redhat.com>
40203 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
40204 read-rtl-function.o, and selftest-rtl.o.
40205 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
40206 (selftest::aarch64_test_loading_full_dump): New function.
40207 (selftest::aarch64_run_selftests): New function.
40208 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
40209 selftest::aarch64_run_selftests.
40210 * config/i386/i386.c
40211 (selftest::ix86_test_loading_dump_fragment_1): New function.
40212 (selftest::ix86_test_loading_call_insn): New function.
40213 (selftest::ix86_test_loading_full_dump): New function.
40214 (selftest::ix86_test_loading_unspec): New function.
40215 (selftest::ix86_run_selftests): Call the new functions.
40216 * emit-rtl.c (maybe_set_max_label_num): New function.
40217 * emit-rtl.h (maybe_set_max_label_num): New decl.
40218 * function.c (instantiate_decls): Guard call to
40219 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
40220 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
40222 * gensupport.c (gen_reader::gen_reader): Pass "false"
40223 for new "compact" param of rtx_reader.
40224 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
40225 rather than an empty string for NULL strings.
40226 * read-md.c: Potentially include config.h rather than bconfig.h.
40227 Wrap include of errors.h with #ifdef GENERATOR_FILE.
40228 (have_error): New global, copied from errors.c.
40229 (md_reader::read_name): Rename to...
40230 (md_reader::read_name_1): ...this, adding "out_loc" param,
40231 and converting "missing name or number" to returning false, rather
40233 (md_reader::read_name): Reimplement in terms of read_name_1.
40234 (md_reader::read_name_or_nil): New function.
40235 (md_reader::read_string): Handle "(nil)" by returning NULL.
40236 (md_reader::md_reader): Add new param "compact".
40237 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
40238 (md_reader::read_file): New method.
40239 * read-md.h (md_reader::md_reader): Add new param "compact".
40240 (md_reader::read_file): New method.
40241 (md_reader::is_compact): New accessor.
40242 (md_reader::read_name): Convert return type from void to file_location.
40243 (md_reader::read_name_or_nil): New decl.
40244 (md_reader::read_name_1): New decl.
40245 (md_reader::m_compact): New field.
40246 (noop_reader::noop_reader): Pass "false" for new "compact" param
40248 (rtx_reader::rtx_reader): Add new "compact" param.
40249 (rtx_reader::read_rtx_operand): Make virtual and convert return
40250 type from void to rtx.
40251 (rtx_reader::read_until): New decl.
40252 (rtx_reader::handle_any_trailing_information): New virtual function.
40253 (rtx_reader::postprocess): New virtual function.
40254 (rtx_reader::finalize_string): New virtual function.
40255 (rtx_reader::m_in_call_function_usage): New field.
40256 (rtx_reader::m_reuse_rtx_by_id): New field.
40257 * read-rtl-function.c: New file.
40258 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
40259 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
40260 (selftest::verify_three_block_rtl_cfg): New decl.
40261 * read-rtl-function.h: New file.
40262 * read-rtl.c: Potentially include config.h rather than bconfig.h.
40263 For host, include function.h, memmodel.h, and emit-rtl.h.
40264 (one_time_initialization): New function.
40265 (struct compact_insn_name): New struct.
40266 (compact_insn_names): New array.
40267 (find_code): Handle insn codes in compact dumps.
40268 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
40269 (bind_subst_iter_and_attr): Likewise.
40270 (add_condition_to_string): Likewise.
40271 (add_condition_to_rtx): Likewise.
40272 (apply_attribute_uses): Likewise.
40273 (add_current_iterators): Likewise.
40274 (apply_iterators): Likewise.
40275 (initialize_iterators): Guard usage of apply_subst_iterator with
40276 #ifdef GENERATOR_FILE.
40277 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
40278 (md_reader::read_mapping): Likewise.
40279 (add_define_attr_for_define_subst): Likewise.
40280 (add_define_subst_attr): Likewise.
40281 (read_subst_mapping): Likewise.
40282 (check_code_iterator): Likewise.
40283 (rtx_reader::read_rtx): Likewise. Move one-time initialization
40285 (one_time_initialization): New function.
40286 (rtx_reader::read_until): New method.
40287 (read_flags): New function.
40288 (parse_reg_note_name): New function.
40289 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
40290 Handle reuse_rtx ids.
40291 Wrap iterator lookup within #ifdef GENERATOR_FILE.
40292 Add parsing support for RTL dumps, mirroring the special-cases in
40293 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
40294 values, and calling handle_any_trailing_information.
40295 (rtx_reader::read_rtx_operand): Convert return type from void
40296 to rtx, returning return_rtx. Handle case 'e'. Call
40297 finalize_string on XSTR and XTMPL fields.
40298 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
40299 "(nil)" values were omitted. Call the postprocess vfunc on the
40301 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
40302 class ctor. Initialize m_in_call_function_usage. Call
40303 one_time_initialization.
40304 * rtl-tests.c (selftest::test_uncond_jump): Call
40305 set_new_first_and_last_insn.
