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 * tree-affine.h (aff_tree::offset): Change from widest_int
7 (wide_int_ext_for_comb): Delete.
8 (aff_combination_const, aff_comb_cannot_overlap_p): Take the
9 constants as poly_widest_int rather than widest_int.
10 (aff_combination_constant_multiple_p): Return the multiplier
12 (aff_combination_zero_p, aff_combination_singleton_var_p): Handle
14 * tree-affine.c (wide_int_ext_for_comb): Make original widest_int
15 version static and add an overload for poly_widest_int.
16 (aff_combination_const, aff_combination_add_cst)
17 (wide_int_constant_multiple_p, aff_comb_cannot_overlap_p): Take
18 the constants as poly_widest_int rather than widest_int.
19 (tree_to_aff_combination): Generalize INTEGER_CST case to
21 (aff_combination_to_tree): Track offsets as poly_widest_ints.
22 (aff_combination_add_product, aff_combination_mult): Handle
24 (aff_combination_constant_multiple_p): Return the multiplier
26 * tree-predcom.c (determine_offset): Return the offset as a
28 (split_data_refs_to_components, suitable_component_p): Update
30 (valid_initializer_p): Update call to
31 aff_combination_constant_multiple_p.
32 * tree-ssa-address.c (addr_to_parts): Handle polynomial offsets.
33 * tree-ssa-loop-ivopts.c (get_address_cost_ainc): Take the step
34 as a poly_int64 rather than a HOST_WIDE_INT.
35 (get_address_cost): Handle polynomial offsets.
36 (iv_elimination_compare_lt): Likewise.
37 (rewrite_use_nonlinear_expr): Likewise.
39 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
40 Alan Hayward <alan.hayward@arm.com>
41 David Sherwood <david.sherwood@arm.com>
43 * tree-dfa.h (get_addr_base_and_unit_offset_1): Return the offset
44 as a poly_int64_pod rather than a HOST_WIDE_INT.
45 (get_addr_base_and_unit_offset): Likewise.
46 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
47 (get_addr_base_and_unit_offset): Likewise.
48 * doc/match-and-simplify.texi: Change off from HOST_WIDE_INT
49 to poly_int64 in example.
50 * fold-const.c (fold_binary_loc): Update call to
51 get_addr_base_and_unit_offset.
52 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
53 (maybe_canonicalize_mem_ref_addr): Likewise.
54 (gimple_fold_stmt_to_constant_1): Likewise.
55 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
57 * ipa-param-manipulation.c (ipa_modify_call_arguments): Likewise.
59 * omp-low.c (lower_omp_target): Likewise.
60 * tree-sra.c (build_ref_for_offset): Likewise.
61 (build_debug_ref_for_model): Likewise.
62 * tree-ssa-address.c (maybe_fold_tmr): Likewise.
63 * tree-ssa-alias.c (ao_ref_init_from_ptr_and_size): Likewise.
64 * tree-ssa-ccp.c (optimize_memcpy): Likewise.
65 * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Likewise.
66 (constant_pointer_difference): Likewise.
67 * tree-ssa-loop-niter.c (expand_simple_operations): Likewise.
68 * tree-ssa-phiopt.c (jump_function_from_stmt): Likewise.
69 * tree-ssa-pre.c (create_component_ref_by_pieces_1): Likewise.
70 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
71 (vn_reference_maybe_forwprop_address, vn_reference_lookup_3): Likewise.
72 (set_ssa_val_to): Likewise.
73 * tree-ssa-strlen.c (get_addr_stridx, addr_stridxptr)
74 (maybe_diag_stxncpy_trunc): Likewise.
75 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
76 * tree.c (build_simple_mem_ref_loc): Likewise.
77 (array_at_struct_end_p): Likewise.
79 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
80 Alan Hayward <alan.hayward@arm.com>
81 David Sherwood <david.sherwood@arm.com>
83 * tree-dfa.h (get_ref_base_and_extent): Return the base, size and
84 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
85 (get_ref_base_and_extent_hwi): Declare.
86 * tree-dfa.c (get_ref_base_and_extent): Return the base, size and
87 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
88 (get_ref_base_and_extent_hwi): New function.
89 * cfgexpand.c (expand_debug_expr): Update call to
90 get_ref_base_and_extent.
91 * dwarf2out.c (add_var_loc_to_decl): Likewise.
92 * gimple-fold.c (get_base_constructor): Return the offset as a
93 poly_int64_pod rather than a HOST_WIDE_INT.
94 (fold_const_aggregate_ref_1): Track polynomial sizes and offsets.
95 * ipa-polymorphic-call.c
96 (ipa_polymorphic_call_context::set_by_invariant)
97 (extr_type_from_vtbl_ptr_store): Track polynomial offsets.
98 (ipa_polymorphic_call_context::ipa_polymorphic_call_context)
99 (check_stmt_for_type_change): Use get_ref_base_and_extent_hwi
100 rather than get_ref_base_and_extent.
101 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
102 * ipa-prop.c (ipa_load_from_parm_agg, compute_complex_assign_jump_func)
103 (get_ancestor_addr_info, determine_locally_known_aggregate_parts):
105 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Update
106 call to get_ref_base_and_extent.
107 * tree-sra.c (create_access, get_access_for_expr): Likewise.
108 * tree-ssa-alias.c (ao_ref_base, aliasing_component_refs_p)
109 (stmt_kills_ref_p): Likewise.
110 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
111 * tree-ssa-scopedtables.c (avail_expr_hash, equal_mem_array_ref_p):
113 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
114 Use get_ref_base_and_extent_hwi rather than get_ref_base_and_extent
115 when calling native_encode_expr.
116 * tree-ssa-structalias.c (get_constraint_for_component_ref): Update
117 call to get_ref_base_and_extent.
118 (do_structure_copy): Use get_ref_base_and_extent_hwi rather than
119 get_ref_base_and_extent.
120 * var-tracking.c (track_expr_p): Likewise.
122 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
123 Alan Hayward <alan.hayward@arm.com>
124 David Sherwood <david.sherwood@arm.com>
126 * ipa-param-manipulation.h (ipa_parm_adjustment::offset): Change from
127 HOST_WIDE_INT to poly_int64_pod.
128 * ipa-param-manipulation.c (ipa_modify_call_arguments): Track
129 polynomail parameter offsets.
131 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
132 Alan Hayward <alan.hayward@arm.com>
133 David Sherwood <david.sherwood@arm.com>
135 * gengtype.c (main): Handle poly_int64_pod.
136 * dwarf2out.h (dw_cfi_oprnd_cfa_loc): New dw_cfi_oprnd_type.
137 (dw_cfi_oprnd::dw_cfi_cfa_loc): New field.
138 (dw_cfa_location::offset, dw_cfa_location::base_offset): Change
139 from HOST_WIDE_INT to poly_int64_pod.
140 * dwarf2cfi.c (queued_reg_save::cfa_offset): Likewise.
141 (copy_cfa): New function.
142 (lookup_cfa_1): Use the cached dw_cfi_cfa_loc, if it exists.
143 (cfi_oprnd_equal_p): Handle dw_cfi_oprnd_cfa_loc.
144 (cfa_equal_p, dwarf2out_frame_debug_adjust_cfa)
145 (dwarf2out_frame_debug_cfa_offset, dwarf2out_frame_debug_expr)
146 (initial_return_save): Treat offsets as poly_ints.
147 (def_cfa_0): Likewise. Cache the CFA in dw_cfi_cfa_loc if either
148 offset is nonconstant.
149 (reg_save): Take the offset as a poly_int64. Fall back to
150 DW_CFA_expression for nonconstant offsets.
151 (queue_reg_save): Take the offset as a poly_int64.
152 * dwarf2out.c (dw_cfi_oprnd2_desc): Handle DW_CFA_def_cfa_expression.
154 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
155 Alan Hayward <alan.hayward@arm.com>
156 David Sherwood <david.sherwood@arm.com>
158 * rtl.h (operand_subword, operand_subword_force): Take the offset
159 as a poly_uint64 an unsigned int.
160 * emit-rtl.c (operand_subword, operand_subword_force): Likewise.
162 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
163 Alan Hayward <alan.hayward@arm.com>
164 David Sherwood <david.sherwood@arm.com>
166 * doc/rtl.texi: Update documentation of SUBREG_BYTE. Document the
167 'p' format code. Use INT_LIST rather than SUBREG as the example of
168 a code with an XINT and an XEXP. Remove the implication that
169 accessing an rtx field using XINT is expected to work.
170 * rtl.def (SUBREG): Change format from "ei" to "ep".
171 * rtl.h (rtunion::rt_subreg): New field.
172 (XCSUBREG): New macro.
173 (SUBREG_BYTE): Use it.
174 (subreg_shape): Change offset from an unsigned int to a poly_uint16.
175 Update constructor accordingly.
176 (subreg_shape::operator ==): Update accordingly.
177 (subreg_shape::unique_id): Return an unsigned HOST_WIDE_INT rather
178 than an unsigned int.
179 (subreg_lsb, subreg_lowpart_offset, subreg_highpart_offset): Return
180 a poly_uint64 rather than an unsigned int.
181 (subreg_lsb_1): Likewise. Take the offset as a poly_uint64 rather
182 than an unsigned int.
183 (subreg_size_offset_from_lsb, subreg_size_lowpart_offset)
184 (subreg_size_highpart_offset): Return a poly_uint64 rather than
185 an unsigned int. Take the sizes as poly_uint64s.
186 (subreg_offset_from_lsb): Return a poly_uint64 rather than
187 an unsigned int. Take the shift as a poly_uint64 rather than
189 (subreg_regno_offset, subreg_offset_representable_p): Take the offset
190 as a poly_uint64 rather than an unsigned int.
191 (simplify_subreg_regno): Likewise.
192 (byte_lowpart_offset): Return the memory offset as a poly_int64
194 (subreg_memory_offset): Likewise. Take the subreg offset as a
195 poly_uint64 rather than an unsigned int.
196 (simplify_subreg, simplify_gen_subreg, subreg_get_info)
197 (gen_rtx_SUBREG, validate_subreg): Take the subreg offset as a
198 poly_uint64 rather than an unsigned int.
199 * rtl.c (rtx_format): Describe 'p' in comment.
200 (copy_rtx, rtx_equal_p_cb, rtx_equal_p): Handle 'p'.
201 * emit-rtl.c (validate_subreg, gen_rtx_SUBREG): Take the subreg
202 offset as a poly_uint64 rather than an unsigned int.
203 (byte_lowpart_offset): Return the memory offset as a poly_int64
205 (subreg_memory_offset): Likewise. Take the subreg offset as a
206 poly_uint64 rather than an unsigned int.
207 (subreg_size_lowpart_offset, subreg_size_highpart_offset): Take the
208 mode sizes as poly_uint64s rather than unsigned ints. Return a
209 poly_uint64 rather than an unsigned int.
210 (subreg_lowpart_p): Treat subreg offsets as poly_ints.
211 (copy_insn_1): Handle 'p'.
212 * rtlanal.c (set_noop_p): Treat subregs offsets as poly_uint64s.
213 (subreg_lsb_1): Take the subreg offset as a poly_uint64 rather than
214 an unsigned int. Return the shift in the same way.
215 (subreg_lsb): Return the shift as a poly_uint64 rather than an
217 (subreg_size_offset_from_lsb): Take the sizes and shift as
218 poly_uint64s rather than unsigned ints. Return the offset as
220 (subreg_get_info, subreg_regno_offset, subreg_offset_representable_p)
221 (simplify_subreg_regno): Take the offset as a poly_uint64 rather than
223 * rtlhash.c (add_rtx): Handle 'p'.
224 * genemit.c (gen_exp): Likewise.
225 * gengenrtl.c (type_from_format, gendef): Likewise.
226 * gensupport.c (subst_pattern_match, get_alternatives_number)
227 (collect_insn_data, alter_predicate_for_insn, alter_constraints)
228 (subst_dup): Likewise.
229 * gengtype.c (adjust_field_rtx_def): Likewise.
230 * genrecog.c (find_operand, find_matching_operand, validate_pattern)
231 (match_pattern_2): Likewise.
232 (rtx_test::SUBREG_FIELD): New rtx_test::kind_enum.
233 (rtx_test::subreg_field): New function.
234 (operator ==, safe_to_hoist_p, transition_parameter_type)
235 (print_nonbool_test, print_test): Handle SUBREG_FIELD.
236 * genattrtab.c (attr_rtx_1): Say that 'p' is deliberately not handled.
237 * genpeep.c (match_rtx): Likewise.
238 * print-rtl.c (print_poly_int): Include if GENERATOR_FILE too.
239 (rtx_writer::print_rtx_operand): Handle 'p'.
240 (print_value): Handle SUBREG.
241 * read-rtl.c (apply_int_iterator): Likewise.
242 (rtx_reader::read_rtx_operand): Handle 'p'.
243 * alias.c (rtx_equal_for_memref_p): Likewise.
244 * cselib.c (rtx_equal_for_cselib_1, cselib_hash_rtx): Likewise.
245 * caller-save.c (replace_reg_with_saved_mem): Treat subreg offsets
247 * calls.c (expand_call): Likewise.
248 * combine.c (combine_simplify_rtx, expand_field_assignment): Likewise.
249 (make_extraction, gen_lowpart_for_combine): Likewise.
250 * loop-invariant.c (hash_invariant_expr_1, invariant_expr_equal_p):
252 * cse.c (remove_invalid_subreg_refs): Take the offset as a poly_uint64
253 rather than an unsigned int. Treat subreg offsets as poly_ints.
254 (exp_equiv_p): Handle 'p'.
255 (hash_rtx_cb): Likewise. Treat subreg offsets as poly_ints.
256 (equiv_constant, cse_insn): Treat subreg offsets as poly_ints.
257 * dse.c (find_shift_sequence): Likewise.
258 * dwarf2out.c (rtl_for_decl_location): Likewise.
259 * expmed.c (extract_low_bits): Likewise.
260 * expr.c (emit_group_store, undefined_operand_subword_p): Likewise.
261 (expand_expr_real_2): Likewise.
262 * final.c (alter_subreg): Likewise.
263 (leaf_renumber_regs_insn): Handle 'p'.
264 * function.c (assign_parm_find_stack_rtl, assign_parm_setup_stack):
265 Treat subreg offsets as poly_ints.
266 * fwprop.c (forward_propagate_and_simplify): Likewise.
267 * ifcvt.c (noce_emit_move_insn, noce_emit_cmove): Likewise.
268 * ira.c (get_subreg_tracking_sizes): Likewise.
269 * ira-conflicts.c (go_through_subreg): Likewise.
270 * ira-lives.c (process_single_reg_class_operands): Likewise.
271 * jump.c (rtx_renumbered_equal_p): Likewise. Handle 'p'.
272 * lower-subreg.c (simplify_subreg_concatn): Take the subreg offset
273 as a poly_uint64 rather than an unsigned int.
274 (simplify_gen_subreg_concatn, resolve_simple_move): Treat
275 subreg offsets as poly_ints.
276 * lra-constraints.c (operands_match_p): Handle 'p'.
277 (match_reload, curr_insn_transform): Treat subreg offsets as poly_ints.
278 * lra-spills.c (assign_mem_slot): Likewise.
279 * postreload.c (move2add_valid_value_p): Likewise.
280 * recog.c (general_operand, indirect_operand): Likewise.
281 * regcprop.c (copy_value, maybe_mode_change): Likewise.
282 (copyprop_hardreg_forward_1): Likewise.
283 * reginfo.c (simplifiable_subregs_hasher::hash, simplifiable_subregs)
284 (record_subregs_of_mode): Likewise.
285 * rtlhooks.c (gen_lowpart_general, gen_lowpart_if_possible): Likewise.
286 * reload.c (operands_match_p): Handle 'p'.
287 (find_reloads_subreg_address): Treat subreg offsets as poly_ints.
288 * reload1.c (alter_reg, choose_reload_regs): Likewise.
289 (compute_reload_subreg_offset): Likewise, and return an poly_int64.
290 * simplify-rtx.c (simplify_truncation, simplify_binary_operation_1):
291 (test_vector_ops_duplicate): Treat subreg offsets as poly_ints.
292 (simplify_const_poly_int_tests<N>::run): Likewise.
293 (simplify_subreg, simplify_gen_subreg): Take the subreg offset as
294 a poly_uint64 rather than an unsigned int.
295 * valtrack.c (debug_lowpart_subreg): Likewise.
296 * var-tracking.c (var_lowpart): Likewise.
297 (loc_cmp): Handle 'p'.
299 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
300 Alan Hayward <alan.hayward@arm.com>
301 David Sherwood <david.sherwood@arm.com>
303 * ira.c (get_subreg_tracking_sizes): New function.
304 (init_live_subregs): Take an integer size rather than a register.
305 (build_insn_chain): Use get_subreg_tracking_sizes. Update calls
306 to init_live_subregs.
308 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
309 Alan Hayward <alan.hayward@arm.com>
310 David Sherwood <david.sherwood@arm.com>
312 * expr.c (store_constructor_field): Change bitsize from a
313 unsigned HOST_WIDE_INT to a poly_uint64 and bitpos from a
314 HOST_WIDE_INT to a poly_int64.
315 (store_constructor): Change size from a HOST_WIDE_INT to
317 (store_field): Likewise bitsize and bitpos.
319 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
320 Alan Hayward <alan.hayward@arm.com>
321 David Sherwood <david.sherwood@arm.com>
323 * expmed.h (store_bit_field): Change bitregion_start and
324 bitregion_end from unsigned HOST_WIDE_INT to poly_uint64.
325 * expmed.c (adjust_bit_field_mem_for_reg, strict_volatile_bitfield_p)
326 (store_bit_field_1, store_integral_bit_field, store_bit_field)
327 (store_fixed_bit_field, store_split_bit_field): Likewise.
328 * expr.c (store_constructor_field, store_field): Likewise.
329 (optimize_bitfield_assignment_op): Likewise. Make the same change
330 to bitsize and bitpos.
331 * machmode.h (bit_field_mode_iterator): Change m_bitregion_start
332 and m_bitregion_end from HOST_WIDE_INT to poly_int64. Make the
333 same change in the constructor arguments.
334 (get_best_mode): Change bitregion_start and bitregion_end from
335 unsigned HOST_WIDE_INT to poly_uint64.
336 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
337 Change bitregion_start and bitregion_end from HOST_WIDE_INT to
339 (bit_field_mode_iterator::next_mode): Update for new types
340 of m_bitregion_start and m_bitregion_end.
341 (get_best_mode): Change bitregion_start and bitregion_end from
342 unsigned HOST_WIDE_INT to poly_uint64.
344 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
345 Alan Hayward <alan.hayward@arm.com>
346 David Sherwood <david.sherwood@arm.com>
348 * rtl.h (simplify_gen_subreg): Add a temporary overload that
349 accepts poly_uint64 offsets.
350 * expmed.h (extract_bit_field): Take bitsize and bitnum as
351 poly_uint64s rather than unsigned HOST_WIDE_INTs.
352 * expmed.c (lowpart_bit_field_p): Likewise.
353 (extract_bit_field_as_subreg): New function, split out from...
354 (extract_bit_field_1): ...here. Take bitsize and bitnum as
355 poly_uint64s rather than unsigned HOST_WIDE_INTs. For vector
356 extractions, check that BITSIZE matches the size of the extracted
357 value and that BITNUM is an exact multiple of that size.
358 If all else fails, try forcing the value into memory if
359 BITNUM is variable, and adjusting the address so that the
360 offset is constant. Split the part that can only handle constant
361 bitsize and bitnum out into...
362 (extract_integral_bit_field): ...this new function.
363 (extract_bit_field): Take bitsize and bitnum as poly_uint64s
364 rather than unsigned HOST_WIDE_INTs.
366 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
367 Alan Hayward <alan.hayward@arm.com>
368 David Sherwood <david.sherwood@arm.com>
370 * expmed.h (store_bit_field): Take bitsize and bitnum as
371 poly_uint64s rather than unsigned HOST_WIDE_INTs.
372 * expmed.c (simple_mem_bitfield_p): Likewise. Add a parameter
373 that returns the byte size.
374 (store_bit_field_1): Take bitsize and bitnum as
375 poly_uint64s rather than unsigned HOST_WIDE_INTs. Update call
376 to simple_mem_bitfield_p. Split the part that can only handle
377 constant bitsize and bitnum out into...
378 (store_integral_bit_field): ...this new function.
379 (store_bit_field): Take bitsize and bitnum as poly_uint64s rather
380 than unsigned HOST_WIDE_INTs.
381 (extract_bit_field_1): Update call to simple_mem_bitfield_p.
383 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
384 Alan Hayward <alan.hayward@arm.com>
385 David Sherwood <david.sherwood@arm.com>
387 * lra-int.h (lra_reg): Change offset from int to poly_int64.
388 (lra_insn_recog_data): Change sp_offset from HOST_WIDE_INT
390 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
391 update_sp_offset from a HOST_WIDE_INT to a poly_int64.
392 (lra_update_reg_val_offset, lra_reg_val_equal_p): Take the
393 offset as a poly_int64 rather than an int.
394 * lra-assigns.c (find_hard_regno_for_1): Handle poly_int64 offsets.
395 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
396 * lra-constraints.c (equiv_address_substitution): Track offsets
398 (emit_inc): Check poly_int_rtx_p instead of CONST_INT_P.
399 (curr_insn_transform): Handle the new form of sp_offset.
400 * lra-eliminations.c (lra_elim_table): Change previous_offset
401 and offset from HOST_WIDE_INT to poly_int64.
402 (print_elim_table, update_reg_eliminate): Update accordingly.
403 (self_elim_offsets): Change from HOST_WIDE_INT to poly_int64_pod.
404 (get_elimination): Update accordingly.
405 (form_sum): Check poly_int_rtx_p instead of CONST_INT_P.
406 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
407 update_sp_offset from a HOST_WIDE_INT to a poly_int64. Handle
408 poly_int64 offsets generally.
409 (curr_sp_change): Change from HOST_WIDE_INT to poly_int64.
410 (mark_not_eliminable, init_elimination): Update accordingly.
411 (remove_reg_equal_offset_note): Return a bool and pass the new
412 offset back by pointer as a poly_int64.
413 * lra-remat.c (change_sp_offset): Take sp_offset as a poly_int64
414 rather than a HOST_WIDE_INT.
415 (do_remat): Track offsets poly_int64s.
416 * lra.c (lra_update_insn_recog_data, setup_sp_offset): Likewise.
418 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
419 Alan Hayward <alan.hayward@arm.com>
420 David Sherwood <david.sherwood@arm.com>
422 * rtl.h (mem_attrs): Add a default constructor. Change size and
423 offset from HOST_WIDE_INT to poly_int64.
424 * emit-rtl.h (set_mem_offset, set_mem_size, adjust_address_1)
425 (adjust_automodify_address_1, set_mem_attributes_minus_bitpos)
426 (widen_memory_access): Take the sizes and offsets as poly_int64s
427 rather than HOST_WIDE_INTs.
428 * alias.c (ao_ref_from_mem): Handle the new form of MEM_OFFSET.
429 (offset_overlap_p): Take poly_int64s rather than HOST_WIDE_INTs
431 (adjust_offset_for_component_ref): Change the offset from a
432 HOST_WIDE_INT to a poly_int64.
433 (nonoverlapping_memrefs_p): Track polynomial offsets and sizes.
434 * cfgcleanup.c (merge_memattrs): Update after mem_attrs changes.
435 * dce.c (find_call_stack_args): Likewise.
436 * dse.c (record_store): Likewise.
437 * dwarf2out.c (tls_mem_loc_descriptor, dw_sra_loc_expr): Likewise.
438 * print-rtl.c (rtx_writer::print_rtx): Likewise.
439 * read-rtl-function.c (test_loading_mem): Likewise.
440 * rtlanal.c (may_trap_p_1): Likewise.
441 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
442 * var-tracking.c (int_mem_offset, track_expr_p): Likewise.
443 * emit-rtl.c (mem_attrs_eq_p, get_mem_align_offset): Likewise.
444 (mem_attrs::mem_attrs): New function.
445 (set_mem_attributes_minus_bitpos): Change bitpos from a
446 HOST_WIDE_INT to poly_int64.
447 (set_mem_alias_set, set_mem_addr_space, set_mem_align, set_mem_expr)
448 (clear_mem_offset, clear_mem_size, change_address)
449 (get_spill_slot_decl, set_mem_attrs_for_spill): Directly
450 initialize mem_attrs.
451 (set_mem_offset, set_mem_size, adjust_address_1)
452 (adjust_automodify_address_1, offset_address, widen_memory_access):
453 Likewise. Take poly_int64s rather than HOST_WIDE_INT.
455 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
456 Alan Hayward <alan.hayward@arm.com>
457 David Sherwood <david.sherwood@arm.com>
459 * rtlanal.c (rtx_addr_can_trap_p_1): Take the offset and size
460 as poly_int64s rather than HOST_WIDE_INTs. Use a size of -1
461 rather than 0 to represent an unknown size. Assert that the size
462 is known when the mode isn't BLKmode.
463 (may_trap_p_1): Use -1 for unknown sizes.
464 (rtx_addr_can_trap_p): Likewise. Pass BLKmode rather than VOIDmode.
466 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
467 Alan Hayward <alan.hayward@arm.com>
468 David Sherwood <david.sherwood@arm.com>
470 * dse.c (store_info): Change offset and width from HOST_WIDE_INT
471 to poly_int64. Update commentary for positions_needed.large.
472 (read_info_type): Change offset and width from HOST_WIDE_INT
474 (set_usage_bits): Likewise.
475 (canon_address): Return the offset as a poly_int64 rather than
476 a HOST_WIDE_INT. Use strip_offset_and_add.
477 (set_all_positions_unneeded, any_positions_needed_p): Use
478 positions_needed.large to track stores with non-constant widths.
479 (all_positions_needed_p): Likewise. Take the offset and width
480 as poly_int64s rather than ints. Assert that rhs is nonnull.
481 (record_store): Cope with non-constant offsets and widths.
482 Nullify the rhs of an earlier store if we can't tell which bytes
484 (find_shift_sequence): Take the access_size and shift as poly_int64s
486 (get_stored_val): Take the read_offset and read_width as poly_int64s
487 rather than HOST_WIDE_INTs.
488 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Handle
489 non-constant offsets and widths.
491 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
492 Alan Hayward <alan.hayward@arm.com>
493 David Sherwood <david.sherwood@arm.com>
495 * inchash.h (inchash::hash::add_poly_int): New function.
496 * tree-ssa-alias.h (ao_ref::offset, ao_ref::size, ao_ref::max_size):
497 Use poly_int64 rather than HOST_WIDE_INT.
498 (ao_ref::max_size_known_p): New function.
499 * tree-ssa-sccvn.h (vn_reference_op_struct::off): Use poly_int64_pod
500 rather than HOST_WIDE_INT.
501 * tree-ssa-alias.c (ao_ref_base): Apply get_ref_base_and_extent
502 to temporaries until its interface is adjusted to match.
503 (ao_ref_init_from_ptr_and_size): Handle polynomial offsets and sizes.
504 (aliasing_component_refs_p, decl_refs_may_alias_p)
505 (indirect_ref_may_alias_decl_p, indirect_refs_may_alias_p): Take
506 the offsets and max_sizes as poly_int64s instead of HOST_WIDE_INTs.
507 (refs_may_alias_p_1, stmt_kills_ref_p): Adjust for changes to
509 * alias.c (ao_ref_from_mem): Likewise.
510 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
511 * tree-ssa-dse.c (valid_ao_ref_for_dse, normalize_ref)
512 (clear_bytes_written_by, setup_live_bytes_from_ref, compute_trims)
513 (maybe_trim_complex_store, maybe_trim_constructor_store)
514 (live_bytes_read, dse_classify_store): Likewise.
515 * tree-ssa-sccvn.c (vn_reference_compute_hash, vn_reference_eq):
516 (copy_reference_ops_from_ref, ao_ref_init_from_vn_reference)
517 (fully_constant_vn_reference_p, valueize_refs_1): Likewise.
518 (vn_reference_lookup_3): Likewise.
519 * tree-ssa-uninit.c (warn_uninitialized_vars): Likewise.
521 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
522 Alan Hayward <alan.hayward@arm.com>
523 David Sherwood <david.sherwood@arm.com>
525 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p)
526 (indirect_refs_may_alias_p): Use ranges_may_overlap_p
527 instead of ranges_overlap_p.
529 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
530 Alan Hayward <alan.hayward@arm.com>
531 David Sherwood <david.sherwood@arm.com>
533 * tree-ssa-alias.c (same_addr_size_stores_p): Take the offsets and
534 sizes as poly_int64s rather than HOST_WIDE_INTs.
536 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
537 Alan Hayward <alan.hayward@arm.com>
538 David Sherwood <david.sherwood@arm.com>
540 * gimple-fold.h (fold_ctor_reference): Take the offset and size
541 as poly_uint64 rather than unsigned HOST_WIDE_INT.
542 * gimple-fold.c (fold_ctor_reference): Likewise.
544 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
545 Alan Hayward <alan.hayward@arm.com>
546 David Sherwood <david.sherwood@arm.com>
548 * target.def (dwarf_poly_indeterminate_value): New hook.
549 * targhooks.h (default_dwarf_poly_indeterminate_value): Declare.
550 * targhooks.c (default_dwarf_poly_indeterminate_value): New function.
551 * doc/tm.texi.in (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Document.
552 * doc/tm.texi: Regenerate.
553 * dwarf2out.h (build_cfa_loc, build_cfa_aligned_loc): Take the
554 offset as a poly_int64.
555 * dwarf2out.c (new_reg_loc_descr): Move later in file. Take the
556 offset as a poly_int64.
557 (loc_descr_plus_const, loc_list_plus_const, build_cfa_aligned_loc):
558 Take the offset as a poly_int64.
559 (build_cfa_loc): Likewise. Use loc_descr_plus_const.
560 (frame_pointer_fb_offset): Change to a poly_int64.
561 (int_loc_descriptor): Take the offset as a poly_int64. Use
562 targetm.dwarf_poly_indeterminate_value for polynomial offsets.
563 (based_loc_descr): Take the offset as a poly_int64.
564 Use strip_offset_and_add to handle (plus X (const)).
565 Use new_reg_loc_descr instead of an open-coded version of the
566 previous implementation.
567 (mem_loc_descriptor): Handle CONST_POLY_INT.
568 (compute_frame_pointer_to_fb_displacement): Take the offset as a
569 poly_int64. Use strip_offset_and_add to handle (plus X (const)).
571 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
572 Alan Hayward <alan.hayward@arm.com>
573 David Sherwood <david.sherwood@arm.com>
575 * rtl.h (reg_attrs::offset): Change from HOST_WIDE_INT to poly_int64.
576 (gen_rtx_REG_offset): Take the offset as a poly_int64.
577 * inchash.h (inchash::hash::add_poly_hwi): New function.
578 * gengtype.c (main): Register poly_int64.
579 * emit-rtl.c (reg_attr_hasher::hash): Use inchash. Treat the
580 offset as a poly_int.
581 (reg_attr_hasher::equal): Use must_eq to compare offsets.
582 (get_reg_attrs, update_reg_offset, gen_rtx_REG_offset): Take the
583 offset as a poly_int64.
584 (set_reg_attrs_from_value): Treat the offset as a poly_int64.
585 * print-rtl.c (print_poly_int): New function.
586 (rtx_writer::print_rtx_operand_code_r): Treat REG_OFFSET as
588 * var-tracking.c (track_offset_p, get_tracked_reg_offset): New
590 (var_reg_set, var_reg_delete_and_set, var_reg_delete): Use them.
591 (same_variable_part_p, track_loc_p): Take the offset as a poly_int64.
592 (vt_get_decl_and_offset): Return the offset as a poly_int64.
593 Enforce track_offset_p for parts of a PARALLEL.
594 (vt_add_function_parameter): Use const_offset for the final
595 offset to track. Use get_tracked_reg_offset for the parts
598 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
599 Alan Hayward <alan.hayward@arm.com>
600 David Sherwood <david.sherwood@arm.com>
602 * target.def (truly_noop_truncation): Take poly_uint64s instead of
603 unsigned ints. Change default to hook_bool_puint64_puint64_true.
604 * doc/tm.texi: Regenerate.
605 * hooks.h (hook_bool_uint_uint_true): Delete.
606 (hook_bool_puint64_puint64_true): Declare.
607 * hooks.c (hook_bool_uint_uint_true): Delete.
608 (hook_bool_puint64_puint64_true): New function.
609 * config/mips/mips.c (mips_truly_noop_truncation): Take poly_uint64s
610 instead of unsigned ints.
611 * config/spu/spu.c (spu_truly_noop_truncation): Likewise.
612 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): Likewise.
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 * optabs.h (expand_operand): Add an int_value field.
619 (create_expand_operand): Add an int_value parameter and use it
620 to initialize the new expand_operand field.
621 (create_integer_operand): Replace with a declaration of a function
622 that accepts poly_int64s. Move the implementation to...
623 * optabs.c (create_integer_operand): ...here.
624 (maybe_legitimize_operand): For EXPAND_INTEGER, check whether
625 the mode preserves the value of int_value, instead of calling
626 const_int_operand on the rtx. Use gen_int_mode to generate
629 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
630 Alan Hayward <alan.hayward@arm.com>
631 David Sherwood <david.sherwood@arm.com>
633 * dumpfile.h (dump_dec): Declare.
634 * dumpfile.c (dump_dec): New function.
635 * pretty-print.h (pp_wide_integer): Turn into a function and
636 declare a poly_int version.
637 * pretty-print.c (pp_wide_integer): New function for poly_ints.
639 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
640 Alan Hayward <alan.hayward@arm.com>
641 David Sherwood <david.sherwood@arm.com>
643 * doc/generic.texi (POLY_INT_CST): Document.
644 * tree.def (POLY_INT_CST): New tree code.
645 * treestruct.def (TS_POLY_INT_CST): New tree layout.
646 * tree-core.h (tree_poly_int_cst): New struct.
647 (tree_node): Add a poly_int_cst field.
648 * tree.h (POLY_INT_CST_P, POLY_INT_CST_COEFF): New macros.
649 (wide_int_to_tree, force_fit_type): Take a poly_wide_int_ref
650 instead of a wide_int_ref.
651 (build_int_cst, build_int_cst_type): Take a poly_int64 instead
653 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
654 instead of an unsigned HOST_WIDE_INT.
655 (build_poly_int_cst, tree_fits_poly_int64_p, tree_fits_poly_uint64_p)
656 (ptrdiff_tree_p): Declare.
657 (tree_to_poly_int64, tree_to_poly_uint64): Likewise. Provide
658 extern inline implementations if the target doesn't use POLY_INT_CST.
659 (poly_int_tree_p): New function.
660 (wi::unextended_tree): New class.
661 (wi::int_traits <unextended_tree>): New override.
662 (wi::extended_tree): Add a default constructor.
663 (wi::extended_tree::get_tree): New function.
664 (wi::widest_extended_tree, wi::offset_extended_tree): New typedefs.
665 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): Use them.
666 (wi::tree_to_poly_widest_ref, wi::tree_to_poly_offset_ref)
667 (wi::tree_to_poly_wide_ref): New typedefs.
668 (wi::ints_for): Provide overloads for extended_tree and
670 (poly_int_cst_value, wi::to_poly_widest, wi::to_poly_offset)
671 (wi::to_wide): New functions.
672 (wi::fits_to_boolean_p, wi::fits_to_tree_p): Handle poly_ints.
673 * tree.c (poly_int_cst_hasher): New struct.
674 (poly_int_cst_hash_table): New variable.
675 (tree_node_structure_for_code, tree_code_size, simple_cst_equal)
676 (valid_constant_size_p, add_expr, drop_tree_overflow): Handle
678 (initialize_tree_contains_struct): Handle TS_POLY_INT_CST.
679 (init_ttree): Initialize poly_int_cst_hash_table.
680 (build_int_cst, build_int_cst_type, build_invariant_address): Take
681 a poly_int64 instead of a HOST_WIDE_INT.
682 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
683 instead of an unsigned HOST_WIDE_INT.
684 (wide_int_to_tree): Rename to...
685 (wide_int_to_tree_1): ...this.
686 (build_new_poly_int_cst, build_poly_int_cst): New functions.
687 (force_fit_type): Take a poly_wide_int_ref instead of a wide_int_ref.
688 (wide_int_to_tree): New function that takes a poly_wide_int_ref.
689 (ptrdiff_tree_p, tree_to_poly_int64, tree_to_poly_uint64)
690 (tree_fits_poly_int64_p, tree_fits_poly_uint64_p): New functions.
691 * lto-streamer-out.c (DFS::DFS_write_tree_body, hash_tree): Handle
693 * tree-streamer-in.c (lto_input_ts_poly_tree_pointers): Likewise.
694 (streamer_read_tree_body): Likewise.
695 * tree-streamer-out.c (write_ts_poly_tree_pointers): Likewise.
696 (streamer_write_tree_body): Likewise.
697 * tree-streamer.c (streamer_check_handled_ts_structures): Likewise.
698 * asan.c (asan_protect_global): Require the size to be an INTEGER_CST.
699 * cfgexpand.c (expand_debug_expr): Handle POLY_INT_CST.
700 * expr.c (expand_expr_real_1, const_vector_from_tree): Likewise.
701 * gimple-expr.h (is_gimple_constant): Likewise.
702 * gimplify.c (maybe_with_size_expr): Likewise.
703 * print-tree.c (print_node): Likewise.
704 * tree-data-ref.c (data_ref_compare_tree): Likewise.
705 * tree-pretty-print.c (dump_generic_node): Likewise.
706 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
707 * tree-vect-data-refs.c (dr_group_sort_cmp): Likewise.
708 * tree-vrp.c (compare_values_warnv): Likewise.
709 * tree-ssa-loop-ivopts.c (determine_base_object, constant_multiple_of)
710 (get_loop_invariant_expr, add_candidate_1, get_computation_aff_1)
711 (force_expr_to_var_cost): Likewise.
712 * tree-ssa-loop.c (for_each_index): Likewise.
713 * fold-const.h (build_invariant_address, size_int_kind): Take a
714 poly_int64 instead of a HOST_WIDE_INT.
715 * fold-const.c (fold_negate_expr_1, const_binop, const_unop)
716 (fold_convert_const, multiple_of_p, fold_negate_const): Handle
718 (size_binop_loc): Likewise. Allow int_const_binop_1 to fail.
719 (int_const_binop_2): New function, split out from...
720 (int_const_binop_1): ...here. Handle POLY_INT_CST.
721 (size_int_kind): Take a poly_int64 instead of a HOST_WIDE_INT.
722 * expmed.c (make_tree): Handle CONST_POLY_INT_P.
723 * gimple-ssa-strength-reduction.c (slsr_process_add)
724 (slsr_process_mul): Check for INTEGER_CSTs before using them
726 * stor-layout.c (bits_from_bytes): New function.
727 (bit_from_pos): Use it.
728 (layout_type): Likewise. For vectors, multiply the TYPE_SIZE_UNIT
729 by BITS_PER_UNIT to get the TYPE_SIZE.
730 * tree-cfg.c (verify_expr, verify_types_in_gimple_reference): Allow
731 MEM_REF and TARGET_MEM_REF offsets to be a POLY_INT_CST.
733 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
734 Alan Hayward <alan.hayward@arm.com>
735 David Sherwood <david.sherwood@arm.com>
737 * doc/rtl.texi (const_poly_int): Document. Also document the
738 rtl sharing behavior.
739 * gengenrtl.c (excluded_rtx): Return true for CONST_POLY_INT.
740 * rtl.h (const_poly_int_def): New struct.
741 (rtx_def::u): Add a cpi field.
742 (CASE_CONST_UNIQUE, CASE_CONST_ANY): Add CONST_POLY_INT.
743 (CONST_POLY_INT_P, CONST_POLY_INT_COEFFS): New macros.
744 (wi::rtx_to_poly_wide_ref): New typedef
745 (const_poly_int_value, wi::to_poly_wide, rtx_to_poly_int64)
746 (poly_int_rtx_p): New functions.
747 (trunc_int_for_mode): Declare a poly_int64 version.
748 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
749 (immed_wide_int_const): Take a poly_wide_int_ref rather than
751 (strip_offset): Declare.
752 (strip_offset_and_add): New function.
753 * rtl.def (CONST_POLY_INT): New rtx code.
754 * rtl.c (rtx_size): Handle CONST_POLY_INT.
755 (shared_const_p): Use poly_int_rtx_p.
756 * emit-rtl.h (gen_int_mode): Take a poly_int64 instead of a
758 (gen_int_shift_amount): Likewise.
759 * emit-rtl.c (const_poly_int_hasher): New class.
760 (const_poly_int_htab): New variable.
761 (init_emit_once): Initialize it when NUM_POLY_INT_COEFFS > 1.
762 (const_poly_int_hasher::hash): New function.
763 (const_poly_int_hasher::equal): Likewise.
764 (gen_int_mode): Take a poly_int64 instead of a HOST_WIDE_INT.
765 (immed_wide_int_const): Rename to...
766 (immed_wide_int_const_1): ...this and make static.
767 (immed_wide_int_const): New function, taking a poly_wide_int_ref
768 instead of a wide_int_ref.
769 (gen_int_shift_amount): Take a poly_int64 instead of a HOST_WIDE_INT.
770 (gen_lowpart_common): Handle CONST_POLY_INT.
771 * cse.c (hash_rtx_cb, equiv_constant): Likewise.
772 * cselib.c (cselib_hash_rtx): Likewise.
773 * dwarf2out.c (const_ok_for_output_1): Likewise.
774 * expr.c (convert_modes): Likewise.
775 * print-rtl.c (rtx_writer::print_rtx, print_value): Likewise.
776 * rtlhash.c (add_rtx): Likewise.
777 * explow.c (trunc_int_for_mode): Add a poly_int64 version.
778 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
779 Handle existing CONST_POLY_INT rtxes.
780 * expmed.h (expand_shift): Take a poly_int64 instead of a
782 * expmed.c (expand_shift): Likewise.
783 * rtlanal.c (strip_offset): New function.
784 (commutative_operand_precedence): Give CONST_POLY_INT the same
785 precedence as CONST_DOUBLE and put CONST_WIDE_INT between that
787 * rtl-tests.c (const_poly_int_tests): New struct.
788 (rtl_tests_c_tests): Use it.
789 * simplify-rtx.c (simplify_const_unary_operation): Handle
791 (simplify_const_binary_operation): Likewise.
792 (simplify_binary_operation_1): Fold additions of symbolic constants
794 (simplify_subreg): Handle extensions and truncations of
796 (simplify_const_poly_int_tests): New struct.
797 (simplify_rtx_c_tests): Use it.
798 * wide-int.h (storage_ref): Add default constructor.
799 (wide_int_ref_storage): Likewise.
800 (trailing_wide_ints): Use GTY((user)).
801 (trailing_wide_ints::operator[]): Add a const version.
802 (trailing_wide_ints::get_precision): New function.
803 (trailing_wide_ints::extra_size): Likewise.
805 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
806 Alan Hayward <alan.hayward@arm.com>
807 David Sherwood <david.sherwood@arm.com>
809 * emit-rtl.h (gen_int_shift_amount): Declare.
810 * emit-rtl.c (gen_int_shift_amount): New function.
811 * asan.c (asan_emit_stack_protection): Use gen_int_shift_amount
813 * calls.c (shift_return_value): Likewise.
814 * cse.c (fold_rtx): Likewise.
815 * dse.c (find_shift_sequence): Likewise.
816 * expmed.c (init_expmed_one_mode, store_bit_field_1, expand_shift_1)
817 (expand_shift, expand_smod_pow2): Likewise.
818 * lower-subreg.c (shift_cost): Likewise.
819 * optabs.c (expand_superword_shift, expand_doubleword_mult)
820 (expand_unop, expand_binop, shift_amt_for_vec_perm_mask)
821 (expand_vec_perm_var): Likewise.
822 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
823 (simplify_binary_operation_1): Likewise.
824 * combine.c (try_combine, find_split_point, force_int_to_mode)
825 (simplify_shift_const_1, simplify_shift_const): Likewise.
826 (change_zero_ext): Likewise. Use simplify_gen_binary.
828 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
830 * poly-int.h (multiple_p): Fix handling of two non-poly_ints.
832 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
834 * doc/invoke.texi (ARM Options): Document accepted extension options
835 for -march=armv8.3-a.
837 2017-12-20 Richard Earnshaw <rearnsha@arm.com>
840 * config.gcc (arm*-*-linux*): When configured with --with-float=hard
841 or --with-float=softfp, set the default CPU to arm10e.
843 2017-12-20 Eric Botcazou <ebotcazou@adacore.com>
845 * config/visium/constraints.md (J, K, L): Use IN_RANGE macro.
846 * config/visium/predicates.md (const_shift_operand): Likewise.
847 * config/visium/visium.c (visium_legitimize_address): Fix oversight.
848 (visium_legitimize_reload_address): Likewise.
850 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
853 * config/arm/arm.h (TEST_REGNO): Adjust comment as expected in r255830.
855 2017-12-20 Jakub Jelinek <jakub@redhat.com>
858 * calls.c (compute_argument_addresses): Ignore TYPE_EMPTY_P arguments.
860 2017-12-20 Julia Koval <julia.koval@intel.com>
862 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VPCLMULQDQ_SET,
863 OPTION_MASK_ISA_VPCLMULQDQ_UNSET): New.
864 (ix86_handle_option): Handle -mvpclmulqdq, move cx6 to flags2.
865 * config.gcc: Include vpclmulqdqintrin.h.
866 * config/i386/cpuid.h: Handle bit_VPCLMULQDQ.
867 * config/i386/driver-i386.c (host_detect_local_cpu): Handle
869 * config/i386/i386-builtin.def (__builtin_ia32_vpclmulqdq_v2di,
870 __builtin_ia32_vpclmulqdq_v4di, __builtin_ia32_vpclmulqdq_v8di): New.
871 * config/i386/i386-c.c (__VPCLMULQDQ__): New.
872 * config/i386/i386.c (isa2_opts): Add -mcx16.
873 (isa_opts): Add -mpclmulqdq, remove -mcx16.
874 (ix86_option_override_internal): Move mcx16 to flags2.
875 (ix86_valid_target_attribute_inner_p): Add vpclmulqdq.
876 (ix86_expand_builtin): Handle OPTION_MASK_ISA_VPCLMULQDQ.
877 * config/i386/i386.h (TARGET_VPCLMULQDQ, TARGET_VPCLMULQDQ_P): New.
878 * config/i386/i386.opt: Add mvpclmulqdq, move mcx16 to flags2.
879 * config/i386/immintrin.h: Include vpclmulqdqintrin.h.
880 * config/i386/sse.md (vpclmulqdq_<mode>): New pattern.
881 * config/i386/vpclmulqdqintrin.h (_mm512_clmulepi64_epi128,
882 _mm_clmulepi64_epi128, _mm256_clmulepi64_epi128): New intrinsics.
883 * doc/invoke.texi: Add -mvpclmulqdq.
885 2017-12-20 Tom de Vries <tom@codesourcery.com>
888 * config/i386/i386.c (ix86_static_chain): Move DECL_STATIC_CHAIN test ...
889 * calls.c (rtx_for_static_chain): ... here. New function.
890 * calls.h (rtx_for_static_chain): Declare.
891 * builtins.c (expand_builtin_setjmp_receiver): Use rtx_for_static_chain
892 instead of targetm.calls.static_chain.
893 * df-scan.c (df_get_entry_block_def_set): Same.
895 2017-12-19 Tom de Vries <tom@codesourcery.com>
897 PR tree-optimization/83493
898 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for): Unshare ub
901 2017-12-19 Jakub Jelinek <jakub@redhat.com>
903 * gimple-ssa-sprintf.c (format_directive): Use inform_n instead of
904 inform with hardcoded english plural handling.
906 2017-12-18 Jeff Law <law@redhat.com>
908 PR tree-optimization/83477
909 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis): For
910 a non-virtual PHI, always push a new range.
912 2017-12-19 Martin Sebor <msebor@redhat.com>
915 * builtins.c (compute_objsize): Handle non-constant offsets.
917 2017-12-19 Jakub Jelinek <jakub@redhat.com>
919 PR tree-optimization/83444
920 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): For the
921 character load case, if get_stridx on MEM_REF's operand doesn't
922 look usable, retry with get_addr_stridx.
924 2017-12-19 Alexandre Oliva <aoliva@redhat.com>
927 * var-tracking.c (vt_debug_insns_local): Do not drop markers.
928 (variable_tracking_main_1): Keep markers even when VTA fails.
931 * cfgexpand.c (expand_gimple_basic_block): Expand label first,
932 even if there are markers before it.
933 * cfgrtl.c (rtl_verify_bb_layout): Reject DEBUG_INSNs outside BBs.
935 2017-12-19 Jakub Jelinek <jakub@redhat.com>
937 * read-rtl.c (parse_reg_note_name): Replace Yoda conditions with
938 typical order conditions.
939 * sel-sched.c (extract_new_fences_from): Likewise.
940 * config/visium/constraints.md (J, K, L): Likewise.
941 * config/visium/predicates.md (const_shift_operand): Likewise.
942 * config/visium/visium.c (visium_legitimize_address,
943 visium_legitimize_reload_address): Likewise.
944 * config/m68k/m68k.c (output_reg_adjust, emit_reg_adjust): Likewise.
945 * config/arm/arm.c (arm_block_move_unaligned_straight): Likewise.
946 * config/avr/constraints.md (Y01, Ym1, Y02, Ym2): Likewise.
947 * config/avr/avr-log.c (avr_vdump, avr_log_set_avr_log,
948 SET_DUMP_DETAIL): Likewise.
949 * config/avr/predicates.md (const_8_16_24_operand): Likewise.