40306 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
40307 * selftest-rtl.c: New file.
40308 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
40309 (selftest::get_insn_by_uid): New decl.
40310 * selftest-run-tests.c (selftest::run_tests): Call
40311 read_rtl_function_c_tests.
40312 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
40313 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
40316 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
40318 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
40319 operands in a special way. Assert that pos+len <= mode precision.
40321 2017-01-05 Jakub Jelinek <jakub@redhat.com>
40323 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
40324 3 argument Alias with unlimited for the negative form.
40325 (fno-vect-cost-model): Removed.
40327 2017-01-05 Martin Liska <mliska@suse.cz>
40329 * hsa-gen.c (gen_hsa_divmod): New function.
40330 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
40332 2017-01-05 Martin Liska <mliska@suse.cz>
40335 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
40338 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
40340 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
40341 small constant length operands.
40343 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
40345 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
40346 between loop iterations.
40348 2017-01-05 Martin Liska <mliska@suse.cz>
40351 * gimplify.c (gimplify_decl_expr): Compare to
40352 asan_poisoned_variables instread of checking flags.
40353 (gimplify_target_expr): Likewise.
40354 (gimplify_expr): Likewise.
40355 (gimplify_function_tree): Conditionally initialize
40356 asan_poisoned_variables.
40358 2017-01-04 Jeff Law <law@redhat.com>
40360 PR tree-optimizatin/78812
40361 * rtl.h (contains_mem_rtx_p): Prototype.
40362 * ifcvt.c (containts_mem_rtx_p): Move from here to...
40363 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
40364 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
40365 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
40366 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
40368 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
40370 * input.c (assert_char_at_range): Default-initialize actual_range.
40372 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
40374 * df-scan.c (df_ref_create_structure): Make regno unsigned,
40375 to match the caller.
40377 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
40379 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
40380 insns after final jump in test to emit dummy move.
40382 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
40384 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
40385 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
40387 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
40389 * multiple_target.c (create_dispatcher_calls): Init e_next.
40390 * tree-ssa-loop-split.c (split_loop): Init border.
40391 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
40394 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
40399 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
40400 (altivec_register_operand): Do not return true if the operand
40401 contains a SUBREG mixing SImode and SFmode.
40402 (vsx_register_operand): Likewise.
40403 (vsx_reg_sfsubreg_ok): New predicate.
40404 (vfloat_operand): Do not return true if the operand contains a
40405 SUBREG mixing SImode and SFmode.
40406 (vint_operand): Likewise.
40407 (vlogical_operand): Likewise.
40408 (gpc_reg_operand): Likewise.
40409 (int_reg_operand): Likewise.
40410 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
40411 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
40412 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
40414 (rs6000_emit_move_si_sf_subreg): New helper function.
40415 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
40416 fixup SUBREGs involving SImode and SFmode.
40417 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
40418 numbers for the new peephole2 optimization.
40419 (peephole2 for SFmode unions): New peephole2 to optimize cases in
40420 the GLIBC math library that do AND/IOR/XOR operations on single
40421 precision floating point.
40422 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
40423 target macros to say whether we need to avoid SUBREGs mixing
40425 (TARGET_ALLOW_SF_SUBREG): Likewise.
40426 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
40427 (UNSPEC_SI_FROM_SF): Likewise.
40428 (iorxor): Change spacing.
40429 (and_ior_xor): New iterator for AND, IOR, and XOR.
40430 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
40431 (movdi_from_sf_zero_ext): Likewise.
40432 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
40433 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
40434 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
40435 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
40436 (fms<mode>4): Likewise.
40437 (fnma<mode>4): Likewise.
40438 (fnms<mode>4): Likewise.
40439 (nfma<mode>4): Likewise.
40440 (nfms<mode>4): Likewise.
40442 2017-01-04 Marek Polacek <polacek@redhat.com>
40445 * doc/invoke.texi: Document -Wpointer-compare.
40447 2017-01-04 Jakub Jelinek <jakub@redhat.com>
40449 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
40452 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
40453 descriptions for -gdwarf-5 and emit them as uleb128 instead of
40456 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
40459 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
40460 documentation of the powerpc_popcntb_ok attribute.
40461 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
40462 code to issue warning messages if a requested CPU configuration is
40463 not supported by the binary (assembler and loader) toolchain.
40464 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
40465 made to define a built-in function that has been disabled.
40466 (paired_init_builtins): Add assertion to prevent ICE if attempt is
40467 made to define a built-in function that has been disabled.
40468 (altivec_init_builtins): Add comment explaining why definition
40469 of the DST built-in functions is not preceded by an assertion
40470 check. Add assertions to prevent ICE if attempts are made to
40471 define an altivec predicate or an abs* built-in function that has
40473 (htm_init_builtins): Add comment explaining why definition of the
40474 htm built-in functions is not preceded by an assertion check.