950 * config/avr/avr.c (STR_PREFIX_P, avr_popcount_each_byte,
951 avr_is_casesi_sequence, avr_casei_sequence_check_operands,
952 avr_set_core_architecture, avr_set_current_function,
953 avr_legitimize_reload_address, avr_asm_len, avr_print_operand,
954 output_movqi, output_movsisf, avr_out_plus, avr_out_bitop,
955 avr_out_fract, avr_adjust_insn_length, avr_encode_section_info,
956 avr_2word_insn_p, output_reload_in_const, avr_has_nibble_0xf,
957 avr_map_decompose, avr_fold_builtin): Likewise.
958 * config/avr/driver-avr.c (avr_devicespecs_file): Likewise.
959 * config/avr/gen-avr-mmcu-specs.c (str_prefix_p, print_mcu): Likewise.
960 * config/i386/i386.c (ix86_parse_stringop_strategy_string): Likewise.
961 * config/m32c/m32c-pragma.c (m32c_pragma_memregs): Likewise.
962 * config/m32c/m32c.c (m32c_conditional_register_usage,
963 m32c_address_cost): Likewise.
964 * config/m32c/predicates.md (shiftcount_operand,
965 longshiftcount_operand): Likewise.
966 * config/iq2000/iq2000.c (iq2000_expand_prologue): Likewise.
967 * config/nios2/nios2.c (nios2_handle_custom_fpu_insn_option,
968 can_use_cdx_ldstw): Likewise.
969 * config/nios2/nios2.h (CDX_REG_P): Likewise.
970 * config/cr16/cr16.h (RETURN_ADDR_RTX, REGNO_MODE_OK_FOR_BASE_P):
972 * config/cr16/cr16.md (*mov<mode>_double): Likewise.
973 * config/cr16/cr16.c (cr16_create_dwarf_for_multi_push): Likewise.
974 * config/h8300/h8300.c (h8300_rtx_costs, get_shift_alg): Likewise.
975 * config/vax/constraints.md (U06, U08, U16, CN6, S08, S16): Likewise.
976 * config/vax/vax.c (adjacent_operands_p): Likewise.
977 * config/ft32/constraints.md (L, b, KA): Likewise.
978 * config/ft32/ft32.c (ft32_load_immediate, ft32_expand_prologue):
980 * cfgexpand.c (expand_stack_alignment): Likewise.
981 * gcse.c (insert_expr_in_table): Likewise.
982 * print-rtl.c (rtx_writer::print_rtx_operand_codes_E_and_V): Likewise.
983 * cgraphunit.c (cgraph_node::expand): Likewise.
984 * ira-build.c (setup_min_max_allocno_live_range_point): Likewise.
985 * emit-rtl.c (add_insn): Likewise.
986 * input.c (dump_location_info): Likewise.
987 * passes.c (NEXT_PASS): Likewise.
988 * read-rtl-function.c (parse_note_insn_name,
989 function_reader::read_rtx_operand_r, function_reader::parse_mem_expr):
991 * sched-rgn.c (sched_rgn_init): Likewise.
992 * diagnostic-show-locus.c (layout::show_ruler): Likewise.
993 * combine.c (find_split_point, simplify_if_then_else, force_to_mode,
994 if_then_else_cond, simplify_shift_const_1, simplify_comparison): Likewise.
995 * explow.c (eliminate_constant_term): Likewise.
996 * final.c (leaf_renumber_regs_insn): Likewise.
997 * cfgrtl.c (print_rtl_with_bb): Likewise.
998 * genhooks.c (emit_init_macros): Likewise.
999 * poly-int.h (maybe_ne, maybe_le, maybe_lt): Likewise.
1000 * tree-data-ref.c (conflict_fn): Likewise.
1001 * selftest.c (assert_streq): Likewise.
1002 * expr.c (store_constructor_field, expand_expr_real_1): Likewise.
1003 * fold-const.c (fold_range_test, extract_muldiv_1, fold_truth_andor,
1004 fold_binary_loc, multiple_of_p): Likewise.
1005 * reload.c (push_reload, find_equiv_reg): Likewise.
1006 * et-forest.c (et_nca, et_below): Likewise.
1007 * dbxout.c (dbxout_symbol_location): Likewise.
1008 * reorg.c (relax_delay_slots): Likewise.
1009 * dojump.c (do_compare_rtx_and_jump): Likewise.
1010 * gengtype-parse.c (type): Likewise.
1011 * simplify-rtx.c (simplify_gen_ternary, simplify_gen_relational,
1012 simplify_const_relational_operation): Likewise.
1013 * reload1.c (do_output_reload): Likewise.
1014 * dumpfile.c (get_dump_file_info_by_switch): Likewise.
1015 * gengtype.c (type_for_name): Likewise.
1016 * gimple-ssa-sprintf.c (format_directive): Likewise.
1018 2017-12-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1021 * config/arm/arm.h (TEST_REGNO): Check reg_renumber is set before
1022 accessing it. Adjust comment.
1024 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1027 * predict.c (zero_one_minusone): New function.
1028 (apply_return_prediction): Avoid return prediction for functions
1029 returning only -1, 0 and 1 values, unless they only return -1 and 0
1032 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
1034 * config/arc/arc.c (legitimate_scaled_address_p): Clean
1035 fall-through warning.
1036 (arc_compute_frame_size): Remove unused variables.
1037 (arc_print_operand): Fix fprintif format.
1038 (arc_can_follow_jump): Clean fall-through warning.
1040 2017-12-19 Marek Polacek <polacek@redhat.com>
1043 * config/i386/i386.c (init_cumulative_args): Don't check TYPE_EMPTY_P
1046 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
1048 * config/arc/arc.c (overriderregs): New variable.
1049 (arc_override_options): Track fixed/call saved/call options.
1050 (arc_conditional_register_usage): Check against overrideregs
1051 variable whenever we change register properties.
1053 2017-12-19 Nathan Sidwell <nathan@acm.org>
1055 * opts.c (finish_options): Don't prefix dump_base_name if it
1056 already contains directories.
1058 2017-12-19 Martin Liska <mliska@suse.cz>
1060 PR rtl-optimization/82675
1061 * loop-unroll.c (unroll_loop_constant_iterations): Allocate one
1062 more element in sbitmap.
1064 2017-12-19 Paolo Carlini <paolo.carlini@oracle.com>
1066 * gimplify.c (gimplify_expr): Use error_operand_p.
1068 2017-12-19 Sebastian Huber <sebastian.huber@embedded-brains.de>
1071 * config/rs6000/t-rtems (MULTILIB_REQUIRED): Remove 64-bit soft-float
1074 2017-12-19 Daniel Cederman <cederman@gaisler.com>
1076 * config/sparc/sparc.c (sparc_do_work_around_errata): Make sure
1077 the jump is to a label.
1079 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1081 PR tree-optimization/83444
1082 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): Optimize
1087 * ipa-inline.c (flatten_remove_node_hook): New function.
1088 (ipa_inline): Keep only nodes with flatten attribute at the end of
1089 the array in the order from ipa_reverse_postorder, only walk that
1090 portion of array for flattening, if there is more than one such
1091 node, temporarily register a removal hook and ignore removed nodes.
1093 PR tree-optimization/80631
1094 * tree-vect-loop.c (vect_create_epilog_for_reduction): Compare
1095 induc_code against MAX_EXPR or MIN_EXPR instead of reduc_fn against
1096 IFN_REDUC_MAX or IFN_REDUC_MIN.
1098 2017-12-18 Jeff Law <law@redhat.com>
1100 * tree-ssa-dom.c (record_equivalences_from_phis): Do not
1101 record symbolic equivalences from backedges in the CFG.
1104 2017-11-19 Jeff Law <law@redhat.com>
1106 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
1107 of degenerates resulting from ignoring an edge.
1109 2017-12-18 Martin Sebor <msebor@redhat.com>
1112 PR tree-optimization/78450
1113 * tree-ssa-strlen.c (maybe_set_strlen_range): New function.
1114 (handle_builtin_strlen): Call it.
1116 2017-12-18 Segher Boessenkool <segher@kernel.crashing.org>
1118 PR rtl-optimization/83424
1119 * rtlanal.c (dead_or_set_regno_p): Handle CLOBBER just like SET.
1121 2017-12-18 Marek Polacek <polacek@redhat.com>
1124 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
1125 Check if TYPE is INTEGRAL_TYPE_P before accessing its min/max
1128 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
1130 * config/arc/arc.md (maddsidi4, maddsidi4_split): Update pattern.
1131 (umaddsidi4, umaddsidi_split): Likewise.
1133 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
1135 * config/arc/arc.c (arc_legitimate_constant_p): Always check all
1138 2017-12-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1141 * config/s390/s390.c (s390_option_override): Avoid strncpy.
1143 2017-12-18 Richard Biener <rguenther@suse.de>
1145 PR tree-optimization/81877
1146 * tree-ssa-loop-im.c (ref_indep_loop_p): Remove safelen parameters.
1147 (outermost_indep_loop): Adjust.
1148 (ref_indep_loop_p_1): Likewise. Remove safelen handling again.
1149 (can_sm_ref_p): Adjust.
1151 2017-12-18 Richard Biener <rguenther@suse.de>
1154 * tree.c (array_at_struct_end_p): Return true if the underlying
1155 object has space for at least one element in excess of what
1156 the array domain specifies.
1158 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
1160 * doc/extend.texi (x86 Function Attributes): Reformat nocf_check
1161 example to avoid overfull hbox.
1162 * doc/invoke.texi (Option Summary): Add missing @gol.
1163 (C++ Dialect Options): Reformat -Wnoexcept-type example to avoid
1166 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
1167 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1169 * doc/invoke.texi (Option Summary): Add -mverbose-cost-dump
1170 to AArch64 and ARM lists, plus missing -mflip-thumb for ARM.
1171 (AArch64 Options): Document -mverbose-cost-dump.
1172 (ARM Options): Likewise, plus -mflip-thumb.
1174 2017-12-17 Martin Sebor <msebor@redhat.com>
1177 * gimple-ssa-warn-restrict.c
1178 (builtin_memref::offset_out_of_bounds): Correct the handling of
1181 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
1183 * doc/invoke.texi (ARC Options): Add missing -mlra entry.
1185 2017-12-17 John David Anglin <danglin@gcc.gnu.org>
1187 * config/pa/pa.c (pa_som_asm_init_sections): Fix comment.
1189 2017-12-17 Markus Trippelsdorf <markus@trippelsdorf.de>
1191 * x86-tune-costs.h (skylake_cost, core_cost): Decrease r64 multiply
1194 2017-12-16 Sandra Loosemore <sandra@codesourcery.com>
1196 * doc/invoke.texi: Fix some typos.
1198 2017-12-16 Martin Sebor <msebor@redhat.com>
1200 PR tree-optimization/78918
1201 * Makefile.in (OBJS): Add gimple-ssa-warn-restrict.o.
1202 * builtins.c (check_sizes): Rename...
1203 (check_access): ...to this. Rename function arguments for clarity.
1204 (check_memop_sizes): Adjust names.
1205 (expand_builtin_memchr, expand_builtin_memcpy): Same.
1206 (expand_builtin_memmove, expand_builtin_mempcpy): Same.
1207 (expand_builtin_strcat, expand_builtin_stpncpy): Same.
1208 (check_strncat_sizes, expand_builtin_strncat): Same.
1209 (expand_builtin_strncpy, expand_builtin_memset): Same.
1210 (expand_builtin_bzero, expand_builtin_memcmp): Same.
1211 (expand_builtin_memory_chk, maybe_emit_chk_warning): Same.
1212 (maybe_emit_sprintf_chk_warning): Same.
1213 (expand_builtin_strcpy): Adjust.
1214 (expand_builtin_stpcpy): Same.
1215 (expand_builtin_with_bounds): Detect out-of-bounds accesses
1216 in pointer-checking forms of memcpy, memmove, and mempcpy.
1217 (gcall_to_tree_minimal, max_object_size): Define new functions.
1218 * builtins.h (max_object_size): Declare.
1219 * calls.c (alloc_max_size): Call max_object_size instead of
1220 hardcoding ssizetype limit.
1221 (get_size_range): Handle new argument.
1222 * calls.h (get_size_range): Add a new argument.
1223 * cfgexpand.c (expand_call_stmt): Propagate no-warning bit.
1224 * doc/invoke.texi (-Wrestrict): Adjust, add example.
1225 * gimple-fold.c (gimple_fold_builtin_memory_op): Detect overlapping
1227 (gimple_fold_builtin_memory_chk): Same.
1228 (gimple_fold_builtin_stxcpy_chk): New function.
1229 * gimple-ssa-warn-restrict.c: New source.
1230 * gimple-ssa-warn-restrict.h: New header.
1231 * gimple.c (gimple_build_call_from_tree): Propagate location.
1232 * passes.def (pass_warn_restrict): Add new pass.
1233 * tree-pass.h (make_pass_warn_restrict): Declare.
1234 * tree-ssa-strlen.c (handle_builtin_strcpy): Detect overlapping
1236 (handle_builtin_strcat): Same.
1237 (strlen_optimize_stmt): Rename...
1238 (strlen_check_and_optimize_stmt): ...to this. Handle strncat,
1239 stpncpy, strncpy, and their checking forms.
1241 2017-12-16 Jan Hubicka <hubicka@ucw.cz>
1243 PR rtl-optimization/82849
1244 * modulo-sched.c (sms_schedule): Use get_estimated_loop_iterations_int
1245 and get_max_loop_iterations_int.
1247 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1248 Alan Hayward <alan.hayward@arm.com>
1249 David Sherwood <david.sherwood@arm.com>
1251 * machmode.h (mode_for_size, int_mode_for_size, float_mode_for_size)
1252 (smallest_mode_for_size, smallest_int_mode_for_size): Take the mode
1253 size as a poly_uint64.
1254 (mode_for_vector, mode_for_int_vector): Take the number of vector
1255 elements as a poly_uint64.
1256 * stor-layout.c (mode_for_size, smallest_mode_for_size): Take the mode
1257 size as a poly_uint64.
1258 (mode_for_vector, mode_for_int_vector): Take the number of vector
1259 elements as a poly_uint64.
1261 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1262 Alan Hayward <alan.hayward@arm.com>
1263 David Sherwood <david.sherwood@arm.com>
1265 * machmode.h (MACRO_MODE): New macro.
1266 * addresses.h (base_reg_class, ok_for_base_p_1): Use it.
1267 * targhooks.c (default_libcall_value, default_secondary_reload)
1268 (default_memory_move_cost, default_register_move_cost)
1269 (default_class_max_nregs): Likewise.
1271 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1272 Alan Hayward <alan.hayward@arm.com>
1273 David Sherwood <david.sherwood@arm.com>
1275 * doc/sourcebuild.texi: Document IN_TARGET_CODE.
1276 * genattrtab.c (write_header): Define IN_TARGET_CODE to 1 in the
1278 * genautomata.c (main): Likewise.
1279 * genconditions.c (write_header): Likewise.
1280 * genemit.c (main): Likewise.
1281 * genextract.c (print_header): Likewise.
1282 * genopinit.c (main): Likewise.
1283 * genoutput.c (output_prologue): Likewise.
1284 * genpeep.c (main): Likewise.
1285 * genpreds.c (write_insn_preds_c): Likewise.
1286 * genrecog.c (writer_header): Likewise.
1287 * config/aarch64/aarch64-builtins.c (IN_TARGET_CODE): Define.
1288 * config/aarch64/aarch64-c.c (IN_TARGET_CODE): Likewise.
1289 * config/aarch64/aarch64.c (IN_TARGET_CODE): Likewise.
1290 * config/aarch64/cortex-a57-fma-steering.c (IN_TARGET_CODE): Likewise.
1291 * config/aarch64/driver-aarch64.c (IN_TARGET_CODE): Likewise.
1292 * config/alpha/alpha.c (IN_TARGET_CODE): Likewise.
1293 * config/alpha/driver-alpha.c (IN_TARGET_CODE): Likewise.
1294 * config/arc/arc-c.c (IN_TARGET_CODE): Likewise.
1295 * config/arc/arc.c (IN_TARGET_CODE): Likewise.
1296 * config/arc/driver-arc.c (IN_TARGET_CODE): Likewise.
1297 * config/arm/aarch-common.c (IN_TARGET_CODE): Likewise.
1298 * config/arm/arm-builtins.c (IN_TARGET_CODE): Likewise.
1299 * config/arm/arm-c.c (IN_TARGET_CODE): Likewise.
1300 * config/arm/arm.c (IN_TARGET_CODE): Likewise.
1301 * config/arm/driver-arm.c (IN_TARGET_CODE): Likewise.
1302 * config/avr/avr-c.c (IN_TARGET_CODE): Likewise.
1303 * config/avr/avr-devices.c (IN_TARGET_CODE): Likewise.
1304 * config/avr/avr-log.c (IN_TARGET_CODE): Likewise.
1305 * config/avr/avr.c (IN_TARGET_CODE): Likewise.
1306 * config/avr/driver-avr.c (IN_TARGET_CODE): Likewise.
1307 * config/avr/gen-avr-mmcu-specs.c (IN_TARGET_CODE): Likewise.
1308 * config/bfin/bfin.c (IN_TARGET_CODE): Likewise.
1309 * config/c6x/c6x.c (IN_TARGET_CODE): Likewise.
1310 * config/cr16/cr16.c (IN_TARGET_CODE): Likewise.
1311 * config/cris/cris.c (IN_TARGET_CODE): Likewise.
1312 * config/darwin.c (IN_TARGET_CODE): Likewise.
1313 * config/epiphany/epiphany.c (IN_TARGET_CODE): Likewise.
1314 * config/epiphany/mode-switch-use.c (IN_TARGET_CODE): Likewise.
1315 * config/epiphany/resolve-sw-modes.c (IN_TARGET_CODE): Likewise.
1316 * config/fr30/fr30.c (IN_TARGET_CODE): Likewise.
1317 * config/frv/frv.c (IN_TARGET_CODE): Likewise.
1318 * config/ft32/ft32.c (IN_TARGET_CODE): Likewise.
1319 * config/h8300/h8300.c (IN_TARGET_CODE): Likewise.
1320 * config/i386/djgpp.c (IN_TARGET_CODE): Likewise.
1321 * config/i386/driver-i386.c (IN_TARGET_CODE): Likewise.
1322 * config/i386/driver-mingw32.c (IN_TARGET_CODE): Likewise.
1323 * config/i386/host-cygwin.c (IN_TARGET_CODE): Likewise.
1324 * config/i386/host-i386-darwin.c (IN_TARGET_CODE): Likewise.
1325 * config/i386/host-mingw32.c (IN_TARGET_CODE): Likewise.
1326 * config/i386/i386-c.c (IN_TARGET_CODE): Likewise.
1327 * config/i386/i386.c (IN_TARGET_CODE): Likewise.
1328 * config/i386/intelmic-mkoffload.c (IN_TARGET_CODE): Likewise.
1329 * config/i386/msformat-c.c (IN_TARGET_CODE): Likewise.
1330 * config/i386/winnt-cxx.c (IN_TARGET_CODE): Likewise.
1331 * config/i386/winnt-stubs.c (IN_TARGET_CODE): Likewise.
1332 * config/i386/winnt.c (IN_TARGET_CODE): Likewise.
1333 * config/i386/x86-tune-sched-atom.c (IN_TARGET_CODE): Likewise.
1334 * config/i386/x86-tune-sched-bd.c (IN_TARGET_CODE): Likewise.
1335 * config/i386/x86-tune-sched-core.c (IN_TARGET_CODE): Likewise.
1336 * config/i386/x86-tune-sched.c (IN_TARGET_CODE): Likewise.
1337 * config/ia64/ia64-c.c (IN_TARGET_CODE): Likewise.
1338 * config/ia64/ia64.c (IN_TARGET_CODE): Likewise.
1339 * config/iq2000/iq2000.c (IN_TARGET_CODE): Likewise.
1340 * config/lm32/lm32.c (IN_TARGET_CODE): Likewise.
1341 * config/m32c/m32c-pragma.c (IN_TARGET_CODE): Likewise.
1342 * config/m32c/m32c.c (IN_TARGET_CODE): Likewise.
1343 * config/m32r/m32r.c (IN_TARGET_CODE): Likewise.
1344 * config/m68k/m68k.c (IN_TARGET_CODE): Likewise.
1345 * config/mcore/mcore.c (IN_TARGET_CODE): Likewise.
1346 * config/microblaze/microblaze-c.c (IN_TARGET_CODE): Likewise.
1347 * config/microblaze/microblaze.c (IN_TARGET_CODE): Likewise.
1348 * config/mips/driver-native.c (IN_TARGET_CODE): Likewise.
1349 * config/mips/frame-header-opt.c (IN_TARGET_CODE): Likewise.
1350 * config/mips/mips.c (IN_TARGET_CODE): Likewise.
1351 * config/mmix/mmix.c (IN_TARGET_CODE): Likewise.
1352 * config/mn10300/mn10300.c (IN_TARGET_CODE): Likewise.
1353 * config/moxie/moxie.c (IN_TARGET_CODE): Likewise.
1354 * config/msp430/driver-msp430.c (IN_TARGET_CODE): Likewise.
1355 * config/msp430/msp430-c.c (IN_TARGET_CODE): Likewise.
1356 * config/msp430/msp430.c (IN_TARGET_CODE): Likewise.
1357 * config/nds32/nds32-cost.c (IN_TARGET_CODE): Likewise.
1358 * config/nds32/nds32-fp-as-gp.c (IN_TARGET_CODE): Likewise.
1359 * config/nds32/nds32-intrinsic.c (IN_TARGET_CODE): Likewise.
1360 * config/nds32/nds32-isr.c (IN_TARGET_CODE): Likewise.
1361 * config/nds32/nds32-md-auxiliary.c (IN_TARGET_CODE): Likewise.
1362 * config/nds32/nds32-memory-manipulation.c (IN_TARGET_CODE): Likewise.
1363 * config/nds32/nds32-pipelines-auxiliary.c (IN_TARGET_CODE): Likewise.
1364 * config/nds32/nds32-predicates.c (IN_TARGET_CODE): Likewise.
1365 * config/nds32/nds32.c (IN_TARGET_CODE): Likewise.
1366 * config/nios2/nios2.c (IN_TARGET_CODE): Likewise.
1367 * config/nvptx/mkoffload.c (IN_TARGET_CODE): Likewise.
1368 * config/nvptx/nvptx.c (IN_TARGET_CODE): Likewise.
1369 * config/pa/pa.c (IN_TARGET_CODE): Likewise.
1370 * config/pdp11/pdp11.c (IN_TARGET_CODE): Likewise.
1371 * config/powerpcspe/driver-powerpcspe.c (IN_TARGET_CODE): Likewise.
1372 * config/powerpcspe/host-darwin.c (IN_TARGET_CODE): Likewise.
1373 * config/powerpcspe/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
1374 * config/powerpcspe/powerpcspe-c.c (IN_TARGET_CODE): Likewise.
1375 * config/powerpcspe/powerpcspe-linux.c (IN_TARGET_CODE): Likewise.
1376 * config/powerpcspe/powerpcspe.c (IN_TARGET_CODE): Likewise.
1377 * config/riscv/riscv-builtins.c (IN_TARGET_CODE): Likewise.
1378 * config/riscv/riscv-c.c (IN_TARGET_CODE): Likewise.
1379 * config/riscv/riscv.c (IN_TARGET_CODE): Likewise.
1380 * config/rl78/rl78-c.c (IN_TARGET_CODE): Likewise.
1381 * config/rl78/rl78.c (IN_TARGET_CODE): Likewise.
1382 * config/rs6000/driver-rs6000.c (IN_TARGET_CODE): Likewise.
1383 * config/rs6000/host-darwin.c (IN_TARGET_CODE): Likewise.
1384 * config/rs6000/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
1385 * config/rs6000/rs6000-c.c (IN_TARGET_CODE): Likewise.
1386 * config/rs6000/rs6000-linux.c (IN_TARGET_CODE): Likewise.
1387 * config/rs6000/rs6000-p8swap.c (IN_TARGET_CODE): Likewise.
1388 * config/rs6000/rs6000-string.c (IN_TARGET_CODE): Likewise.
1389 * config/rs6000/rs6000.c (IN_TARGET_CODE): Likewise.
1390 * config/rx/rx.c (IN_TARGET_CODE): Likewise.
1391 * config/s390/driver-native.c (IN_TARGET_CODE): Likewise.
1392 * config/s390/s390-c.c (IN_TARGET_CODE): Likewise.
1393 * config/s390/s390.c (IN_TARGET_CODE): Likewise.
1394 * config/sh/sh-c.c (IN_TARGET_CODE): Likewise.
1395 * config/sh/sh-mem.cc (IN_TARGET_CODE): Likewise.
1396 * config/sh/sh.c (IN_TARGET_CODE): Likewise.
1397 * config/sh/sh_optimize_sett_clrt.cc (IN_TARGET_CODE): Likewise.
1398 * config/sh/sh_treg_combine.cc (IN_TARGET_CODE): Likewise.
1399 * config/sparc/driver-sparc.c (IN_TARGET_CODE): Likewise.
1400 * config/sparc/sparc-c.c (IN_TARGET_CODE): Likewise.
1401 * config/sparc/sparc.c (IN_TARGET_CODE): Likewise.
1402 * config/spu/spu-c.c (IN_TARGET_CODE): Likewise.
1403 * config/spu/spu.c (IN_TARGET_CODE): Likewise.
1404 * config/stormy16/stormy16.c (IN_TARGET_CODE): Likewise.
1405 * config/tilegx/mul-tables.c (IN_TARGET_CODE): Likewise.
1406 * config/tilegx/tilegx-c.c (IN_TARGET_CODE): Likewise.
1407 * config/tilegx/tilegx.c (IN_TARGET_CODE): Likewise.
1408 * config/tilepro/mul-tables.c (IN_TARGET_CODE): Likewise.
1409 * config/tilepro/tilepro-c.c (IN_TARGET_CODE): Likewise.
1410 * config/tilepro/tilepro.c (IN_TARGET_CODE): Likewise.
1411 * config/v850/v850-c.c (IN_TARGET_CODE): Likewise.
1412 * config/v850/v850.c (IN_TARGET_CODE): Likewise.
1413 * config/vax/vax.c (IN_TARGET_CODE): Likewise.
1414 * config/visium/visium.c (IN_TARGET_CODE): Likewise.
1415 * config/vms/vms-c.c (IN_TARGET_CODE): Likewise.
1416 * config/vms/vms-f.c (IN_TARGET_CODE): Likewise.
1417 * config/vms/vms.c (IN_TARGET_CODE): Likewise.
1418 * config/xtensa/xtensa.c (IN_TARGET_CODE): Likewise.
1420 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1421 Alan Hayward <alan.hayward@arm.com>
1422 David Sherwood <david.sherwood@arm.com>
1424 * tree-ssa-sccvn.c (vn_reference_lookup_3): Avoid repeated
1427 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1428 Alan Hayward <alan.hayward@arm.com>
1429 David Sherwood <david.sherwood@arm.com>
1431 * doc/generic.texi (VEC_SERIES_EXPR): Document.
1432 * doc/md.texi (vec_series@var{m}): Document.
1433 * tree.def (VEC_SERIES_EXPR): New tree code.
1434 * tree.h (build_vec_series): Declare.
1435 * tree.c (build_vec_series): New function.
1436 * cfgexpand.c (expand_debug_expr): Handle VEC_SERIES_EXPR.
1437 * tree-pretty-print.c (dump_generic_node): Likewise.
1438 * gimple-pretty-print.c (dump_binary_rhs): Likewise.
1439 * tree-inline.c (estimate_operator_cost): Likewise.
1440 * expr.c (expand_expr_real_2): Likewise.
1441 * optabs-tree.c (optab_for_tree_code): Likewise.
1442 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
1443 * fold-const.c (const_binop): Fold VEC_SERIES_EXPRs of constants.
1444 * expmed.c (make_tree): Handle VEC_SERIES.
1445 * optabs.def (vec_series_optab): New optab.
1446 * optabs.h (expand_vec_series_expr): Declare.
1447 * optabs.c (expand_vec_series_expr): New function.
1448 * tree-vect-generic.c (expand_vector_operations_1): Check that
1449 the operands also have vector type.
1451 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1452 Alan Hayward <alan.hawyard@arm.com>
1453 David Sherwood <david.sherwood@arm.com>
1455 * doc/generic.texi (VEC_DUPLICATE_EXPR): Document.
1456 (VEC_COND_EXPR): Add missing @tindex.
1457 * doc/md.texi (vec_duplicate@var{m}): Document.
1458 * tree.def (VEC_DUPLICATE_EXPR): New tree codes.
1459 * tree.c (build_vector_from_val): Add stubbed-out handling of
1460 variable-length vectors, using VEC_DUPLICATE_EXPR.
1461 (uniform_vector_p): Handle VEC_DUPLICATE_EXPR.
1462 * cfgexpand.c (expand_debug_expr): Likewise.
1463 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
1464 * tree-inline.c (estimate_operator_cost): Likewise.
1465 * tree-pretty-print.c (dump_generic_node): Likewise.
1466 * tree-vect-generic.c (ssa_uniform_vector_p): Likewise.
1467 * fold-const.c (const_unop): Fold VEC_DUPLICATE_EXPRs of a constant.
1468 (test_vec_duplicate_folding): New function.
1469 (fold_const_c_tests): Call it.
1470 * optabs.def (vec_duplicate_optab): New optab.
1471 * optabs-tree.c (optab_for_tree_code): Handle VEC_DUPLICATE_EXPR.
1472 * optabs.h (expand_vector_broadcast): Declare.
1473 * optabs.c (expand_vector_broadcast): Make non-static. Try using
1474 vec_duplicate_optab.
1475 * expr.c (store_constructor): Try using vec_duplicate_optab for
1477 (expand_expr_real_2): Handle VEC_DUPLICATE_EXPR.
1479 2017-12-15 Markus Trippelsdorf <markus@trippelsdorf.de>
1482 * config/i386/x86-tune-costs.h (skylake_cost, core_cost): Increase
1483 div/mod latencies a bit.
1485 2017-12-15 Jeff Law <law@redhat.com>
1487 PR tree-optimization/36550
1488 * tree-ssa-threadupdate.c (count_stmts_and_phis_in_block): New.
1489 (mark_threaded_blocks): Rewrite code to avoid block copying when
1490 optimizing for size. Don't pessimize blocks which will be
1491 copied, but all the statements will be dead.
1493 2017-12-15 Alexandre Oliva <aoliva@redhat.com>
1495 PR tree-optimization/81165
1496 * tree-ssa-threadupdate.c (uses_in_bb): New.
1497 (estimate_threading_killed_stmts): New.
1498 * tree-ssa-threadupdate.h (estimate_threading_killed_stmts): Prototype.
1499 * tree-ssa-threadedge.c
1500 (record_temporary_equivalences_from_stmts_at_dest): Expand limit
1503 2017-12-15 Jeff Law <law@redhat.com>
1505 PR tree-optimization/83410
1506 * tree-ssa-threadupdate.c (thread_block_1): Avoid certain jump
1507 threads when parallelizing loops.
1509 2017-12-15 Jakub Jelinek <jakub@redhat.com>
1511 * tree-core.h (struct attribute_spec): Swap affects_type_identity and
1513 * config/alpha/alpha.c (vms_attribute_table): Swap
1514 affects_type_identity and handler fields, adjust comments.
1515 * config/mips/mips.c (mips_attribute_table): Likewise.
1516 * config/visium/visium.c (visium_attribute_table): Likewise.
1517 * config/epiphany/epiphany.c (epiphany_attribute_table): Likewise.
1518 * config/microblaze/microblaze.c (microblaze_attribute_table):
1520 * config/spu/spu.c (spu_attribute_table): Likewise.
1521 * config/mcore/mcore.c (mcore_attribute_table): Likewise.
1522 * config/arc/arc.c (arc_attribute_table): Likewise.
1523 * config/m68k/m68k.c (m68k_attribute_table): Likewise.
1524 * config/v850/v850.c (v850_handle_interrupt_attribute,
1525 v850_handle_data_area_attribute): Formatting fixes.
1526 (v850_attribute_table): Swap affects_type_identity and handler
1527 fields, adjust comments.
1528 * config/m32r/m32r.c (m32r_attribute_table): Likewise.
1529 * config/arm/arm.c (arm_attribute_table): Likewise.
1530 * config/avr/avr.c (avr_attribute_table): Likewise.
1531 * config/s390/s390.c (s390_attribute_table): Likewise.
1532 * config/sh/sh.c (sh_attribute_table): Likewise.
1533 * config/i386/i386.c (ix86_handle_cconv_attribute,
1534 ix86_handle_callee_pop_aggregate_return): Formatting fixes.
1535 (ix86_attribute_table): Swap affects_type_identity and handler
1536 fields, adjust comments.
1537 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
1538 * config/sparc/sparc.c (sparc_attribute_table): Likewise.
1539 * config/m32c/m32c.c (m32c_attribute_table): Likewise.
1540 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Likewise.
1541 * config/ia64/ia64.c (ia64_attribute_table): Likewise.
1542 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
1543 * config/rx/rx.c (rx_attribute_table): Likewise.
1544 * config/cr16/cr16.c (cr16_attribute_table): Likewise.
1545 * config/h8300/h8300.c (h8300_attribute_table): Likewise.
1546 * config/nvptx/nvptx.c (nvptx_attribute_table): Likewise.
1547 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Likewise.
1548 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
1549 * config/stormy16/stormy16.c (xstormy16_attribute_table): Likewise.
1550 * config/bfin/bfin.c (bfin_attribute_table): Likewise.
1551 * config/rs6000/rs6000.c (rs6000_attribute_table): Likewise.
1552 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
1553 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
1554 * doc/plugins.texi (user_attr): Likewise. Add NULL for
1556 * attribs.c (empty_attribute_table): Swap affects_type_identity and
1558 (register_scoped_attributes, decl_attributes): Formatting fixes.
1560 PR tree-optimization/83269
1561 * fold-const.c (fold_binary_loc): Perform (-A) - B -> (-B) - A
1562 subtraction in arg0's type if type is signed and arg0 is unsigned.
1566 * match.pd ((T)(P + A) - (T)P -> (T) A): Use @@0 instead of @0 and
1567 convert? on @0 instead of convert. Check type of @1, not @0.
1568 ((T)P - (T)(P + A) -> -(T) A): Use @@0 instead of @0 and
1569 convert? on @0 instead of convert. Check type of @1, not @0.
1570 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Use @@0 instead of @0,
1571 only optimize if either both @1 and @2 types are narrower
1572 precision, or both are wider or equal precision, and in the former
1573 case only if both have undefined overflow.
1575 2017-12-15 Richard Biener <rguenther@suse.de>
1578 * internal-fn.def (IFN_NOP): Add.
1579 * internal-fn.c (expand_NOP): Do nothing.
1580 * lto-streamer-in.c (input_function): Instead of removing
1581 sanitizer calls replace them with IFN_NOP calls.
1583 2017-12-15 Richard Sandiford <richard.sandiford@linaro.org>
1584 Alan Hayward <alan.hayward@arm.com>
1585 David Sherwood <david.sherwood@arm.com>
1587 * dse.c (store_info, read_info_type): Replace begin and end with
1589 (print_range): New function.
1590 (set_all_positions_unneeded, any_positions_needed_p)
1591 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Update
1593 (record_store): Likewise. Optimize the case in which all positions
1595 (get_stored_val): Replace read_begin and read_end with read_offset
1597 (replace_read): Update call accordingly.
1599 2017-12-15 Bin Cheng <bin.cheng@arm.com>
1601 * gimple-loop-interchange.cc (STMT_COST_RATIO): New macro.
1602 (loop_cand::m_num_stmts, loop_cand::m_const_init_reduc): New members.
1603 (loop_cand::loop_cand): Initialize above members.
1604 (loop_cand::supported_operations): Delete.
1605 (loop_cand::can_interchange_p): Inline above function.
1606 (loop_cand::classify_simple_reduction): Record number of constant
1607 initialized simple reductions.
1608 (should_interchange_loops): New parameters. Check stmt cost of loops
1610 (tree_loop_interchange::interchange): Prepare stmt cost of outer loop.
1611 Update call to should_interchange_loops.
1612 (should_interchange_loop_nest): Update call to
1613 should_interchange_loops.
1615 2017-12-15 Eric Botcazou <ebotcazou@adacore.com>
1618 * ggc-page.c (HOST_BITS_PER_PTR): Do not define here...
1619 * hwint.h (HOST_BITS_PER_PTR): ...but here instead.
1620 * config/i386/xm-mingw32.h (HOST_BITS_PER_PTR): Delete.
1622 2017-12-15 Julia Koval <julia.koval@intel.com>
1624 * config/i386/i386-builtin.def (__builtin_ia32_vaesenclast_v16qi,
1625 __builtin_ia32_vaesenclast_v32qi, __builtin_ia32_vaesenclast_v64qi): New.
1626 * config/i386/sse.md (vaesenclast_<mode>): New pattern.
1627 * config/i386/vaesintrin.h (_mm256_aesenclast_epi128,
1628 _mm512_aesenclast_epi128, _mm_aesenclast_epi128): New intrinsics.
1630 2017-12-15 Julia Koval <julia.koval@intel.com>
1632 * config/i386/i386-builtin.def (__builtin_ia32_vaesenc_v16qi,
1633 __builtin_ia32_vaesenc_v32qi, __builtin_ia32_vaesenc_v64qi): New.
1634 * config/i386/sse.md (vaesenc_<mode>): New pattern.
1635 * config/i386/vaesintrin.h (_mm256_aesenc_epi128, _mm512_aesenc_epi128,
1636 _mm_aesenc_epi128): New intrinsics.
1638 2017-12-15 Julia Koval <julia.koval@intel.com>
1640 * config/i386/i386-builtin.def (__builtin_ia32_vaesdeclast_v16qi,
1641 __builtin_ia32_vaesdeclast_v32qi, __builtin_ia32_vaesdeclast_v64qi): New.
1642 * config/i386/sse.md (vaesdeclast_<mode>): New pattern.
1643 * config/i386/vaesintrin.h (_mm256_aesdeclast_epi128,
1644 _mm512_aesdeclast_epi128, _mm_aesdeclast_epi128): New intrinsics.
1646 2017-12-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1648 * gimple-ssa-strength-reduction.c (analyze_increments):
1649 Distinguish replacement costs for constant strides from those for
1652 2017-12-14 Jakub Jelinek <jakub@redhat.com>
1654 * var-tracking.c (variable_tracking_main_1): Formatting fix.
1656 2017-12-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
1658 * doc/invoke.texi: Document -Wcast-function-type.
1659 * recog.h (stored_funcptr): Change signature.
1660 * tree-dump.c (dump_node): Avoid warning.
1661 * typed-splay-tree.h (typed_splay_tree): Avoid warning.
1663 2017-12-14 Qing Zhao <qing.zhao@oracle.com>
1666 * gimple-fold.c (get_range_strlen): Add the handling of non-member
1669 2017-12-14 David Malcolm <dmalcolm@redhat.com>
1671 PR tree-optimization/83312
1672 * domwalk.h (dom_walker::dom_walker): Fix typo in comment.
1673 * tree-cfg.c (find_taken_edge): Update to handle NULL_TREE for
1674 "val" param, and to cope with arbitrary basic blocks.
1675 (find_taken_edge_cond_expr): Add "cond_stmt" param and use it to
1676 handle NULL_TREE for "val", dropping "bb" param.
1677 (find_taken_edge_switch_expr): Make "switch_stmt" param const and
1678 drop "bb" param. Handle NULL_TREE for "val".
1679 (find_case_label_for_value): Make "switch_stmt" param const.
1680 * tree-vrp.c (class check_array_bounds_dom_walker): New subclass
1682 (vrp_prop::check_all_array_refs): Reimplement as...
1683 (check_array_bounds_dom_walker::before_dom_children): ...this new
1684 vfunc. Replace linear search through BB block list, excluding
1685 those with non-executable in-edges via dominator walk.
1687 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1689 * config/arm/arm.opt (mverbose-cost-dump): New option.
1690 * config/arm/arm.c (arm_rtx_costs): Use it.
1692 2017-12-14 Andreas Schwab <schwab@linux-m68k.org>
1695 * reload1.c (emit_input_reload_insns): Skip debug markers.
1697 2017-12-14 Alexandre Oliva <aoliva@redhat.com>
1699 * config/i386/i386.c (rest_of_insert_endbranch): Use call loc
1703 * config/arc/arc.c (hwloop_optimize): Skip debug insns.
1704 * config/sh/sh-protos.h (sh_find_set_of_reg): Adjust.
1705 * config/sh/sh.c: Skip debug insns besides notes.
1706 * config/sh/sh.md: Likewise.
1707 * config/sh/sh_treg_combine.cc: Likewise.
1708 * config/sh/sync.md: Likewise.
1710 2017-12-14 Tom de Vries <tom@codesourcery.com>
1712 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
1713 Add item for weak_undefined.
1715 2017-12-14 Richard Biener <rguenther@suse.de>
1717 PR tree-optimization/67842
1718 * sese.h (bb_in_region): Remove #if 0'ed code.
1720 2017-12-14 Richard Biener <rguenther@suse.de>
1722 PR tree-optimization/83326
1723 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
1724 may_be_zero parameter and handle it by not marking the first
1725 peeled copy as not exiting the loop.
1726 (try_peel_loop): Likewise.
1727 (canonicalize_loop_induction_variables): Use number_of_iterations_exit
1728 to handle the case of constant or zero iterations and perform
1729 loop header copying on-the-fly.
1731 2017-12-14 Richard Biener <rguenther@suse.de>
1733 PR tree-optimization/83418
1734 * vr-values.c (vr_values::extract_range_for_var_from_comparison_expr):
1735 Instead of asserting we don't get unfolded comparisons deal with
1738 2017-12-14 Jakub Jelinek <jakub@redhat.com>
1741 * var-tracking.c (vt_initialize): Ignore non-DEBUG_INSNs outside of
1742 basic blocks. Assert debug bind insns don't appear outside of bbs,
1743 don't reset them. Assert insns without BLOCK_FOR_INSN are outside of
1746 PR tree-optimization/83198
1747 * gimple-ssa-sprintf.c (format_floating): Set type solely based on
1748 dir.modifier, regardless of TREE_TYPE (arg). Assume non-REAL_CST
1749 value if arg is a REAL_CST with incompatible type.
1751 2017-12-14 Sudakshina Das <sudi.das@arm.com>
1752 Bin Cheng <bin.cheng@arm.com>
1755 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Move LTGT
1757 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Add
1760 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1762 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75,
1763 cortex-a75.cortex-a55): Specify AARCH64_FL_F16 in the arch features.
1765 2017-12-14 Richard Sandiford <richard.sandiford@linaro.org>
1766 Alan Hayward <alan.hayward@arm.com>
1767 David Sherwood <david.sherwood@arm.com>
1769 * poly-int.h: New file.
1770 * poly-int-types.h: Likewise.
1771 * coretypes.h: Include them.
1772 (POLY_INT_CONVERSION): Define.
1773 * target.def (estimated_poly_value): New hook.
1774 * doc/tm.texi.in (TARGET_ESTIMATED_POLY_VALUE): New hook.
1775 * doc/tm.texi: Regenerate.
1776 * doc/poly-int.texi: New file.
1777 * doc/gccint.texi: Include it.
1778 * doc/rtl.texi: Describe restrictions on subreg modes.
1779 * Makefile.in (TEXI_GCCINT_FILES): Add poly-int.texi.
1780 * genmodes.c (NUM_POLY_INT_COEFFS): Provide a default definition.
1781 (emit_insn_modes_h): Emit a definition of NUM_POLY_INT_COEFFS.
1782 * targhooks.h (default_estimated_poly_value): Declare.
1783 * targhooks.c (default_estimated_poly_value): New function.
1784 * target.h (estimated_poly_value): Likewise.
1785 * wide-int.h (WI_UNARY_RESULT): Use wi::binary_traits.
1786 (wi::unary_traits): Delete.
1787 (wi::binary_traits::signed_shift_result_type): Define for
1788 offset_int << HOST_WIDE_INT, etc.
1789 (generic_wide_int::operator <<=): Define for all types and use
1790 wi::lshift instead of <<.
1791 (wi::hwi_with_prec): Add a default constructor.
1792 (wi::ints_for): New class.
1793 (operator <<): Define for all wide-int types.
1794 (operator /): New function.
1795 (operator %): Likewise.
1796 * selftest.h (ASSERT_KNOWN_EQ, ASSERT_KNOWN_EQ_AT, ASSERT_MAYBE_NE)
1797 (ASSERT_MAYBE_NE_AT): New macros.
1799 2017-12-13 Eric Botcazou <ebotcazou@adacore.com>
1800 Dominik Vogt <vogt@linux.vnet.ibm.com>
1803 * emit-rtl.c (init_emit): Remove ??? comment.
1804 * explow.c (get_dynamic_stack_size): Take known alignment of stack
1805 pointer + STACK_DYNAMIC_OFFSET into account in lieu of STACK_BOUNDARY.
1806 * config/sparc/sparc.h (INIT_EXPANDERS): In 32-bit mode, lower the
1807 alignment of 3 virtual registers to BITS_PER_WORD.
1809 * config/sparc/sparc.c (sparc_compute_frame_size): Simplify.
1811 2017-12-13 Peter Bergner <bergner@vnet.ibm.com>
1813 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_HTM_NO_SUSPEND): New define.
1814 * config/rs6000/rs6000.c (cpu_supports_info): Use it.
1816 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
1819 * reload1.c (eliminate_regs_in_insn): Skip debug markers.
1821 2017-12-13 Jakub Jelinek <jakub@redhat.com>
1823 * tree-cfg.c (verify_gimple_in_cfg): Verify no non-label stmts
1824 with the exception of debug begin stmt markers appear before
1828 * final.c (rest_of_handle_final): Call variable_tracking_main only
1829 if !flag_var_tracking.
1831 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
1832 Jakub Jelinek <jakub@redhat.com>
1836 * tree-cfgcleanup.c (remove_forwarder_block): Keep after
1837 labels debug stmts that can only appear after labels.
1839 2017-12-13 Alexander Monakov <amonakov@ispras.ru>
1841 PR rtl-optimization/82398
1842 * sel-sched.c (sel_rank_for_schedule): Fix check for zero
1843 EXPR_USEFULNESS in priority comparison.
1845 2017-12-13 Segher Boessenkool <segher@kernel.crashing.org>
1847 PR rtl-optimization/83393
1848 * combine.c (move_deaths): If reg_stat points to a too new insn in
1849 last_death, do not use it: find the proper insn instead.
1851 2017-12-12 Jeff Law <law@redhat.com>
1853 PR tree-optimization/83298
1854 PR tree-optimization/83362
1855 PR tree-optimization/83383
1856 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Make
1857 push_value_range a public interface. Add new argument to
1858 record_ranges_from_stmt.
1859 * gimple-ssa-evrp-analyze.c
1860 (evrp_range_analyzer::record_ranges_from_stmt): Add new argument.
1861 Update comments. Handle recording temporary equivalences.
1862 * tree-ssa-dom.c (dom_opt_opt_walker::before_dom_children): Add
1863 new argument to call to evrp_range_analyzer::record_ranges_from_stmt.
1864 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Likewise.
1865 * tree-ssa-threadedge.c: Include alloc-pool.h, vr-values.h and
1866 gimple-ssa-evrp-analyze.h.
1867 (record_temporary_equivalences_from_phis): Add new argument. When
1868 the PHI arg is an SSA_NAME, set the result's range to the range
1870 (record_temporary_equivalences_from_stmts_at_dest): Record ranges
1871 from statements too.
1872 (thread_through_normal_block): Accept new argument, evrp_range_analyzer.
1873 Pass it down to children as needed.
1874 (thread_outgoing_edges): Likewise.
1875 (thread_across_edge): Likewise. Push/pop range state as needed.
1876 * tree-ssa-threadedge.h (thread_outgoing_edges): Update prototype.
1878 2017-12-12 Julia Koval <julia.koval@intel.com>
1880 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
1881 (PTA_CANNONLAKE): Remove PTA_CLWB.
1883 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1886 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
1887 (vcondv2div2df): Likewise.
1888 (vconduv2dfv2di): Likewise.
1890 2017-12-12 Jakub Jelinek <jakub@redhat.com>
1892 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
1893 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
1894 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
1895 ATTR_MATHFN_FPROUNDING_ERRNO.
1897 2017-12-12 Richard Biener <rguenther@suse.de>
1899 PR tree-optimization/83385
1900 * graphite-scop-detection.c (get_order, order): Remove.
1901 (bb_to_rpo): New global.
1903 (build_scops): Sort pbbs in RPO order.
1905 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
1907 * combine.c (simplify_set): Do not transform subregs to zero_extends
1908 if the destination mode is a vector mode.
1910 2017-12-12 Jakub Jelinek <jakub@redhat.com>
1912 PR tree-optimization/83359
1913 * tree-cfg.h (fold_loop_internal_call): Declare.
1914 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
1915 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
1916 (find_loop_dist_alias): New function.
1917 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
1918 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
1921 PR tree-optimization/80631
1922 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
1923 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
1924 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
1925 hardcoding zero as the value if COND_EXPR is never true. For
1926 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
1927 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
1928 hardcoding MAX_EXPR as the reduction operation.
1929 (is_nonwrapping_integer_induction): Allow negative step.
1930 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
1931 vect_create_epilog_for_reduction, if no value is suitable, don't
1932 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
1934 2017-12-12 Richard Biener <rguenther@suse.de>
1936 PR tree-optimization/81889
1937 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
1938 range info from the non-wrapping IV instead of just the range
1941 2017-12-12 Julia Koval <julia.koval@intel.com>
1943 * config.gcc: Add vaesintrin.h.
1944 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
1946 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
1947 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
1949 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
1950 * config/i386/immintrin.h: Include vaesintrin.h.
1951 * config/i386/sse.md (vaesdec_<mode>): New pattern.
1952 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
1953 _mm_aesdec_epi128): New intrinsics.