40476 2017-01-04 Jeff Law <law@redhat.com>
40478 PR tree-optimizatin/67955
40479 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
40480 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
40481 the points-to solution does not include pt_null. Use DECL_PT_UID
40484 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
40486 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
40487 Use gen_int_mode instead of gen_lopwart for const_int operands.
40489 2017-01-04 Jakub Jelinek <jakub@redhat.com>
40491 PR tree-optimization/71563
40492 * match.pd: Simplify X << Y into X if Y is known to be 0 or
40493 out of range value - has low bits known to be zero.
40495 2017-01-04 Alan Modra <amodra@gmail.com>
40497 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
40498 * configure: Regenerate.
40499 * config.in: Regenerate.
40501 2017-01-04 Jakub Jelinek <jakub@redhat.com>
40504 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
40505 a substring of the message, but strcmp with the whole message. Ifdef
40506 ENABLE_NLS, translate the message first using dgettext.
40508 2017-01-03 Jeff Law <law@redhat.com>
40510 PR tree-optimizatin/78856
40511 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
40512 (mark_threaded_blocks): Remove code to truncate thread paths that
40513 cross multiple loop headers. Instead invalidate the cached loop
40514 iteration information and handle case of a thread path walking
40515 into an irreducible region.
40517 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
40520 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
40521 assertions. Add support for doing the signbit if the IEEE 128-bit
40522 floating point value is in a GPR.
40523 * config/rs6000/rs6000.md (Fsignbit): Delete.
40524 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
40525 Update the length attribute if the value is in a GPR.
40526 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
40527 the sign or zero extension instruction, since the value is always 0/1.
40528 (signbit<mode>2_dm2): Delete using <Fsignbit>.
40531 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
40532 extracting SImode to a GPR register so that we can generate a
40533 store, limit the vector to be in a traditional Altivec register
40534 for the vextuwrx instruction.
40536 2017-01-03 Ian Lance Taylor <iant@google.com>
40538 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
40540 2017-01-03 Martin Sebor <msebor@redhat.com>
40542 PR tree-optimization/78696
40543 * gimple-ssa-sprintf.c (format_floating): Correct handling of
40544 precision. Use MPFR for %f for greater fidelity. Correct handling
40546 (pass_sprintf_length::compute_format_length): Set width and precision
40547 specified by asrerisk to void_node for vararg functions.
40548 (try_substitute_return_value): Adjust dump output.
40550 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
40552 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
40554 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
40556 * doc/invoke.texi (SPARC options): Document -mlra as the default.
40557 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
40558 -mlra/-mno-lra was passed to the compiler.
40560 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
40562 PR rtl-optimization/65618
40563 * emit-rtl.c (try_split): Move initialization of "before" and
40564 "after" to just before the call to emit_insn_after_setloc.
40566 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
40568 * doc/md.texi (Standard Names): Remove reference to Java frontend.
40570 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
40572 * dwarf2out.c (gen_enumeration_type_die): When
40573 -gno-strict-dwarf, add a DW_AT_encoding attribute.
40575 2017-01-03 Jakub Jelinek <jakub@redhat.com>
40577 PR tree-optimization/78965
40578 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
40579 Change first argument from const call_info & to call_info &. For %n
40580 set info.nowrite to false.
40582 PR middle-end/78901
40583 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
40584 possibly throwing calls.
40586 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
40587 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
40588 and fns handling, rather than in a separate case SSA_NAME.
40590 2017-01-02 Jeff Law <law@redhat.com>
40592 * config/darwin-driver.c (darwin_driver_init): Const-correctness
40593 fixes for first_period and second_period variables.
40595 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
40598 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
40599 (*insvqi_1): New insn pattern.
40600 (*insvqi_1_mem_rex64): Ditto.
40601 (*insvqi_2): Ditto.
40602 (*insvqi_3): Rename from *insvqi.
40604 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
40606 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
40608 * doc/cfg.texi (Edges): Remove reference to Java.
40609 (Maintaining the CFG): Ditto.
40611 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40613 PR middle-end/77674
40614 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
40615 transparent aliases.
40617 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40619 PR middle-end/77484
40620 * predict.def (PRED_CALL): Update hitrate.
40621 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
40622 * predict.c (tree_estimate_probability_bb): Split CALL predictor
40623 into direct/indirect/polymorphic variants.
40625 2017-01-01 Jakub Jelinek <jakub@redhat.com>
40627 Update copyright years.
40629 * gcc.c (process_command): Update copyright notice dates.
40630 * gcov-dump.c (print_version): Ditto.
40631 * gcov.c (print_version): Ditto.
40632 * gcov-tool.c (print_version): Ditto.
40633 * gengtype.c (create_file): Ditto.
40634 * doc/cpp.texi: Bump @copying's copyright year.
40635 * doc/cppinternals.texi: Ditto.
40636 * doc/gcc.texi: Ditto.
40637 * doc/gccint.texi: Ditto.
40638 * doc/gcov.texi: Ditto.
40639 * doc/install.texi: Ditto.
40640 * doc/invoke.texi: Ditto.
40642 Copyright (C) 2017 Free Software Foundation, Inc.
40644 Copying and distribution of this file, with or without modification,
40645 are permitted in any medium without royalty provided the copyright
40646 notice and this notice are preserved.