1955 2017-12-12 Julia Koval <julia.koval@intel.com>
1957 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
1958 OPTION_MASK_ISA_VAES_UNSET): New.
1959 (ix86_handle_option): Handle -mvaes.
1960 * config/i386/cpuid.h: Define bit_VAES.
1961 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
1962 * config/i386/i386-c.c (__VAES__): New.
1963 * config/i386/i386.c (ix86_target_string): Add -mvaes.
1964 (ix86_valid_target_attribute_inner_p): Ditto.
1965 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
1966 * config/i386/i386.opt: Add -mvaes.
1967 * doc/invoke.texi: Ditto.
1969 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
1971 * debug.h (gcc_debug_hooks): Add inline_entry.
1972 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
1973 * debug.c (do_nothing_debug_hooks): Likewise.
1974 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
1975 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
1976 (dwarf2_lineno_debug_hooks): Likewise.
1978 * common.opt (gstatement-frontiers): New, setting
1979 debug_nonbind_markers_p.
1980 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
1981 * toplev.c (process_options): Autodetect value for debug statement
1983 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
1984 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
1986 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
1987 markers. Integrate source bind into debug stmt expand loop.
1988 (pass_expand::execute): Check debug marker limit. Avoid deep
1989 TER and expand debug locations for debug bind insns only.
1990 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
1992 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
1993 * final.c (reemit_insn_block_notes): Take current block from
1994 nonbind markers. Declare note where it's first set.
1995 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
1996 begin stmt markers if enabled.
1997 (notice_source_line): Handle nonbind markers. Fail if their
1998 location is unknown or that of builtins.
1999 (rest_of_handle_final): Convert begin stmt markers to notes if
2000 var-tracking didn't run.
2001 (rest_of_clean_state): Skip begin stmt markers.
2002 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
2004 * function.c (allocate_struct_function): Set begin_stmt_markers.
2005 * function.h (struct function): Add debug_marker_count counter
2006 and debug_nonbind_markers flag.
2007 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
2008 * gimple-low.c (lower_function_body): Adjust
2009 debug_nonbind_markers.
2010 (lower_stmt): Drop or skip gimple debug stmts.
2011 (lower_try_catch): Skip debug stmts.
2012 * gimple.c (gimple_build_debug_begin_stmt): New.
2013 (gimple_copy): Increment debug_marker_count if copying one.
2014 * gimple.h (gimple_build_debug_begin_stmt): Declare.
2015 * gimplify.c (rexpr_location): New.
2016 (rexpr_has_location): New.
2017 (warn_switch_unreachable_r): Handle gimple debug stmts.
2018 (shortcut_cond_r): Call expr_location.
2020 (find_goto_label): New.
2021 (shortcut_cond_expr): Call expr_has_location, expr_location, and
2023 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
2025 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
2026 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
2027 (LANG_HOOKS_INITIALIZER): ... this.
2028 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
2029 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
2031 (update_ebb_live_info): Skip debug insn markers.
2032 * lra.c (debug_insn_static_data): Rename to...
2033 (debug_bind_static_data): ... this.
2034 (debug_marker_static_data): New.
2035 (lra_set_insn_recog_data): Select one of the above depending
2037 (lra_update_isn_regno_info): Don't assume debug insns have
2039 (push_insns): Skip debug insns.
2040 * lto-streamer-in.c (input_function): Drop debug stmts
2041 depending on active options. Adjust debug_nonbind_markers.
2042 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
2043 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
2044 begin stmt marker notes.
2045 (print_insn): Likewise.
2046 * recog.c (extract_insn): Recognize rtl for debug markers.
2047 * rtl.def (DEBUG_MARKER): New.
2048 * tree-inline.c: Include params.h.
2049 (remap_gimple_stmt): Handle nonbind markers.
2050 (maybe_move_debug_stmts_to_successors): Likewise.
2051 (copy_debug_stmt): Likewise.
2052 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
2053 markers regardless of no side effects.
2054 (tsi_link_before): Don't update container's side effects when adding
2055 a begin stmt marker.
2056 (tsi_link_after): Likewise.
2057 (expr_first): Skip begin stmt markers.
2058 (expr_last): Likewise.
2059 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
2060 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
2061 Disregard nonbind markers.
2062 * tree.c (make_node_stat): Don't set side effects for begin stmt
2064 (build1_stat): Likewise.
2065 * tree.def (DEBUG_BEGIN_STMT): New.
2066 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
2067 * var-tracking.c (delete_debug_insns): Renamed to...
2068 (delete_vta_debug_insns): ... this.
2069 (reemit_marker_as_note): New.
2070 (vt_initialize): Reemit markers.
2071 (delete_vta_debug_insns): Likewise.
2072 (vt_debug_insns_local): Reemit or delete markers.
2073 (variable_tracking_main_1): Likewise.
2074 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
2075 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
2076 (gimple_debug_nonbind_marker_p): New.
2077 (gimple_build_debug_bind): Adjust.
2078 (gimple_build_debug_begin_stmt): New.
2079 * doc/invoke.texi (max-debug-marker-count): New param.
2080 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
2081 (debug_parameter_ref, debug_marker): New.
2082 (NOTE_INSN_BEGIN_STMT): New.
2083 (DEBUG_INSN): Describe begin stmt markers.
2085 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
2086 without debug insns, we wouldn't, but clean up debug insns
2087 after a control flow insn nevertheless.
2089 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
2090 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
2091 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
2092 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
2093 (rtl_tidy_fallthru_edge): Likewise.
2094 (rtl_verify_fallthru): Likewise.
2095 (rtl_verify_bb_layout): Likewise.
2096 (skip_insns_after_block): Likewise.
2097 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
2098 * dwarf2out.c: Include print-rtl.h.
2099 (dwarf2out_next_real_insn): New.
2100 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
2101 Dump debug binds in asm comments.
2102 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
2103 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
2104 callers to use gsi_start_nondebug_bb instead.
2105 (gsi_after_labels): Skip gimple debug stmts.
2106 (gsi_start_nondebug): New.
2107 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
2108 (proper_loop_form_for_interchange): Adjust.
2109 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
2110 * gimple.h (gimple_seq_last_nondebug_stmt): New.
2111 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
2112 (collect_fallthrough_labels): Likewise.
2113 (should_warn_for_implicit_fallthrough): Likewise.
2114 (warn_implicit_fallthrough_r): Likewise.
2115 (expand_FALLTHROUGH_r): Likewise.
2116 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
2117 (graphite_copy_stmts_from_block): Skip nonbind markers.
2118 * haifa-sched.c (sched_extend_bb): Skip debug insns.
2119 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
2120 * jump.c (clean_barriers): Skip debug insns.
2121 * omp-expand.c (expand_parallel_call): Skip debug insns.
2122 (expand_task_call): Likewise.
2123 (remove_exit_barrier): Likewise.
2124 (expand_omp_taskreg): Likewise.
2125 (expand_omp_for_init_counts): Likewise.
2126 (expand_omp_for_generic): Likewise.
2127 (expand_omp_for_static_nochunk): Likewise.
2128 (expand_omp_for_static_chunk): Likewise.
2129 (expand_omp_simd): Likewise.
2130 (expand_omp_taskloop_for_outer): Likewise.
2131 (expand_omp_taskloop_for_inner): Likewise.
2132 (expand_oacc_for): Likewise.
2133 (expand_omp_sections): Likewise.
2134 (expand_omp_single): Likewise.
2135 (expand_omp_synch): Likewise.
2136 (expand_omp_atomic_load): Likewise.
2137 (expand_omp_atomic_store): Likewise.
2138 (expand_omp_atomic_fetch_op): Likewise.
2139 (expand_omp_atomic_pipeline): Likewise.
2140 (expand_omp_atomic_mutex): Likewise.
2141 (expand_omp_target): Likewise.
2142 (grid_expand_omp_for_loop): Likewise.
2143 (grid_expand_target_grid_body): Likewise.
2144 (build_omp_regions_1): Likewise.
2145 * omp-low.c (check_combined_parallel): Skip debug stmts.
2146 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
2147 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
2148 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
2150 * tree-cfg.c (make_blobs_1): Skip debug stmts.
2151 (make_edges): Likewise.
2152 (cleanup_dead_labels): Likewise.
2153 (gimple_can_merge_blocks_p): Likewise.
2154 (stmt_starts_bb_p): Likewise.
2155 (gimple_block_label): Likewise.
2156 (gimple_redirect_edge_and_branch): Likewise.
2157 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
2159 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
2161 * tree-pretty-print (print_declaration): Omit initializer in slim
2163 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
2165 (eliminate_unnecessary_stmts): Stabilize block removal order.
2166 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
2167 * var-tracking.c (get_first_insn): New.
2168 (vt_emit_notes): Call it.
2169 (vt_initialize): Walk any insns before the first BB.
2170 (delete_debug_insns): Likewise.
2172 * gimple.h (enum gimple_debug_subcode): Add
2173 GIMPLE_DEBUG_BEGIN_STMT.
2174 (gimple_debug_begin_stmt_p): New.
2175 (gimple_debug_nonbind_marker_p): New.
2176 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
2177 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
2178 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
2179 * insn-notes.def (BEGIN_STMT): New.
2180 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
2181 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
2182 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
2183 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
2184 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
2185 (INSN_DEBUG_MARKER_KIND): New.
2186 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
2187 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
2188 (INSN_VAR_LOCATION_PTR): New.
2189 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
2191 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
2192 deps only if debug bind insns are enabled.
2193 (pass_expand::execute): Avoid deep TER and expand
2194 debug locations for debug bind insns only.
2195 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
2196 debug stmts special handling down to debug bind stmts.
2197 * combine.c (try_combine): Narrow debug insns special handling
2198 down to debug bind insns.
2199 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
2200 Narrow debug insns preexisting special handling down to debug
2202 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
2203 handling down to debug bind insns.
2204 * function.c (instantiate_virtual_regs): Skip debug markers,
2205 adjust handling of debug binds.
2206 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
2207 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
2208 * haifa-sched.c (schedule_insn): Narrow special handling of debug
2209 insns to debug bind insns.
2210 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
2211 special handling of debug stmts to debug bind stmts.
2212 * ipa-split.c (split_function): Likewise.
2213 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
2214 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
2216 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
2217 * regrename.c (build_def_use): Likewise.
2218 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
2219 (pass_cprop_hardreg): Narrow special casing of debug insns to
2221 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
2222 * reload1.c (reload): Likewise.
2223 * sese.c (sese_insert_phis_for_liveouts): Narrow special
2224 casing of debug stmts to debug bind stmts.
2225 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
2226 * ssa-iterators.h (num_imm_uses): Likewise.
2227 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
2228 debug stmts to debug bind stmts.
2229 * tree-inline.c (tree_function_versioning): Narrow special casing
2230 of debug stmts to debug bind stmts.
2231 * tree-loop-distribution.c (generate_loops_for_partition):
2232 Narrow special casing of debug stmts to debug bind stmts.
2233 * tree-sra.c (analyze_access_subtree): Narrow special casing
2234 of debug stmts to debug bind stmts.
2235 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
2236 stmts to debug bind stmts.
2237 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
2238 casing of debug stmts to debug bind stmts.
2239 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
2240 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
2241 casing of debug stmts to debug bind stmts.
2242 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
2244 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
2245 debug stmts to debug bind stmts.
2246 (gimple_replace_ssa_lhs): Likewise.
2247 (insert_debug_temp_for_var_def): Likewise.
2248 (insert_debug_temps_for_defs): Likewise.
2249 (reset_debug_uses): Likewise.
2250 * tree-ssanames.c (release_ssa_name_fn): Likewise.
2251 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
2252 (adjust_debug_stmts): Likewise.
2253 (adjust_phi_and_debug_stmts): Likewise.
2254 (vect_do_peeling): Likewise.
2255 * tree-vect-loop.c (vect_transform_loop): Likewise.
2256 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
2257 * var-tracking.c (adjust_mems): Narrow special casing of debug
2258 insns to debug bind insns.
2259 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
2260 (compute_bb_dataflow, vt_find_locations): Likewise.
2261 (vt_expand_loc, emit_notes_for_changes): Likewise.
2262 (vt_init_cfa_base): Likewise.
2263 (vt_emit_notes): Likewise.
2264 (vt_initialize): Likewise.
2265 (vt_finalize): Likewise.
2267 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
2268 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
2269 (next_nonnote_nondebug_insn_bb): New.
2270 (prev_nonnote_nondebug_insn_bb): New.
2271 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
2272 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
2273 (prev_nonnote_nondebug_insn_bb): Declare.
2274 (next_nonnote_nondebug_insn_bb): Declare.
2275 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
2276 * cfgrtl.c (get_last_bb_insn): Likewise.
2277 * lra.c (push_insns): Likewise.
2279 2017-12-11 David Malcolm <dmalcolm@redhat.com>
2282 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
2284 * selftest.c (selftest::test_runner::test_runner): New ctor.
2285 (selftest::test_runner::~test_runner): New dtor.
2286 * selftest.h (class selftest::test_runner): New class.
2288 2017-12-11 Carl Love <cel@us.ibm.com>
2290 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
2291 vec_extract_fp32_from_shortl]): Add #defines.
2292 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
2293 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
2294 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
2295 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
2296 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
2297 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
2298 * doc/extend.texi: Add documentation for the added builtins.
2300 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
2302 PR rtl-optimization/80693
2303 PR rtl-optimization/81019
2304 PR rtl-optimization/81020
2305 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
2306 are not mentioned in i3. Place the REG_UNUSED note on i2,
2307 possibly modified to REG_DEAD, if it did not originate in i3.
2309 2017-12-11 Jakub Jelinek <jakub@redhat.com>
2311 * recog.c (store_data_bypass_p_1): New function.
2312 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
2313 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
2315 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
2317 PR rtl-optimization/83361
2318 * ifcvt.c (if_convert): Call fixup_partitions.
2320 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
2322 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
2323 early folding of splat_u{8,16,32}.
2325 2017-12-11 Jakub Jelinek <jakub@redhat.com>
2327 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
2328 output_operand_lossage first argument with capital letter.
2329 (aarch64_override_options): Don't start error and sorry first argument
2330 with capital letter.
2332 2017-12-11 Andi Kleen <ak@linux.intel.com>
2334 PR gcov-profile/83355
2335 * auto-profile.c (string_table::get_index_by_decl): Don't
2336 recurse when abstract origin points to itself.
2338 2017-12-11 Bin Cheng <bin.cheng@arm.com>
2340 PR tree-optimization/83320
2341 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
2342 (prune_datarefs_not_in_loop): Ditto.
2344 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
2346 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
2348 2017-12-10 Jakub Jelinek <jakub@redhat.com>
2350 PR tree-optimization/83337
2351 * gimple-loop-interchange.cc (compute_access_stride): Handle
2352 bitfield DRs properly.
2354 2017-12-09 Jakub Jelinek <jakub@redhat.com>
2356 PR tree-optimization/83338
2357 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
2358 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
2359 vector of unsigned integers to vector of signed integers.
2361 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
2363 PR rtl-optimization/83317
2364 * lra-constraints.c (process_address_1): Add insn code check.
2366 2017-12-08 Michael Matz <matz@suse.de>
2368 Fix PR tree-optimization/83323
2369 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
2370 head-controlled loops and loop BBs.
2371 * common.opt (funroll-and-jam): Remove, instead ...
2372 (floop-unroll-and-jam): ... reuse this option.
2373 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
2374 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
2375 (-floop-unroll-and-jam): ... this option.
2377 2017-12-08 Jakub Jelinek <jakub@redhat.com>
2379 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
2380 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
2381 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
2383 2017-12-08 Richard Biener <rguenther@suse.de>
2386 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
2387 handle accesses outside of zero-sized vars.
2389 2017-12-08 Martin Jambor <mjambor@suse.cz>
2391 PR tree-optimization/83141
2392 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
2393 test for MEM_REFs implicitely changing types with padding. Remove
2395 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
2397 2017-12-08 Jakub Jelinek <jakub@redhat.com>
2399 * config/arc/arc.c (arc_attribute_table): Add exclusions to
2401 * config/avr/avr.c (avr_attribute_table): Likewise.
2402 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
2403 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
2404 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
2405 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
2406 of struct attribute_spec.
2407 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
2409 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
2412 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
2414 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
2417 * profile-count.c (profile_count::from_gcov_type): Move from
2418 profile-count.h; handle overflow.
2419 * profile-count.h (profile_count::from_gcov_type): Move offline.
2421 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
2423 PR rtl-optimization/83304
2424 * combine.c (move_deaths): If we do not know where a register died,
2427 2017-12-08 Richard Biener <rguenther@suse.de>
2429 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
2430 Provide -fopt-info-loop feedback when we interchange in a nest.
2432 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
2434 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
2435 for armv6 ARM CPU IDs.
2437 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
2439 * common/config/arm/arm-common.c: Include <algorithm>.
2440 (INCLUDE_VECTOR): Define.
2441 (compare_opt_names): New function.
2442 (arm_rewrite_selected_arch): Only strip out extensions that can be
2443 expressed through -mfpu. Sort the remaining extensions
2446 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
2448 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
2449 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
2450 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
2451 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
2453 2017-06-08 Tristan Gingold <gindold@adacore.com>
2456 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
2457 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
2459 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2461 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
2462 of struct attribute_spec.
2464 2017-12-08 Julia Koval <julia.koval@intel.com>
2466 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
2467 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
2468 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
2469 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
2470 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
2471 _mm_maskz_dpwssds_epi32): Ditto.
2473 2017-12-08 Richard Biener <rguenther@suse.de>
2475 PR tree-optimization/81303
2476 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
2477 conditions try to create a comparison vector type matching
2478 the data vector type.
2479 (vectorizable_condition): Adjust.
2480 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
2481 Leave invariant conditions alone in case we can vectorize those.
2483 2017-12-08 Julia Koval <julia.koval@intel.com>
2485 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
2486 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
2487 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
2488 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
2489 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
2491 2017-12-08 Julia Koval <julia.koval@intel.com>
2493 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
2494 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
2495 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
2496 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
2497 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
2498 _mm_maskz_dpbusds_epi32): New intrinsics.
2500 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
2502 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
2504 (ld<bh>io_signed): Likewise.
2505 (st<bhw>io): Likewise.
2506 * config/nios2/predicates.md (ldstio_memory_operand): Allow
2507 SMALL_INT12 constant integer operand.
2509 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
2511 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
2513 (nios2_large_constant_allowed): ...this. Adjust uses.
2514 (nios2_plus_symbolic_constant_p): Rename to...
2515 (nios2_plus_large_constant_p): ...this. Adjust uses.
2516 (nios2_legitimate_address_p): Correct CONST_INT handling.
2517 (nios2_symbolic_memory_operand_p): Rename to...
2518 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
2519 (nios2_large_constant_p): Check for large constant integers too.
2520 (nios2_split_large_constant): Handle constant integers.
2521 (nios2_split_symbolic_memory_operand): Rename to...
2522 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
2523 (nios2_legitimize_constant_address): Handle constant integers.
2524 (r0rel_constant_p): Handle small constant integers.
2525 (nios2_print_operand_address): Handle r0-relative integer addresses.
2526 * config/nios2/nios2-protos.h: Adjust for renamed functions.
2527 * config/nios2/nios2.md: Adjust for renamed functions.
2529 2017-12-07 Andrew Waterman <andrew@sifive.com>
2531 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
2532 (TARGET_HAVE_SRODATA_SECTION): New define.
2533 (riscv_select_section): New function.
2535 2017-12-08 Joseph Myers <joseph@codesourcery.com>
2536 Alexander Monakov <amonakov@ispras.ru>
2537 Jakub Jelinek <jakub@redhat.com>
2540 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
2542 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2544 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
2545 a bool success value. Don't call output_operand_lossage here.
2546 (aarch64_print_ldpstp_address): Return a bool success value.
2547 (aarch64_print_operand_address): Call output_addr_const if
2548 aarch64_print_address_internal fails.
2549 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
2550 'y'; call output_operand_lossage instead. Call output_operand_lossage
2551 if aarch64_print_ldpstp_address fails.
2553 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2555 * tree-vector-builder.h
2556 (tree_vector_builder::binary_encoded_nelts): Declare.
2557 * tree-vector-builder.c
2558 (tree_vector_builder::binary_encoded_nelts): New function.
2559 * fold-const.c (negate_expr_p): Likewise.
2560 (operand_equal_p, fold_checksum_tree): Likewise.
2561 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
2562 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
2563 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
2564 (uniform_vector_p): Likewise.
2565 * varasm.c (const_hash_1, compare_constant): Likewise.
2566 * tree-ssa-ccp.c: Include tree-vector-builder.h.
2567 (valid_lattice_transition): Operate directly on the VECTOR_CST
2569 * ipa-icf.c: Include tree-vector-builder.h.
2570 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
2571 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
2573 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2575 * tree.c (build_vector): Delete.
2576 * tree.h (build_vector): Make static and move into the self-testing
2579 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2581 * vector-builder.h (vector_builder::derived): New const overload.
2582 (vector_builder::elt): New function.
2583 * tree-vector-builder.h (tree_vector_builder::type): New function.
2584 (tree_vector_builder::apply_step): Declare.
2585 * tree-vector-builder.c (tree_vector_builder::apply_step): New
2587 * gimple-fold.h (tree_vector_builder): Declare.
2588 (gimple_build_vector): Take a tree_vector_builder instead of a
2589 type and vector of elements.
2590 * gimple-fold.c (gimple_build_vector): Likewise.
2591 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
2593 (get_initial_defs_for_reduction): Likewise.
2594 (vectorizable_induction): Likewise.
2596 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2598 * tree-vector-builder.h
2599 (tree_vector_builder::new_binary_operation): Declare.
2600 * tree-vector-builder.c
2601 (tree_vector_builder::new_binary_operation): New function.
2602 * fold-const.c (fold_relational_const): Use it.
2603 (const_binop): Likewise. Check that both input vectors have
2604 the same number of elements, thus excluding things like WIDEN_SUM.
2605 Check whether it is possible to operate directly on the encodings
2608 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2610 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
2611 new_unary_operation, operating only on the encoded elements.
2612 (const_unop): Likewise.
2613 (exact_inverse): Likewise.
2614 (distributes_over_addition_p): New function.
2615 (const_binop): Use tree_vector_builder and new_unary_operation
2616 for combinations of VECTOR_CST and INTEGER_CST. Operate only
2617 on the encoded elements unless the encoding is strided and the
2618 operation does not distribute over addition.
2619 (fold_convert_const): Use tree_vector_builder and
2620 new_unary_operation. Operate only on the encoded elements
2621 for truncating integer conversions, or for non-stepped encodings.
2623 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2625 * config/sparc/sparc.c: Include tree-vector-builder.h.
2626 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
2627 * expmed.c: Include tree-vector-builder.h.
2628 (make_tree): Use tree_vector_builder instead of build_vector.
2629 * fold-const.c: Include tree-vector-builder.h.
2630 (const_binop): Use tree_vector_builder instead of build_vector.
2631 (const_unop): Likewise.
2632 (native_interpret_vector): Likewise.
2633 (fold_vec_perm): Likewise.
2634 (fold_ternary_loc): Likewise.
2635 * gimple-fold.c: Include tree-vector-builder.h.
2636 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
2638 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
2639 (simplify_vector_constructor): Use tree_vector_builder instead
2641 * tree-vect-generic.c: Include tree-vector-builder.h.
2642 (add_rshift): Use tree_vector_builder instead of build_vector.
2643 (expand_vector_divmod): Likewise.
2644 (optimize_vector_constructor): Likewise.
2645 * tree-vect-loop.c: Include tree-vector-builder.h.
2646 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
2647 of build_vector. Explicitly use a stepped encoding for
2649 * tree-vect-slp.c: Include tree-vector-builder.h.
2650 (vect_get_constant_vectors): Use tree_vector_builder instead
2652 (vect_transform_slp_perm_load): Likewise.
2653 (vect_schedule_slp_instance): Likewise.
2654 * tree-vect-stmts.c: Include tree-vector-builder.h.
2655 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
2656 (vect_gen_perm_mask_any): Likewise.
2657 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
2658 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
2660 (build_vector_from_val): Likewise. Explicitly use a duplicate
2663 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
2665 * doc/generic.texi (VECTOR_CST): Describe new representation of
2667 * vector-builder.h: New file.
2668 * tree-vector-builder.h: Likewise.
2669 * tree-vector-builder.c: Likewise.
2670 * Makefile.in (OBJS): Add tree-vector-builder.o.
2671 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
2672 * tree-core.h (tree_base): Add a vector_cst field to the u union.
2673 (tree_vector): Change the number of elements to
2674 vector_cst_encoded_nelts.
2675 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
2676 (VECTOR_CST_ELTS): Delete.
2677 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
2678 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
2679 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
2680 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
2681 (VECTOR_CST_ENCODED_ELT): Likewise.
2682 (vector_cst_encoded_nelts): New function.
2683 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
2684 VECTOR_CST_NELTS_PER_PATTERN as arguments.
2685 (vector_cst_int_elt, vector_cst_elt): Declare.
2686 * tree.c: Include tree-vector-builder.h.
2687 (tree_code_size): Abort if passed VECTOR_CST.
2688 (tree_size): Update for new VECTOR_CST layout.
2689 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
2690 VECTOR_CST_NELTS_PER_PATTERN as arguments.
2691 (build_vector): Use tree_vector_builder.
2692 (vector_cst_int_elt, vector_cst_elt): New functions.
2693 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
2694 encoded elements and then create the vector in the canonical form.
2695 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
2696 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
2697 (tree_c_tests): Call test_vector_cst_patterns.
2698 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
2700 (hash_tree): Likewise.
2701 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
2702 (streamer_write_tree_header): Likewise.
2703 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
2704 (streamer_alloc_tree): Likewise. Update call to make_vector.
2705 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
2707 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2709 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
2710 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
2711 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
2713 2017-12-07 Bin Cheng <bin.cheng@arm.com>
2714 Richard Biener <rguenther@suse.de>
2716 PR tree-optimization/81303
2717 * Makefile.in (gimple-loop-interchange.o): New object file.
2718 * common.opt (floop-interchange): Reuse the option from graphite.
2719 * doc/invoke.texi (-floop-interchange): Ditto. New document for
2720 -floop-interchange and mention it for -O3.
2721 * opts.c (default_options_table): Enable -floop-interchange at -O3.
2722 * gimple-loop-interchange.cc: New file.
2723 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
2724 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
2725 * passes.def (pass_linterchange): New pass.
2726 * timevar.def (TV_LINTERCHANGE): New time var.
2727 * tree-pass.h (make_pass_linterchange): New declaration.
2728 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
2729 interchange. Record IV before/after increment in new parameters.
2730 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
2731 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
2733 (check_reduction_path): ...New function here.
2734 * tree-vectorizer.h (check_reduction_path): New declaration.
2736 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
2739 PR rtl-optimization/80818
2740 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
2741 always early clobbered.
2742 * lra-lives.c (process_bb_lives): Check input hard regs for early
2743 clobbered non-operand hard reg.
2745 2017-12-07 Jakub Jelinek <jakub@redhat.com>
2748 * tree-cfg.c (verify_gimple_assign_binary): Don't require
2749 types_compatible_p, just that TYPE_MODE is the same.
2751 2017-12-07 Martin Sebor <msebor@redhat.com>
2754 * attribs.c (empty_attribute_table): Initialize new member of
2755 struct attribute_spec.
2756 (decl_attributes): Add argument. Handle mutually exclusive
2757 combinations of attributes.
2758 (selftests::test_attribute_exclusions): New function.
2759 (selftests::attribute_c_tests): Ditto.
2760 * attribs.h (decl_attributes): Add default argument.
2761 * selftest.h (attribute_c_tests): Declare.
2762 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
2763 * tree-core.h (attribute_spec::exclusions, exclude): New type and
2765 * doc/extend.texi (Common Function Attributes): Update const and pure.
2766 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
2767 of struct attribute_spec.
2768 * config/arc/arc.c (arc_attribute_table): Same.
2769 * config/arm/arm.c (arm_attribute_table): Same.
2770 * config/avr/avr.c ( avr_attribute_table): Same.
2771 * config/bfin/bfin.c (bfin_attribute_table): Same.
2772 * config/cr16/cr16.c (cr16_attribute_table): Same.
2773 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
2774 * config/h8300/h8300.c (h8300_attribute_table): Same.
2775 * config/i386/i386.c (ix86_attribute_table): Same.
2776 * config/ia64/ia64.c (ia64_attribute_table): Same.
2777 * config/m32c/m32c.c (m32c_attribute_table): Same.
2778 * config/m32r/m32r.c (m32r_attribute_table): Same.
2779 * config/m68k/m68k.c (m68k_attribute_table): Same.
2780 * config/mcore/mcore.c (mcore_attribute_table): Same.
2781 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
2782 * config/mips/mips.c (mips_attribute_table): Same.
2783 * config/msp430/msp430.c (msp430_attribute_table): Same.
2784 * config/nds32/nds32.c (nds32_attribute_table): Same.
2785 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
2786 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
2787 * config/rl78/rl78.c (rl78__attribute_table): Same.
2788 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
2789 * onfig/rx/rx.c (rx_attribute_table): Same.
2790 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
2791 * config/sh/sh.c (sh_attribute_table): Same.
2792 * config/sparc/sparc.c (sparc_attribute_table): Same.
2793 * config/spu/spu.c (spu_attribute_table): Same.
2794 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
2795 * config/v850/v850.c (v850_attribute_table): Same.
2796 * config/visium/visium.c (visium_attribute_table): Same.
2798 2017-12-07 Tamar Christina <tamar.christina@arm.com>
2801 * config/arm/arm.c (INCLUDE_STRING): Define.
2802 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
2803 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
2806 2017-12-07 Michael Matz <matz@suse.de>
2808 Add unroll and jam pass
2810 * gimple-loop-jam.c: New file.
2811 * Makefile.in (OBJS): Add gimple-loop-jam.o.
2812 * common.opt (funroll-and-jam): New option.
2813 * opts.c (default_options_table): Add unroll-and-jam at -O3.
2814 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
2815 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
2816 * passes.def: Add pass_loop_jam.
2817 * timevar.def (TV_LOOP_JAM): Add.
2818 * tree-pass.h (make_pass_loop_jam): Declare.
2819 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
2820 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
2821 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
2822 to flow_loop_tree_node_add.
2823 (duplicate_subloops, copy_loops_to): Append to sibling list.
2824 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
2825 * doc/invoke.texi (-funroll-and-jam): Document new option.
2826 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
2828 2017-12-07 Richard Biener <rguenther@suse.de>
2830 PR tree-optimization/83296
2831 PR tree-optimization/67769
2832 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
2833 flow sensitive info in an unrelated BB.
2834 (value_replacement): Use reset_flow_sensitive_info.
2835 (minmax_replacement): Reset flow sensitive info on the def
2836 we move. Do not reset flow sensitive info in the whole BB
2837 we move the stmt to.
2838 (abs_replacement): Likewise.
2840 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
2843 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
2844 rs6000_cpu to the given -mcpu=, or to the default processor.
2846 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
2848 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
2849 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
2850 instead of rs6000_cpu_attr.
2851 (rs6000_adjust_cost): Ditto.
2852 (is_microcoded_insn): Ditto.
2853 (rs6000_adjust_priority): Ditto.
2854 (rs6000_issue_rate): Ditto.
2855 (rs6000_use_sched_lookahead): Ditto.
2856 (rs6000_use_sched_lookahead_guard): Ditto.
2857 (rs6000_sched_reorder): Ditto.
2858 (force_new_group): Ditto.
2859 * config/rs6000/rs6000.md (cpu attribute): Ditto.
2860 (group_ending_nop): Ditto.
2862 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
2864 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
2865 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
2866 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
2867 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
2869 (rs6000_reassociation_width): Ditto.
2870 (rs6000_emit_epilogue): Ditto.
2871 (rs6000_adjust_cost): Ditto.
2872 (is_microcoded_insn): Ditto.
2873 (is_cracked_insn): Ditto.
2874 (rs6000_adjust_priority): Ditto.
2875 (rs6000_sched_reorder): Ditto.
2876 (rs6000_sched_reorder2): Ditto.
2877 (insn_must_be_first_in_group): Ditto.
2878 (insn_must_be_last_in_group): Ditto.
2879 (rs6000_register_move_cost): Ditto.
2880 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
2883 2017-12-07 Julia Koval <julia.koval@intel.com>
2885 * config.gcc: Add vaesintrin.h.
2886 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
2888 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
2889 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
2891 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
2892 * config/i386/immintrin.h: Include vaesintrin.h.
2893 * config/i386/sse.md (vaesdec_<mode>): New pattern.
2894 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
2895 _mm_aesdec_epi128): New intrinsics.
2897 2017-12-06 David Malcolm <dmalcolm@redhat.com>
2899 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
2900 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
2901 c-family/c-spellcheck.cc.
2902 (best_macro_match::best_macro_match): Likewise.
2904 (struct edit_distance_traits<cpp_hashnode *>): Move to
2905 c-family/c-spellcheck.h.
2906 (class best_macro_match): Likewise.
2908 2017-12-06 Jakub Jelinek <jakub@redhat.com>
2910 PR tree-optimization/83293
2911 * gimple-ssa-strength-reduction.c (insert_initializers): Use
2912 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
2913 might insert into empty bb.
2916 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
2917 simplify for plus with :c added, and pointer_plus without that.
2918 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
2919 with undefined overflow and the conversion is not widening,
2920 perform negation in utype and only convert to type afterwards.
2921 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
2922 simplify for plus with :c added, and pointer_plus without that.
2923 If type is integral with undefined overflow and the conversion is
2924 not widening, perform minus in utype and only convert to type
2925 afterwards. Move the last pointer_diff_expr simplify into the
2928 2017-12-06 Martin Sebor <msebor@redhat.com>
2930 PR tree-optimization/82646
2931 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
2932 strncpy, not maxlen.
2934 2017-12-06 Martin Sebor <msebor@redhat.com>
2936 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
2939 PR tree-optimization/83075
2940 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
2941 strncat/strncpy don't change length of source string.
2943 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
2946 2017-11-29 Martin Aberg <maberg@gaisler.com>
2948 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
2949 to prevent b2bst errata sequence.
2950 (sqrtdf2_fix): Likewise.
2952 2017-12-06 Jakub Jelinek <jakub@redhat.com>
2954 PR tree-optimization/81945
2955 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
2956 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
2957 to dest_cfun has orig_loop_num set, either remap it to the new
2958 loop number if the loop got moved too, or clear it.
2960 2017-12-05 Steve Ellcey <sellcey@cavium.com>
2962 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
2964 (thunderx2t99_nothing): New insn reservation.
2965 (thunderx2t99_mrs): New insn reservation.
2966 (thunderx2t99_multiple): New insn reservation.
2967 (thunderx2t99_alu_basi): Add bfx to reservation.
2968 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
2970 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
2973 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
2975 2017-12-05 Bin Cheng <bin.cheng@arm.com>
2977 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
2978 tree-ssa-pre.c::remove_dead_inserted_code.
2979 * tree-ssa-dce.h: New file.
2980 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
2981 (remove_dead_inserted_code): Move and rename to function
2982 tree-ssa-dce.c::simple_dce_from_worklist.
2983 (pass_pre::execute): Update use.
2985 2017-12-05 Richard Biener <rguenther@suse.de>
2987 PR tree-optimization/83277
2988 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
2989 to code-gen liveout vars.
2991 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
2993 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
2994 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
2995 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
2998 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
3000 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
3002 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
3003 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
3006 2017-12-05 Richard Biener <rguenther@suse.de>
3008 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
3009 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
3010 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
3011 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
3013 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3015 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
3017 2017-12-05 Martin Liska <mliska@suse.cz>
3018 Jakub Jelinek <jakub@redhat.com>
3020 * doc/invoke.texi: Document the options.
3021 * flag-types.h (enum sanitize_code): Add
3022 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
3023 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
3024 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
3025 * opts.c: Define new sanitizer options.
3026 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
3027 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
3029 2017-12-05 Julia Koval <julia.koval@intel.com>
3031 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
3032 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
3033 (ix86_handle_option): Handle -mavx512vnni.
3034 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
3035 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
3036 * config/i386/i386-c (__AVX512VNNI__): New.
3037 * config/i386/i386.c (ix86_target_string): Handle new option.
3038 (ix86_valid_target_attribute_inner_p): Handle new option.
3039 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
3040 * config/i386/i386.opt (mavx512vnni): New option.
3042 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
3045 * athlon.md: Disable for generic.
3046 * haswell.md: Enable for generic.
3047 * i386.c (ix86_sched_init_global): Add core hooks for generic.
3048 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
3050 (ix86_adjust_cost): Move generic to haswell path.
3052 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
3054 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
3055 instead of MEM_P in a couple more places. Fix formatting issues.
3057 2017-12-04 Jim Wilson <jimw@sifive.com>
3059 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
3060 instead of GP_REG_LAST-1.
3061 (riscv_adjust_libcall_cfi_prologue): Likewise.
3062 (riscv_adjust_libcall_cri_epilogue): Likewise.
3063 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
3066 2017-12-04 Luis Machado <luis.machado@linaro.org>
3068 * ipa-pure-const.c (check_decl): Add missing newline.
3069 (state_from_flags): Likewise.
3071 2017-12-04 Jeff Law <law@redhat.com>
3073 PR tree-optimizatin/78496
3074 * gimple-ssa-evrp-analyze.h
3075 (evrp_range_analyzer::get_vr_values): Simplify.
3076 * gimple-ssa-evrp-analyze.c: Corresponding changes.
3077 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
3078 and gimple-ssa-evrp-analyze.h.
3079 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
3080 (simplify_stmt_for_jump_threading): Copy a blob of code from
3081 tree-vrp.c to use ranges to simplify statements.
3082 (dom_opt_dom_walker::before_dom_children): Call
3083 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
3084 (dom_opt_dom_walker::after_dom_children): Similarly for
3085 evrp_range_analyzer::leave.
3086 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
3089 * gimple-ssa-evrp-analyze.c
3090 (evrp_range_analyzer::extract_range_from_stmt): Always use
3091 vr_values::update_value_range so preexisting range info is
3092 medged with new range info, even if the new range is VR_VARYING.
3094 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
3096 * combine.c: Adjust comment.
3097 (use_crosses_set_p): Delete.
3098 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
3099 (try_combine): Ditto.
3101 2017-12-04 Richard Biener <rguenther@suse.de>
3103 PR tree-optimization/83255
3104 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
3105 Re-add zero-iteration check.
3107 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
3109 PR rtl-optimization/83245
3110 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
3111 hard registers as earlyclobber, also if not in an asm.
3113 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
3117 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3120 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
3121 rs6000_cpu based on cpu_index, not tune_index.
3123 2017-12-04 Richard Biener <rguenther@suse.de>
3125 PR tree-optimization/83238
3126 * graphite-scop-detection.c (scop_detection::merge_sese): Make
3127 code match comment, rejecting invalid SESE regions.
3129 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
3131 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
3132 require base operand is a REG_POINTER prior to reload on targets
3133 with non-equivalent space registers.
3135 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
3137 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
3138 (update_specialized_profile): Fix updating of counts.
3139 (perhaps_add_new_callers): Likewise.
3141 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
3144 * x86-tune.def: Remove obsolette FIXMEs.
3145 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
3146 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
3147 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
3149 (X86_TUNE_PAD_RETURNS): Disable for generic.
3151 2017-12-02 Jakub Jelinek <jakub@redhat.com>
3153 PR tree-optimization/83170
3154 PR tree-optimization/83241
3155 * gimple-ssa-store-merging.c
3156 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
3157 gimple_vuse (ins_stmt) in case it has changed.
3158 (imm_store_chain_info::output_merged_store): Likewise.
3160 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
3164 * tree-cfg.c (pass_warn_function_return::execute): Handle
3165 __builtin_ubsan_handle_missing_return like __builtin_unreachable
3166 with BUILTINS_LOCATION.
3170 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
3171 is BLKmode for vector field with vector raw mode, use TYPE_MODE
3172 instead of DECL_MODE.
3174 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
3175 last argument to rtx pointer.
3176 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
3177 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
3178 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
3179 depending on the chosen ISAs.
3180 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
3181 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
3182 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
3184 * config/i386/sse.md (mov<mode>_internal): Likewise.
3185 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
3187 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3189 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
3191 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3193 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
3194 parameters from prototype.
3195 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
3196 parameters. Don't print a comment.
3197 (emit_fusion_gpr_load): Adjust.
3198 (emit_fusion_load_store): Adjust.
3199 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
3200 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
3201 comment on the second line.
3203 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3206 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
3207 rs6000_cpu based on cpu_index, not tune_index.
3209 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3211 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
3212 which_alternative instead of which_alternative + 1.
3213 (output_asm_insn): Print an extra tab if the template is short.
3215 2017-12-01 Jim Wilson <jimw@sifive.com>
3217 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
3219 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
3220 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
3222 * doc/tm.texi: Regenerate.
3224 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
3227 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
3228 whether we can allocate pseudos before trying to fix an address.
3229 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
3230 memory address is indexed or indirect.
3231 (floatuns_<mode>si2_hw2): Likewise.
3233 2017-12-01 Jason Merrill <jason@redhat.com>
3235 * Makefile.in (TAGS): Add c-family/*.cc.
3237 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
3239 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
3240 (call_value_insn): Likewise.
3241 (sibcall_insn): Likewise.
3242 (sibcall_value_insn): Likewise.
3243 (movsi_aarch64): Likewise.
3244 (movdi_aarch64): Likewise.
3245 (add_losym_): Likewise.
3246 (ldr_got_small_): Likewise.
3247 (ldr_got_small_sidi): Likewise.
3248 (ldr_got_small_28k_): Likewise.
3249 (ldr_got_small_28k_sidi): Likewise.
3250 * config/aarch64/aarch64.c (aarch64_print_address_internal):
3251 Move output_addr_const to symbolic case. Add error check.
3253 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3255 * config/s390/predicates.md (plus16_Q_operand): New predicate.
3256 * config/s390/s390.md: Disable MVC merging peephole if it would
3257 disable operand forwarding.
3258 (new peephole2): Split MVCs if it would turn them into up to 2
3261 2017-12-01 Richard Biener <rguenther@suse.de>
3263 PR tree-optimization/83232
3264 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
3265 detection of same access. Instead of breaking the group here
3266 do not consider the duplicate. Add comment explaining real fix.
3268 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
3270 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
3272 2017-12-01 Sudakshina Das <sudi.das@arm.com>
3274 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
3276 2017-12-01 Jakub Jelinek <jakub@redhat.com>
3278 * function.h (struct function): Remove cilk_frame_decl,
3279 is_cilk_function and calls_cilk_spawn fields.
3280 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
3282 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
3283 cilk_elemental field.
3284 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
3285 * target.def: Adjust comment.
3286 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
3287 Don't test cilk_elemental.
3289 PR tree-optimization/83233
3290 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
3291 bswap_stat name for the struct.
3294 * tree.h (SWITCH_BREAK_LABEL_P): Define.
3295 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
3296 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
3297 SWITCH_BREAK_LABEL_P set on the label.
3298 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
3299 added for default case if it was missing and not all cases covered.
3300 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
3301 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
3303 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
3304 to avoid -Wimplicit-fallthrough warning.
3305 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
3306 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
3308 PR tree-optimization/83221
3309 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
3311 (init_reassoc): Formatting fix.
3314 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
3315 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
3319 * tree-cfg.c: Include asan.h.
3320 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
3322 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
3324 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
3326 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
3328 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
3329 vpcmpeqd instruction.
3331 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
3333 * config/i386/i386.c (standard_sse_constant_opcode): Fix
3334 registers type for 128bit mode.
3336 2017-11-30 Jakub Jelinek <jakub@redhat.com>
3338 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
3340 * gengtype-state.c (read_state_pair): Likewise.
3341 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
3342 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
3343 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
3344 * ifcvt.c (noce_try_store_flag_constants): Likewise.
3345 * tree-ssa-ccp.c (ccp_finalize): Likewise.
3346 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
3347 * builtins.c (fold_builtin_3): Likewise.
3348 * graphite-scop-detection.c
3349 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
3350 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
3353 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
3356 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
3357 parameter. Return true if ignore_decl_rtl_set_p is true and other
3358 conditions are satisfied.
3359 * asan.h (asan_protect_global): Add new parameter.
3360 * varasm.c (categorize_decl_for_section): Pass true as second parameter
3361 to asan_protect_global calls.
3363 2017-11-30 Jim Wilson <jimw@sifive.com>
3365 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
3366 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
3367 -mstrict-align, add info on default value. Delete redundant lines for
3368 -mabi. Add missing -mexplicit-relocs docs.
3370 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
3372 * config/arc/arc.md (trap): New pattern.
3374 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
3376 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
3377 instruction to end into a delay slot.
3378 * config/arc/arc.md (cond_delay_insn): Check if the instruction
3379 can be placed into a delay slot against reg_note.
3381 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
3383 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
3384 labels number of usages.
3386 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
3388 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
3390 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
3392 2017-11-30 Jakub Jelinek <jakub@redhat.com>
3395 * internal-fn.c (expand_mul_overflow): Optimize unsigned
3396 multiplication by power of 2 constant into two shifts + comparison.
3398 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
3401 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
3403 2017-11-30 Richard Biener <rguenther@suse.de>
3405 PR tree-optimization/83202
3406 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
3407 allow_peel argument and guard peeling.
3408 (canonicalize_loop_induction_variables): Likewise.
3409 (canonicalize_induction_variables): Pass false.
3410 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
3411 peeling from cunrolli.
3413 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
3415 * combine.c (try_combine): Print a message to dump file whenever
3416 I0, I1, or I2 cannot be combined into I3.
3418 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
3420 PR rtl-optimization/83156
3421 PR rtl-optimization/82621
3422 * combine.c (try_combine): Don't split an I2 if one of the dests is
3423 set again before I3. Allow unused dests.
3425 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
3427 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
3429 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
3431 PR rtl-optimization/80818
3432 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
3433 recursively. Use insn code for clobber.
3434 (lra_set_insn_recog_data): Pass the new arg to
3435 collect_non_operand_hard_regs.
3436 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
3438 (lra_update_insn_regno_info): Pass insn to
3439 add_regs_to_insn_regno_info.
3441 2017-11-29 Jim Wilson <jimw@sifive.com>
3442 Andrew Waterman <andrew@sifive.com>
3444 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
3445 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
3446 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
3447 use SINGLE_SHIFT_COST.
3448 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
3449 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
3451 2017-11-29 Julia Koval <julia.koval@intel.com>
3453 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
3454 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
3455 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
3456 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
3457 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
3458 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
3459 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
3460 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
3461 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
3462 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
3463 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
3464 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
3465 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
3466 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
3467 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
3468 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
3469 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
3470 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
3471 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
3472 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
3473 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
3474 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
3475 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
3476 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
3477 __builtin_ia32_vpshldv_v2di_mask,
3478 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
3479 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
3480 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
3482 2017-11-29 Julia Koval <julia.koval@intel.com>
3484 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
3485 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
3486 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
3487 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
3488 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
3489 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
3490 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
3491 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
3492 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
3493 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
3494 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
3495 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
3496 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
3497 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
3498 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
3499 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
3500 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
3501 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
3502 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
3503 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
3504 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
3506 2017-11-29 Daniel Cederman <cederman@gaisler.com>
3508 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
3509 movsi_pic_gotdata_op instruction as a load for the UT699 errata
3512 2017-11-29 Martin Aberg <maberg@gaisler.com>
3514 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
3515 to prevent b2bst errata sequence.
3516 (sqrtdf2_fix): Likewise.
3518 2017-11-29 Daniel Cederman <cederman@gaisler.com>
3520 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
3521 (div_sqrt_insn_p): New function.
3522 (sparc_do_work_around_errata): Insert NOP instructions to
3523 prevent sequences that could trigger the TN-0013 errata for
3524 certain LEON3 processors.
3525 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
3526 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
3527 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
3528 (in_branch_delay): Prevent div and sqrt in delay slot if
3530 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
3532 2017-11-29 Daniel Cederman <cederman@gaisler.com>
3534 * config/sparc/sparc.c (atomic_insn_p): New function.
3535 (sparc_do_work_around_errata): Insert NOP instructions to
3536 prevent sequences that could trigger the TN-0010 errata for
3538 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
3539 instruction referable in atomic_insns_p.
3541 2017-11-29 Daniel Cederman <cederman@gaisler.com>
3543 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
3544 (atomic_compare_and_swap_leon3_1): Likewise.
3547 2017-11-29 Daniel Cederman <cederman@gaisler.com>
3549 * config/sparc/sparc.c (fpop_insn_p): New function.
3550 (sparc_do_work_around_errata): Insert NOP instructions to
3551 prevent sequences that could trigger the TN-0012 errata for
3553 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
3554 * config/sparc/sparc.md (fix_gr712rc): New attribute.
3555 (in_branch_annul_delay): Prevent floating-point instructions
3556 in delay slot of annulled integer branch.
3558 2017-11-29 Richard Biener <rguenther@suse.de>
3560 PR tree-optimization/83202
3561 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
3563 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
3564 nodes vectorized to the same stmts multiple times.
3565 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
3566 (vect_analyze_slp_instance): Adjust.
3567 (scalar_stmts_to_slp_tree_map_t): New typedef.
3568 (vect_schedule_slp_instance): Add a map recording the SLP node
3569 representing the vectorized stmts for a set of scalar stmts.
3570 Avoid code-generating redundancies.
3571 (vect_schedule_slp): Allocate map and pass it down.
3573 2017-11-29 Nathan Sidwell <nathan@acm.org>
3576 * tree.c (build_complex_type): Fix canonicalization. Only fill in
3579 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
3581 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
3582 cases for printing LDP/STP memory addresses.
3583 (aarch64_print_address_internal): Renamed from
3584 aarch64_print_operand_address, added parameter, add Pmode check.
3585 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
3586 (aarch64_print_operand_address): Indirect to
3587 aarch64_print_address_internal.
3588 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
3589 'y' operand output specifier.
3591 2017-11-29 Jakub Jelinek <jakub@redhat.com>
3594 * tree.c (build_simple_mem_ref_loc): Handle
3595 get_addr_base_and_unit_offset returning a MEM_REF.
3598 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
3601 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
3604 2017-11-28 David Malcolm <dmalcolm@redhat.com>
3606 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
3608 (layout::move_to_column): Likewise.
3610 2017-11-28 Jakub Jelinek <jakub@redhat.com>
3613 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
3614 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
3616 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3617 Martin Jambor <mjambor@suse.cz>
3620 * tree.h (expr_type_first_operand_type_p): Declare
3621 * tree.c (expr_type_first_operand_type_p): New function.
3622 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
3623 (ipa_value_from_jfunc): Adjust declaration.
3624 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
3625 Use it as result type for arithmetics, unless it is NULL in which case
3626 be more conservative.
3627 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
3628 ipa_get_jf_pass_through_result.
3629 (propagate_vals_across_pass_through): Likewise.
3630 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
3631 is to propagate_vals_across_pass_through.
3632 (propagate_constants_across_call): Pass PARM_TYPE to
3633 propagate_scalar_across_jump_function.
3634 (find_more_scalar_values_for_callers_subset): Pass parameter type to
3635 ipa_value_from_jfunc.
3636 (cgraph_edge_brings_all_scalars_for_node): Likewise.
3637 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
3638 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
3639 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
3640 target_type, pass it to ipa_value_from_jfunc.
3641 (update_indirect_edges_after_inlining): Pass parameter type to
3642 try_make_edge_direct_simple_call.
3644 2017-11-28 Jeff Law <law@redhat.com>
3646 * gimple-ssa-evrp-analyze.c
3647 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
3648 refine ranges if scev_initialized_p returns true.
3649 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
3651 2017-11-28 Julia Koval <julia.koval@intel.com>
3653 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
3654 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
3655 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
3656 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
3657 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
3658 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
3659 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
3660 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
3661 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
3662 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
3663 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
3664 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
3665 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
3666 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
3667 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
3668 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
3669 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
3670 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
3671 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
3672 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
3673 __builtin_ia32_vpshrd_v2di_mask): New builtins.
3674 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
3676 2017-11-28 Julia Koval <julia.koval@intel.com>
3678 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
3679 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
3680 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
3681 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
3682 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
3683 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
3684 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
3685 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
3686 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
3687 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
3688 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
3689 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
3690 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
3691 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
3692 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
3693 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
3694 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
3695 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
3696 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
3697 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
3698 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
3699 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
3700 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
3701 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
3702 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
3703 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
3704 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
3705 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
3706 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
3707 __builtin_ia32_vpshld_v2di_mask): New builtins.
3708 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
3709 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
3711 2017-11-28 Richard Biener <rguenther@suse.de>
3713 PR tree-optimization/80776
3714 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
3716 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
3718 (evrp_range_analyzer::record_ranges_from_incoming_edges):
3719 If the incoming edge is an effective fallthru because the other
3720 edge only reaches a __builtin_unreachable () then record ranges
3721 derived from the controlling condition in SSA info.
3722 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
3723 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
3725 2017-11-28 Olivier Hainque <hainque@adacore.com>
3727 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
3728 on mingw build hosts.
3730 2017-11-28 Jakub Jelinek <jakub@redhat.com>
3733 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
3735 2017-11-28 Richard Biener <rguenther@suse.de>
3738 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
3739 copies generated from memcpy use a character array as reference
3742 2017-11-28 Julia Koval <julia.koval@intel.com>
3743 Sebastian Peryt <sebastian.peryt@intel.com>
3745 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
3746 c-family/c-cilkplus.o, c-family/array-notation-common.o,
3747 cilk-common.o, cilk.h, cilk-common.c): Remove.
3749 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
3750 * builtins.c (is_builtin_name): Remove cilkplus condition.
3751 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
3752 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
3753 cilk-builtins.def, cilkplus.def): Remove.
3754 * cif-code.def (CILK_SPAWN): Remove.
3755 * cilk-builtins.def: Delete.
3756 * cilk-common.c: Ditto.
3758 * cilkplus.def: Ditto.
3759 * config/darwin.h (fcilkplus): Delete.
3760 * cppbuiltin.c: Ditto.
3761 * doc/extend.texi: Remove cilkplus doc.
3762 * doc/generic.texi: Ditto.
3763 * doc/invoke.texi: Ditto.
3764 * doc/passes.texi: Ditto.
3765 * gcc.c (fcilkplus): Remove.
3766 * gengtype.c (cilk.h): Remove.
3767 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
3769 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
3771 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
3773 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
3774 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
3775 cilkplus conditions.
3776 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
3777 inline_read_section): Ditto.
3778 * ipa-inline-analysis.c (cilk.h): Remove.
3779 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
3780 * lto-wrapper.c (merge_and_complain, append_compiler_options,
3781 append_linker_options): Remove condition for fcilkplus.
3782 * lto/lto-lang.c (cilk.h): Remove.
3783 (lto_init): Remove condition for fcilkplus.
3784 * omp-expand.c (expand_cilk_for_call): Delete.
3785 (expand_omp_taskreg, expand_omp_for_static_chunk,
3786 expand_omp_for): Remove cilkplus
3788 (expand_cilk_for): Delete.
3789 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
3790 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
3791 execute_lower_omp, diagnose_sb_0): Ditto.
3792 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
3793 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
3794 * tree-nested.c: Ditto.
3795 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
3796 (dump_generic_node): Ditto.
3797 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
3798 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
3800 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
3802 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
3804 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
3805 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
3806 for vector mode and !TARGET_SIMD.
3808 2017-11-28 Jakub Jelinek <jakub@redhat.com>
3810 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
3812 * tree.h (SWITCH_LABELS): Remove.
3813 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
3814 assert SWITCH_BODY is non-NULL.
3815 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
3817 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
3819 PR tree-optimization/80788
3820 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
3821 has TREE_OVERFLOW set, call drop_tree_overflow.
3823 2017-11-28 Richard Biener <rguenther@suse.de>
3825 PR tree-optimization/83158
3826 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
3828 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
3831 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
3832 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
3834 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
3836 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
3837 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
3839 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
3842 * varasm.c (compare_constant): Take the mode of the constants into
3843 account when comparing floating point constants.
3845 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
3847 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
3848 from explicit instantiation of debug_helper.
3849 * vec.h (DEFINE_DEBUG_VEC): Ditto.
3851 2017-11-27 Richard Biener <rguenther@suse.de>
3853 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
3856 2017-11-27 Richard Biener <rguenther@suse.de>
3858 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
3859 (type_hash_canon): Also clear int_cst_hash_table entry for
3861 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
3863 2017-11-27 Tamar Christina <tamar.christina@arm.com>
3865 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
3866 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
3867 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
3869 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
3871 * hash-map.h (gt_cleare_cache): Avoid UB.
3873 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
3875 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
3876 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
3877 (decide_unroll_runtime_iterations): Likewise.
3878 (decide_unroll_stupid): Likewise.
3880 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3883 * config/i386/i386.md: Add a loop with incssp.
3885 2017-11-27 Martin Jambor <mjambor@suse.cz>
3887 PR tree-optimization/81248
3888 * tree-sra.c (splice_param_accesses): Remove size check.
3889 (decide_one_param_reduction): Fix size check.
3890 * gimple-pretty-print.c (dump_profile): Silence warning.
3891 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
3893 2017-11-27 Jakub Jelinek <jakub@redhat.com>
3896 * dbxout.c (lastlineno): New variable.
3897 (dbx_debug_hooks): Use dbxout_switch_text_section as
3898 switch_text_section debug hook.
3899 (dbxout_function_end): Switch to current_function_section
3900 rather than function_section. If crtl->has_bb_partition,
3901 output just one N_FUN, depending on in_cold_section_p.
3902 (dbxout_source_line): Remember last lineno in lastlineno.
3903 (dbxout_switch_text_section): New function.
3904 (dbxout_function_decl): Adjust dbxout_block caller.
3905 (dbx_block_with_cold_children): New function.
3906 (dbxout_block): Return true if any LBRAC/RBRAC have been
3907 emitted. Use dbx_block_with_cold_children at depth == 0
3908 in second partition. Add PARENT_BLOCKNUM argument, pass
3909 it optionally adjusted to children. Output LBRAC/RBRAC
3910 around recursive call only if the block is in the current
3911 partition, if not and anything was output, emit empty
3913 * final.c (final_scan_insn): Compute cold_function_name
3914 before calling switch_text_section debug hook. Call
3915 that hook even if dwarf2out_do_frame if not emitting
3919 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
3920 TREE_READONLY decls.
3922 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
3924 PR rtl-optimization/82488
3925 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
3927 2017-11-26 Julia Koval <julia.koval@intel.com>
3929 * config/i386/i386.c (processor_target_table): Add skylake_cost for
3931 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
3934 2017-11-26 Julia Koval <julia.koval@intel.com>
3936 * config/i386/driver-i386.c (host_detect_local_cpu):
3937 Detect skylake-avx512.
3939 2017-11-26 Julia Koval <julia.koval@intel.com>
3941 * config.gcc: Add -march=cannonlake.
3942 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
3943 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
3944 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
3945 (PTA_CANNONLAKE): New.
3946 (processor_target_table): Add cannonlake.
3947 (ix86_option_override_internal): Ditto.
3948 (fold_builtin_cpu): Ditto.
3949 (get_builtin_code_for_version): Handle cannonlake.
3950 (M_INTEL_COREI7_CANNONLAKE): New.
3951 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
3952 * doc/invoke.texi: Add -march=cannonlake.
3954 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
3956 * plugin.c (add_new_plugin): Use platform-specific library extensions.
3957 (try_init_one_plugin): Alternative implementation for MinGW.
3958 * Makefile.in (plugin_implib): New.
3959 (gengtype-lex.c): Fix broken AIX workaround.
3960 * configure: Regenerate.
3961 * doc/plugins.texi: Document support for MinGW.
3963 2017-11-25 Jakub Jelinek <jakub@redhat.com>
3965 PR rtl-optimization/81553
3966 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
3967 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
3968 is a shift where C1 has different mode than the whole shift, use C1's
3969 mode for MULT rather than the shift's mode.
3972 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
3973 builtins not enabled in the currently selected ISA.
3975 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
3977 PR tree-optimization/71026
3978 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
3979 (insert_reciprocals): Change to insert reciprocals before a division
3980 by a square and to insert the square of a reciprocal.
3981 (execute_cse_reciprocals_1): Change to consider division by a square.
3982 (register_division_in): Add importance parameter.
3984 2017-11-24 Richard Biener <rguenther@suse.de>
3986 PR tree-optimization/82402
3987 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
3988 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
3990 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
3992 * match.pd (0-ptr): New transformation.
3994 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
3997 * ipa-inline.c (inline_small_functions): Set current badnes correctly
3998 when skipping checking.
4000 2017-11-24 Richard Biener <rguenther@suse.de>
4002 PR tree-optimization/83128
4003 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
4004 (vn_reference_lookup_3): Likewise.
4006 2017-11-24 Jakub Jelinek <jakub@redhat.com>
4009 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
4010 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
4013 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
4014 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
4015 check for the latter with gimple_call_builtin_p. Do not handle
4016 BUILT_IN_STPNCPY_CHK which is not a pass through call.
4018 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
4020 * config/arm/arm_neon.h: Fix pragma GCC push_options before
4023 2017-11-23 Julia Koval <julia.koval@intel.com>
4025 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
4026 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
4027 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
4028 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
4029 _mm512_maskz_expandloadu_epi16): New intrinsics.
4030 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
4031 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
4032 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
4033 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
4034 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
4035 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
4036 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
4037 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
4038 _mm256_maskz_expandloadu_epi8): New intrinsics.
4039 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
4040 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
4041 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
4042 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
4043 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
4044 * config/i386/sse.md (VI248_VLBW): New iterator.
4045 (expand<mode>_mask, expand<mode>_maskz): New patterns.
4047 2017-11-23 Julia Koval <julia.koval@intel.com>
4049 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
4050 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
4051 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
4052 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
4053 _mm512_mask_compressstoreu_epi16): New.
4054 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
4055 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
4056 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
4057 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
4058 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
4059 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
4060 _mm256_mask_compressstoreu_epi8): New.
4061 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
4062 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
4063 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
4064 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
4065 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
4066 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
4067 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
4068 __builtin_ia32_compresshi128_mask,
4069 __builtin_ia32_compressstoreuqi512_mask,
4070 __builtin_ia32_compressstoreuhi512_mask,
4071 __builtin_ia32_compressstoreuqi256_mask,
4072 __builtin_ia32_compressstoreuqi128_mask,
4073 __builtin_ia32_compressstoreuhi256_mask,
4074 __builtin_ia32_compressstoreuhi128_mask): New builtins.
4075 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
4077 (ix86_expand_special_args_builtin): Handle new types.
4078 (s4fma_expand): Handle new builtin array.
4079 * config/i386/immintrin.h: Include new headers.
4080 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
4081 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
4083 2017-11-23 Jakub Jelinek <jakub@redhat.com>
4085 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
4086 fixes. Declare temp and g variables at the top in order to avoid
4087 {} in most of the cases.
4089 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
4091 * match.pd (ptr-0): New transformation.
4093 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
4095 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
4096 (struct addr_mode_cost_table): New.
4097 (struct tune_params): Add field addr_mode_costs.
4098 * config/arm/arm.c (generic_addr_mode_costs): New.
4099 (arm_slowmul_tune): Initialise addr_mode_costs field.
4100 (arm_fastmul_tune): Likewise.
4101 (arm_strongarm_tune): Likewise.
4102 (arm_xscale_tune): Likewise.
4103 (arm_9e_tune): Likewise.
4104 (arm_marvell_pj4_tune): Likewise.
4105 (arm_v6t2_tune): Likewise.
4106 (arm_cortex_tune): Likewise.
4107 (arm_cortex_a8_tune): Likewise.
4108 (arm_cortex_a7_tune): Likewise.
4109 (arm_cortex_a15_tune): Likewise.
4110 (arm_cortex_a35_tune): Likewise.
4111 (arm_cortex_a53_tune): Likewise.
4112 (arm_cortex_a57_tune): Likewise.
4113 (arm_exynosm1_tune): Likewise.
4114 (arm_xgene1_tune): Likewise.
4115 (arm_cortex_a5_tune): Likewise.
4116 (arm_cortex_a9_tune): Likewise.
4117 (arm_cortex_a12_tune): Likewise.
4118 (arm_cortex_a73_tune): Likewise.
4119 (arm_v7m_tune): Likewise.
4120 (arm_cortex_m7_tune): Likewise.
4121 (arm_v6m_tune): Likewise.
4122 (arm_fa726te_tune): Likewise.
4123 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
4125 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
4127 * config/arm/arm.c (arm_mem_costs): New function.
4128 (arm_rtx_costs_internal): Use arm_mem_costs.
4130 2017-11-23 Mark Wielaard <mark@klomp.org>
4132 * dwarf2out.c (init_sections_and_labels): Use generation to create
4133 unique ranges_section_label and ranges_base_label. Return generation.
4134 (output_rnglists): Add generation argument. Use generation to create
4135 unique ranges labels.
4136 (dwarf2out_finish): Get generation from init_sections_and_labels
4137 and pass generation to output_rnglists.
4139 2017-11-23 Mike Stump <mikestump@comcast.net>
4140 Eric Botcazou <ebotcazou@adacore.com>
4142 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
4143 * cfgloop.h (struct loop): Add unroll field.
4144 * function.h (struct function): Add has_unroll bitfield.
4145 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
4146 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
4147 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
4148 (pass_rtl_unroll_loops::gate): Likewise.
4149 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
4150 for which loop->unroll==1.
4151 (decide_unroll_constant_iterations): Use note for consistency and
4152 take loop->unroll into account. Return early if loop->unroll is set.
4153 Fix thinko in existing test.
4154 (decide_unroll_runtime_iterations): Use note for consistency and
4155 take loop->unroll into account.
4156 (decide_unroll_stupid): Likewise.
4157 * lto-streamer-in.c (input_cfg): Read loop->unroll.
4158 * lto-streamer-out.c (output_cfg): Write loop->unroll.
4159 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
4161 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
4162 (print_loop): Print loop->unroll if set.
4163 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
4164 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
4165 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
4167 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
4168 loop->unroll is set and smaller than the trip count. Otherwise bypass
4169 entirely the heuristics if loop->unroll is set. Remove dead note.
4170 Fix off-by-one bug in other note.
4171 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
4172 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
4174 (tree_unroll_loops_completely): Make static.
4175 (pass_complete_unroll::execute): Use correct type for variable.
4176 (pass_complete_unrolli::execute): Fix formatting.
4177 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
4179 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
4181 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
4182 enable when TARGET_PREFER_AVX128 is set.
4184 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
4186 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
4188 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
4189 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
4191 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
4193 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
4194 frequencies. Fix estimation of aggregate parameters.
4196 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
4198 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
4201 2017-11-23 Tom de Vries <tom@codesourcery.com>
4203 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
4205 2017-11-23 Tom de Vries <tom@codesourcery.com>
4207 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
4210 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
4213 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
4214 sibcall_value_pcrel_fdpic): Use local variable instead of
4216 (calli_tbr_rel): Add missing operand 2.
4217 (call_valuei_tbr_rel): Add missing operand 3.
4219 2017-11-23 Jakub Jelinek <jakub@redhat.com>
4222 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
4223 bitpos/bitsize covering the whole destination, use store_expr only if
4224 the complex mode is the same. Otherwise, use expand_normal and if
4225 it returns CONCAT, subreg each part separately instead of trying to
4226 subreg the whole result.
4228 2017-11-23 Richard Biener <rguenther@suse.de>
4230 PR tree-optimization/23094
4231 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
4232 come here from walking over backedges in the first iteration.
4233 (vn_reference_lookup_3): Skip clobbers that store the same value.
4235 2017-11-23 Richard Biener <rguenther@suse.de>
4237 PR tree-optimization/81403
4238 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
4239 a block we need a leader relative to.
4240 (phi_translate_1): For nary processing require a leader from
4241 get_representative_for given we run expression simplification
4242 using match-and-simplify. Remove previous fix.
4244 2017-11-22 Jeff Law <law@redhat.com>
4246 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
4247 Use new method allocate_value_range rather than accessing the
4248 vrp_value_range_pool data member directly.
4249 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
4250 to use extract_range_from_stmt method to avoid need for
4251 extract_range_from_assignment method.
4252 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
4253 method rather than setting values_propgated data member directly.
4254 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
4255 and values propagated data members and extract_range_from_assignment
4256 method. Reorder private data members to conform to standards.
4257 Add new methods set_lattice_propagation_complete and
4258 allocate_value_range.
4260 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
4262 PR rtl-optimization/83030
4263 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
4264 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
4265 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
4266 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
4267 CROSSING_JUMP_P flag is already set before setting it.
4268 * cfgrtl.c (fixup_partition_crossing): Likewise.
4269 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
4272 2017-11-22 Jakub Jelinek <jakub@redhat.com>
4274 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
4275 Handle the case where both arguments are using gen_const_vec_series.
4277 2017-11-22 David Malcolm <dmalcolm@redhat.com>
4280 * pretty-print.c (pp_format): Move quoting implementation to
4281 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
4282 to pass address of "quote" local.
4283 (pp_begin_quote): New function.
4284 (pp_end_quote): New function.
4285 * pretty-print.h (printer_fn): Convert penultimate param from bool
4287 (pp_begin_quote): New decl.
4288 (pp_end_quote): New decl.
4289 * tree-diagnostic.c (default_tree_printer): Convert penultimate
4290 param from bool to bool *.
4291 * tree-diagnostic.h (default_tree_printer): Likewise.
4293 2017-11-22 Jeff Law <law@redhat.com>
4295 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
4296 Initialize vr_values.
4297 (evrp_range_analyzer::try_find_new_range): Call methods attached to
4298 vr_values via vr_values class instance rather than delegators.
4299 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
4300 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
4301 (evrp_range_analyzer::push_value_range): Likewise.
4302 (evrp_range_analyzer::pop_value_range): Likewise.
4303 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
4304 most delegators. Those remaining are exposed as public interfaces.
4305 Make vr_values a pointer and private.
4306 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
4308 (evrp_range_analyzer::get_vr_value): New method.
4309 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
4310 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
4311 (class evrp_dom_walker): Attach evrp_folder class, initialize
4312 it in the ctor. Remove temporary delegators.
4313 (evrp_dom_walker::before_dom_children): Call methods in attached
4314 evrp_range_analyzer class via class instance pointer. Use
4315 free value_range_constant_singleton to remove need for
4316 op_with_constant_singleton_value delegator method. Do not
4317 create a vrp_prop class instance for every call! Narrow
4318 scope of a couple variables.
4319 (evrp_dom_walker::cleanup): Call methods in attached
4320 evrp_range_analyzer class via class instance pointer.
4321 * vr-values.h (class vr_values): Privatize many methods and
4324 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
4326 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
4327 * cfgexpand.c (expand_debug_expr): Remove handling for them.
4328 * expr.c (expand_expr_real_2): Likewise.
4329 * fold-const.c (const_unop): Likewise.
4330 * optabs-tree.c (optab_for_tree_code): Likewise.
4331 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
4332 * tree-inline.c (estimate_operator_cost): Likewise.
4333 * tree-pretty-print.c (dump_generic_node): Likewise.
4334 (op_code_prio): Likewise.
4335 (op_symbol_code): Likewise.
4336 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
4337 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
4338 * internal-fn.c (direct_internal_fn_optab): New function.
4339 (direct_internal_fn_array, direct_internal_fn_supported_p
4340 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
4341 * fold-const-call.c (fold_const_reduction): New function.
4342 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
4344 * tree-vect-loop.c: Include internal-fn.h.
4345 (reduction_code_for_scalar_code): Rename to...
4346 (reduction_fn_for_scalar_code): ...this and return an internal
4348 (vect_model_reduction_cost): Take an internal_fn rather than
4350 (vect_create_epilog_for_reduction): Likewise. Build calls rather
4352 (vectorizable_reduction): Use internal functions rather than tree
4353 codes for the reduction operation. Update calls to the functions
4355 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
4356 Use calls to internal functions rather than REDUC tree codes.
4357 * config/aarch64/aarch64-simd.md: Update comment accordingly.
4359 2017-11-22 Olivier Hainque <hainque@adacore.com>
4361 * config/vxworks.c (vxworks_override_options): Pick default
4362 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
4363 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
4364 DWARF_GNAT_ENCODINGS_DEFAULT.
4365 * config/vxworksae.h: Likewise.
4367 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
4369 PR tree-optimization/83104
4370 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
4373 2017-11-22 Marek Polacek <polacek@redhat.com>
4374 H.J. Lu <hongjiu.lu@intel.com>
4375 Jason Merrill <jason@redhat.com>
4380 * calls.c (initialize_argument_information): Call
4381 warn_parameter_passing_abi target hook.
4382 (store_one_arg): Use 0 for empty record size. Don't push 0 size
4383 argument onto stack.
4384 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
4385 * common.opt: Update -fabi-version description.
4386 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
4387 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
4389 (ix86_is_empty_record): New function.
4390 (ix86_warn_parameter_passing_abi): New function.
4391 (TARGET_EMPTY_RECORD_P): Redefine.
4392 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
4393 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
4394 * doc/tm.texi: Regenerated.
4395 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
4396 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
4397 * dwarf2out.c (get_ultimate_context): Move to tree.c.
4398 * explow.c (hard_function_value): Call arg_int_size_in_bytes
4399 instead of int_size_in_bytes.
4400 * expr.c (copy_blkmode_to_reg): Likewise.
4401 * function.c (aggregate_value_p): Return 0 for empty types.
4402 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
4404 (locate_and_pad_parm): Call arg size_in_bytes instead
4406 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
4407 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
4408 * target.def (empty_record_p, warn_parameter_passing_abi): New target
4410 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
4411 (std_gimplify_va_arg_expr): Skip empty records. Call
4412 arg_size_in_bytes instead size_in_bytes.
4413 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
4414 * tree-core.h (tree_type_common): Add empty_flag.
4415 (tree_decl_common): Update comments.
4416 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
4418 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
4419 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
4421 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
4422 * tree.c (default_is_empty_type): New function.
4423 (default_is_empty_record): New function.
4424 (arg_int_size_in_bytes): New function.
4425 (arg_size_in_bytes): New function.
4426 (get_ultimate_context): New function.
4427 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
4428 TRANSLATION_UNIT_WARN_EMPTY_P.
4429 (default_is_empty_record, arg_int_size_in_bytes,
4430 arg_size_in_bytes, get_ultimate_context): Declare.
4432 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
4434 * config/arm/arm.c (cmse_clear_registers): New function.
4435 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
4436 code by call to cmse_clear_registers.
4437 (cmse_nonsecure_entry_clear_before_return): Likewise.
4439 2017-11-22 Tamar Christina <tamar.christina@arm.com>
4441 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
4442 (vdot_s32, vdotq_s32): New.
4443 (vdot_lane_u32, vdotq_lane_u32): New.
4444 (vdot_lane_s32, vdotq_lane_s32): New.
4447 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
4450 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
4451 for unsigned values with fewer HWIs than the precision.
4452 (test_overflow): New function.
4453 (wide_int_cc_tests): Call it.
4455 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
4457 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
4458 has the same mode class as Pmode.
4460 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
4462 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
4463 padding_bits_to_clear_ptr.
4464 (cmse_nonsecure_entry_clear_before_return): Likewise.
4466 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
4468 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
4469 auto_sbitap instead of integer bitfield to control register needing
4472 2017-11-22 Jakub Jelinek <jakub@redhat.com>
4474 PR tree-optimization/83044
4475 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
4476 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
4477 that need the upper bound. Subtract offset from
4478 get_addr_base_and_unit_offset only if positive and subtract it
4479 before division by eltsize rather than after it.
4482 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
4483 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
4484 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
4488 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
4489 expand_binop with *mul_widen_optab, make sure at least one of the
4490 operands doesn't have VOIDmode.
4493 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
4495 PR rtl-optimization/82044
4496 PR tree-optimization/82042
4497 * dse.c (record_store): Check for overflow.
4498 (check_mem_read_rtx): Properly check for overflow if width == -1, call
4499 add_wild_read instead of clear_rhs_from_active_local_stores on
4500 overflow and log it into dump_file.
4502 2017-11-22 Richard Biener <rguenther@suse.de>
4504 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
4505 fake edges to exit when looking for a place to insert.
4506 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
4508 (insert_into_preds_of_block): Commit edge insertion immediately,
4509 assert that doesn't require new BBs.
4510 (fini_pre): Release expressions.
4511 (pass_pre::execute): Shuffle things around a bit, if the fn
4512 is too large do not compute AVAIL either as this is really the
4515 2017-11-22 Richard Biener <rguenther@suse.de>
4517 PR tree-optimization/83089
4518 * tree-if-conv.c (pass_if_conversion::execute): If anything
4519 changed reset SCEV and free the number of iteration estimates.
4521 2017-11-21 Martin Sebor <msebor@redhat.com>
4523 PR tree-optimization/82945
4524 * calls.h (warn_nonstring_bound): Remove unused function.
4526 2017-11-21 Martin Sebor <msebor@redhat.com>
4528 PR tree-optimization/82945
4529 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
4530 * calls.h (maybe_warn_nonstring_arg): Declare new function.
4531 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
4533 (initialize_argument_information): Call maybe_warn_nonstring_arg.
4534 * calls.h (get_attr_nonstring_decl): Declare new function.
4535 * doc/extend.texi (attribute nonstring): Update.
4536 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
4537 get_attr_nonstring_decl and handle it.
4538 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
4539 detection of nul-termination.
4540 (strlen_to_stridx): Change to a pointer.
4541 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
4542 (pass_strlen::execute): Same.
4544 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
4546 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
4547 for the new option -mprefer-vector-width=[none|128|256|512].
4548 * config/i386/i386.c (ix86_target_string): remove old style options
4549 -mprefer-avx256 and make -mprefer-avx128 as alias.
4550 (ix86_option_override_internal): Apply defaults for the
4551 -mprefer-vector-width=[128|256] option.
4552 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
4553 Implement macros to work with -mprefer-vector-width=.
4554 * config/i386/i386.opt: Implemented option
4555 -mprefer-vector-width=[none|128|256|512].
4556 * doc/invoke.texi: Documentation for
4557 -mprefer-vector-width=[none|128|256|512].
4559 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
4561 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
4563 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
4565 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
4567 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
4569 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
4571 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
4572 * expr.c (expand_expr_real_2): Likewise.
4573 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
4574 fold_binary_loc): Likewise.
4575 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
4576 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
4577 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
4578 MINUS_EXPR transformations.
4579 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
4580 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
4581 * tree-inline.c (estimate_operator_cost): Likewise.
4582 * tree-pretty-print.c (dump_generic_node, op_code_prio,
4583 op_symbol_code): Likewise.
4584 * tree-vect-stmts.c (vectorizable_operation): Likewise.
4585 * vr-values.c (extract_range_from_binary_expr): Likewise.
4586 * varasm.c (initializer_constant_valid_p_1): Likewise.
4587 * tree.def: New tree code POINTER_DIFF_EXPR.
4589 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
4591 * config/i386/i386.md (*bswap<mode>2_movbe): Add
4592 integer suffix to movbe mnemonic.
4593 (*bswaphi2_movbe): Ditto.
4594 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
4596 2017-11-21 Jakub Jelinek <jakub@redhat.com>
4599 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
4600 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
4601 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
4603 2017-11-21 Martin Liska <mliska@suse.cz>
4605 * tree-inline.c (expand_call_inline): Remove not needed
4608 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
4609 Jakub Jelinek <jakub@redhat.com>
4612 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
4613 Remove static keyword from f variable.
4615 2017-11-21 Jakub Jelinek <jakub@redhat.com>
4617 PR tree-optimization/83086
4618 * gimple-ssa-store-merging.c
4619 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
4620 rather than n.base_addr.
4622 2017-11-21 Martin Liska <mliska@suse.cz>
4624 PR rtl-optimization/82044
4625 PR tree-optimization/82042
4626 * dse.c (check_mem_read_rtx): Check for overflow.
4628 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
4630 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
4633 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
4635 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
4636 and its corresponding call arg location note.
4638 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
4639 Andrew Burgess <andrew.burgess@embecosm.com>
4641 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
4643 (arc_return_slot_offset): Likewise.
4644 (arc_eh_return_address_location): New declaration.
4645 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
4646 (MUST_SAVE_REGISTER): Add exception handler case.
4647 (MUST_SAVE_RETURN_ADDR): Likewise.
4648 (arc_frame_pointer_required): Likewise.
4649 (arc_frame_pointer_needed): New function.
4650 (arc_compute_frame_size): Changed.
4651 (arc_expand_prologue): Likewise.
4652 (arc_expand_epilogue): Likewise.
4653 (arc_initial_elimination_offset): Likewise.
4654 (arc_return_slot_offset): Delete.
4655 (arc_eh_return_address_location): New function.
4656 (arc_builtin_setjmp_frame_value): Likewise.
4657 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
4658 (EH_RETURN_STACKADJ_RTX): Define.
4659 (EH_RETURN_HANDLER_RTX): Likewise.
4660 * config/arc/arc.md (eh_return): Delete.
4662 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
4664 * print-tree.h (debug_vec_tree): Remove prototype.
4665 * gdbinit.in (pvt): Remove macro.
4667 2017-11-21 Jakub Jelinek <jakub@redhat.com>
4669 PR tree-optimization/83047
4670 * gimple-ssa-store-merging.c
4671 (imm_store_chain_info::output_merged_store): If the loads with the
4672 same vuse are in different basic blocks, for load_gsi pick a load
4673 location that is dominated by the other loads.
4676 * config/i386/i386.c (ix86_memmodel_check): Start
4677 -Winvalid-memory-model diagnostics with lowercase letter.
4680 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
4681 set in_cold_section_p to the partition containing loc_list->first.
4682 When seeing loc_list->last_before_switch node, update secname and
4683 perform range_across_switch second partition handling only after that.
4686 * run-rtl-passes.c: Include debug.h.
4687 (run_rtl_passes): Call debug_hooks->assembly_start.
4688 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
4692 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
4693 OPTAB_DIRECT in calls to expand_simple_binop.
4695 2017-11-20 David Malcolm <dmalcolm@redhat.com>
4698 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
4700 2017-11-20 David Malcolm <dmalcolm@redhat.com>
4703 * spellcheck.h (best_match::blithely_get_best_candidate): New
4706 2017-11-20 Jakub Jelinek <jakub@redhat.com>
4708 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
4709 with lower case letter.
4711 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
4713 * config/i386/i386.md (bswaphi2): New expander.
4714 (*bswaphi2_movbe): New insn pattern.
4715 (bswaphi -> rorhi pepehole2): New peephole pattern.
4717 2017-11-20 Jeff Law <law@redhat.com>
4719 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
4720 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
4721 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
4722 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
4723 gimple-ssa-evrp-analyze.h.
4725 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
4726 set BB_VISITED here.
4727 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
4729 2017-11-20 Bin Cheng <bin.cheng@arm.com>
4731 * tree-predcom.c: Add general comment on Store-Store chains.
4732 (split_data_refs_to_components): Postpone clearing eliminate_store_p
4734 (get_chain_last_ref_at): Rename into...
4735 (get_chain_last_write_at): ...this.
4736 (get_chain_last_write_before_load): New function.
4737 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
4738 CT_STORE_STORE when write reference is added.
4739 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
4740 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
4741 (initialize_root_vars_store_elim_1): Ditto.
4742 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
4743 definition is created.
4744 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
4745 chain by replacing it with dominant stored value.
4747 2017-11-20 Bin Cheng <bin.cheng@arm.com>
4749 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
4751 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
4753 * vr-values.c (extract_range_from_binary_expr): Use a full range
4756 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
4758 * config/arm/arm.md (R4_REGNUM): Define constant.
4759 (nonsecure_call_internal): Remove r4 clobber.
4760 (nonsecure_call_value_internal): Likewise.
4761 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
4762 clobber and resequence match_operands.
4763 (nonsecure_call_value_reg_thumb1_v5): Likewise.
4764 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
4765 (nonsecure_call_value_reg_thumb2): Likewise.
4767 2017-11-20 Jakub Jelinek <jakub@redhat.com>
4769 PR tree-optimization/78821
4770 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
4771 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
4772 to the address of the base rather than the base itself.
4773 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
4774 (find_bswap_or_nop_finalize): New function.
4775 (find_bswap_or_nop): Use it.
4776 (bswap_replace): Return a tree rather than bool, change first
4777 argument from gimple * to gimple_stmt_iterator, allow inserting
4778 into an empty sequence, allow ins_stmt to be NULL - then emit
4779 all stmts into gsi. Fix up MEM_REF address gimplification.
4780 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
4781 (struct store_immediate_info): Add N and INS_STMT non-static
4783 (store_immediate_info::store_immediate_info): Initialize them
4784 from newly added ctor args.
4785 (merged_store_group::apply_stores): Formatting fixes. Sort by
4787 (stmts_may_clobber_ref_p): For stores call also
4788 refs_anti_dependent_p.
4789 (gather_bswap_load_refs): New function.
4790 (imm_store_chain_info::try_coalesce_bswap): New method.
4791 (imm_store_chain_info::coalesce_immediate_stores): Use it.
4792 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
4793 (imm_store_chain_info::output_merged_store): Fail if number of
4794 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
4795 and NOP_EXPR rhs_code.
4796 (pass_store_merging::process_store): Compute n and ins_stmt, if
4797 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
4798 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
4800 (pass_store_merging::execute): Calculate dominators.
4802 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
4803 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
4804 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
4805 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
4806 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
4807 class pass_optimize_bswap, bswap_replace,
4808 pass_optimize_bswap::execute): Moved to ...
4809 * gimple-ssa-store-merging.c: ... this file.
4810 Include optabs-tree.h.
4811 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
4812 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
4813 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
4814 anonymous namespace, remove static keywords.
4815 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
4816 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
4818 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
4821 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
4823 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
4825 * vec.h (debug_helper): New function.
4826 (DEFINE_DEBUG_VEC): New macro.
4827 * hash-set.h (debug_helper): New function.
4828 (DEFINE_DEBUG_HASH_SET): New macro.
4829 * cfg.c (debug_slim (edge)): New function.
4830 Call DEFINE_DEBUG_VEC for edges.
4831 Call DEFINE_DEBUG_HASH_SET for edges.
4832 * cfghooks.c (debug_slim (basic_block)): New function.
4833 Call DEFINE_DEBUG_VEC for basic blocks.
4834 Call DEFINE_DEBUG_HASH_SET for basic blocks.
4835 * print-tree.c (debug_slim): New function to handle trees.
4836 Call DEFINE_DEBUG_VEC for trees.
4837 Call DEFINE_DEBUG_HASH_SET for trees.
4838 (debug (vec<tree, va_gc>) &): Remove.
4839 (debug (<vec<tree, va_gc>) *): Remove.
4840 * print-rtl.c (debug_slim): New function to handle const_rtx.
4841 Call DEFINE_DEBUG_VEC for rtx_def.
4842 Call DEFINE_DEBUG_VEC for rtx_insn.
4843 Call DEFINE_DEBUG_HASH_SET for rtx_def.
4844 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
4845 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
4846 (debug (vec<rtx_insn *> *ptr): Remove.
4847 (debug_insn_vector): Remove.
4848 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
4850 2017-11-20 Tom de Vries <tom@codesourcery.com>
4852 PR rtl-optimization/82020
4853 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
4854 IF_THEN_ELSE condition.
4856 2017-11-19 Jeff Law <law@redhat.com>
4858 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
4859 of degenerates resulting from ignoring an edge.
4861 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
4864 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
4866 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
4869 * profile-count.c (profile_count::to_sreal_scale): Fix return value
4870 for uninitialied counts.
4872 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
4875 * tree-inline.c (estimate_num_insns): Set to 1 at least.
4877 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
4880 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
4882 2017-11-19 Tom de Vries <tom@codesourcery.com>
4884 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
4887 2017-11-19 Tom de Vries <tom@codesourcery.com>
4889 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
4892 2017-11-19 Tom de Vries <tom@codesourcery.com>
4894 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
4897 2017-11-19 Tom de Vries <tom@codesourcery.com>
4899 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
4901 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
4903 2017-11-19 Tom de Vries <tom@codesourcery.com>
4905 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
4907 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
4908 semicolon after MCORE_EXPORT_NAME call.
4910 2017-11-19 Tom de Vries <tom@codesourcery.com>
4913 * vmsdbgout.c (vmsdbgout_early_finish): New function.
4914 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
4916 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
4918 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
4919 (cgraph_edge::clone): Cleanup updating of profile.
4920 * ipa-cp.c (update_profiling_info): Likewise.
4921 * ipa-inline-transform.c (inline_transform): Likewise.
4922 * ipa-inline.c (inline_small_functions): Add missing space to dump.
4923 * ipa-split.c (execute_split_functions): Do not split when function
4925 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
4926 * profile-count.c (profile_count::dump): Add global0.
4927 (profile_count::to_cgraph_frequency): Do not ICE when entry is
4929 (profile_count::to_sreal_scale): Likewise.
4930 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
4931 (profile_count::combine_with_ipa_count): New function.
4932 * profile-count.h (profile_guessed_global0adjusted): New.
4933 (profile_count::adjusted_zero): New.
4934 (profile_count::global0adjusted): New.
4935 (profile_count::combine_with_ipa_count): New.
4936 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
4937 correct profile of return block of split functions.
4938 (copy_cfg_body): Remove unused profile_count.
4939 (copy_body): Likewise.
4940 (expand_call_inline): Update.
4941 (tree_function_versioning): Update.
4943 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
4945 * hash-set.h (hash_set::empty): New.
4946 * tree-ssa-threadbackward.h: Delete.
4947 * tree-ssa-threadbackward.c (class thread_jumps): New.
4948 Move max_threaded_paths into class.
4949 (fsm_find_thread_path): Remove arguments that are now in class.
4950 (profitable_jump_thread_path): Rename to...
4951 (thread_jumps::profitable_jump_thread_path): ...this.
4952 (convert_and_register_jump_thread_path): Rename to...
4953 (thread_jumps::convert_and_register_current_path): ...this.
4954 (check_subpath_and_update_thread_path): Rename to...
4955 (thread_jumps::check_subpath_and_update_thread_path): ...this.
4956 (register_jump_thread_path_if_profitable): Rename to...
4957 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
4958 (handle_phi): Rename to...
4959 (thread_jumps::handle_phi): ...this.
4960 (handle_assignment): Rename to...
4961 (thread_jumps::handle_assignment): ...this.
4962 (fsm_find_control_statement_thread_paths): Rename to...
4963 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
4964 (find_jump_threads_backwards): Rename to...
4965 (thread_jumps::find_jump_threads_backwards): ...this.
4966 Initialize path local data.
4967 (pass_thread_jumps::execute): Call find_jump_threads_backwards
4968 from within thread_jumps class.
4969 (pass_early_thread_jumps::execute): Same.
4971 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
4973 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
4975 2017-11-17 Steve Ellcey <sellcey@cavium.com>
4977 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
4978 to canonical location.
4980 2017-11-17 Steve Ellcey <sellcey@cavium.com>
4983 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
4985 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
4987 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
4989 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
4990 rcrt1.o%s/grcrt1.o%s for -static-pie.
4992 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
4994 * i386.c (ix86_multiplication_cost, ix86_division_cost,
4995 ix86_shift_rotate_cost): Break out from ...
4996 (ix86_rtx_costs): ... here.
4997 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
5000 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5002 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
5003 when entry block was promoted unlikely.
5004 (estimate_bb_frequencies): Increase frequency scale.
5005 * profile-count.h (profile_count): Export precision info.
5007 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5009 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
5010 disturbin profile of entry block.
5012 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5014 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
5017 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5019 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
5020 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
5022 2017-11-17 Jeff Law <law@redhat.com>
5024 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
5025 from evrp_dom_walker class. Various methods moved into new class.
5026 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
5027 (evrp_range_analyzer::enter): New method.
5028 (evrp_range_analyzer::leave): New method.
5029 (evrp_dom_walker): Remove delegators no longer needed by this class.
5030 Replace vr_values data member with evrp_range_analyzer
5032 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
5033 method extracted from evrp_dom_walker::before_dom_children.
5034 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
5035 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
5037 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
5038 Add private copy constructor and move assignment operators.
5039 Privatize methods and class data where trivially possible.
5040 (evrp_dom_walker::cleanup): New function, extracted from
5041 execute_early_vrp. Simplify access to class data.
5043 * vr-values.h (get_output_for_vrp): Prototype.
5044 * vr-values.c (get_output_for_vrp): New function extracted from
5045 vrp_visit_assignment_or_call and extract_range_from_stmt.
5046 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
5048 2017-11-17 Luis Machado <luis.machado@linaro.org>
5050 * config/aarch64/aarch64.c
5051 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
5052 (qdf24xx_tunings) <autoprefetcher_model>: Set to
5053 tune_params::AUTOPREFETCHER_WEAK.
5055 2017-11-17 Tamar Christina <tamar.christina@arm.com>
5058 * config/arm/arm.c (arm_valid_target_attribute_rec):
5059 Parse "arch=" and "+<ext>".
5060 (arm_valid_target_attribute_tree): Re-init global options.
5061 (arm_option_override): Make non-static.
5062 (arm_options_perform_arch_sanity_checks): Make errors fatal.
5063 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
5064 (__ARM_FEATURE_CRC32): Support undef.
5065 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
5066 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
5068 2017-11-17 David Malcolm <dmalcolm@redhat.com>
5070 * gdbinit.in (break-on-diagnostic): New command.
5072 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5074 * config/i386/i386.c (ix86_expand_epilogue): Change simple
5075 return to indirect jump for EH return if control-flow
5076 protection is enabled. Change explicit 'false' argument in
5077 pro_epilogue_adjust_stack with a value of
5079 * config/i386/i386.md (simple_return_indirect_internal):
5080 Remove SImode restriction to support 64-bit.
5082 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
5084 * combine.c (added_notes_insn): New.
5085 (try_combine): Handle added_notes_insn like added_links_insn.
5086 Rewrite return value code.
5087 (distribute_notes): Set added_notes_insn to the earliest insn we added
5090 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
5092 PR rtl-optimization/82621
5093 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
5094 dest of one of those SETs is unused.
5096 2017-11-17 Richard Biener <rguenther@suse.de>
5099 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
5100 * tree-pretty-print.c (dump_generic_node): Handle
5101 annot_expr_parallel_kind.
5102 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
5103 * gimplify.c (gimple_boolify): Likewise.
5105 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5107 * config.gcc (extra_headers): Add cet.h for x86 targets.
5108 * config/i386/cet.h: New file.
5109 * doc/install.texi: Add --enable-cet/--disable-cet.
5111 2017-11-17 Richard Biener <rguenther@suse.de>
5113 PR tree-optimization/83017
5114 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
5115 (gen_parallel_loop): Properly count iterations.
5116 (parallelize_loops): Handle loop->can_be_parallel independent
5117 of flag_loop_parallelize_all. Make static profitability test match
5119 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
5120 * invoke.texi (parloops-min-per-thread): Document.
5122 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
5124 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
5125 upstreaming review comments.
5127 2017-11-17 Tamar Christina <tamar.christina@arm.com>
5129 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
5130 with fast unaligned access.
5131 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
5133 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
5135 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
5136 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
5138 2017-11-17 Richard Biener <rguenther@suse.de>
5140 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
5141 folding of references.
5143 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
5146 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
5147 of replacing call to strncmp with corresponding call to strcmp when
5150 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
5152 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
5153 option prefer-avx256 for skylake-avx512 configuration.
5154 * config/i386/i386.c (ix86_option_override_internal): Ditto.
5155 (get_builtin_code_for_version): Ditto.
5157 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
5158 Monk Chiang <sh.chiang04@gmail.com>
5160 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
5161 (FIXED_REGISTERS): Reserve more register numbers.
5162 (CALL_USED_REGISTERS): Likewise.
5163 (REG_ALLOC_ORDER): Likewise.
5164 (REG_CLASS_CONTENTS): Likewise.
5165 (REGISTER_NAMES): Likewise.
5167 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
5168 Kito Cheng <kito.cheng@gmail.com>
5170 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
5172 * config/nds32/iterators.md: Add vector mode iterators and attributes.
5174 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
5176 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
5177 parameter list for vec_splats.
5179 2017-11-16 Joseph Myers <joseph@codesourcery.com>
5181 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
5183 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
5185 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
5188 * doc/install.texi: Document incompatibility of Perl >=5.6.26
5189 with the required version of automake 1.11.6.
5191 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
5193 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
5195 (DU_C2_3_power9): Correct reservation combinations.
5196 (FP_DIV_power9, VEC_DIV_power9): New.
5197 (power9-alu): Split out rotate/shift...
5198 (power9-rot): ...to here, correct dispatch resource.
5199 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
5201 (power9-fp): Correct latency.
5202 (power9-sdiv): Add div/sqrt resource.
5203 (power9-ddiv): Correct latency, add div/sqrt resource.
5204 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
5205 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
5207 (power9-qpdiv, power9-qpmul): Adjust resource usage.
5209 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
5211 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
5212 switch statement mapping KF built-ins to TF built-ins if we don't
5213 have the proper ISA 3.0 assembler support.
5215 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5217 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
5218 (lower_emutls_function_body): Do not compute it.
5220 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5222 * ipa-split.c (split_bb_info): Turn time to sreal.
5223 (split_point): Likewise.
5224 (dump_split_point): Likewise.
5225 (fine_split_points): Likewise.
5226 (execute_split_functions): Only zero split_bbs; turn time to sreals.
5228 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5230 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
5233 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5235 * predict.c (combine_predictions_for_bb): Preserve zero predicted
5237 (expensive_function_p): Remove useless assert.
5238 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
5240 2017-11-16 Martin Sebor <msebor@redhat.com>
5242 PR tree-optimization/82588
5243 PR tree-optimization/82583
5244 * tree-vrp.c (check_array_ref): Handle flexible array members,
5245 string literals, and inner indices.
5246 (search_for_addr_array): Add detail to diagnostics.
5248 2017-11-16 Nathan Sidwell <nathan@acm.org>
5252 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
5253 (SET_DECL_ASSEMBLER_NAME): Forward to
5254 overwrite_decl_assembler_name.
5255 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
5256 (overwrite_decl_assembler_name): Declare.
5257 * tree.c (overwrite_decl_assembler_name): New.
5258 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
5259 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
5260 (LANG_HOOKS_INITIALIZER): Add it.
5261 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
5262 * langhooks.c (lhd_set_decl_assembler_name): Use
5263 SET_DECL_ASSEMBLER_NAME.
5264 (lhd_overwrite_decl_assembler_name): Default implementation.
5266 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
5267 Jackson Woodruff <jackson.woodruff@arm.com>
5269 PR tree-optimization/71026
5270 * match.pd: Canonicalize constant multiplies in division.
5272 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
5274 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
5275 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
5276 Remove OPT_fomit_frame_pointer entry.
5277 * common/config/arc/arc-common.c: Likewise.
5278 * common/config/arm/arm-common.c: Likewise.
5279 * common/config/avr/avr-common.c: Likewise.
5280 * common/config/c6x/c6x-common.c: Likewise.
5281 * common/config/cr16/cr16-common.c: Likewise.
5282 * common/config/cris/cris-common.c: Likewise.
5283 * common/config/epiphany/epiphany-common.c: Likewise.
5284 * common/config/fr30/fr30-common.c: Likewise.
5285 * common/config/frv/frv-common.c: Likewise.
5286 * common/config/ia64/ia64-common.c: Likewise.
5287 * common/config/iq2000/iq2000-common.c: Likewise.
5288 * common/config/lm32/lm32-common.c: Likewise.
5289 * common/config/m32r/m32r-common.c: Likewise.
5290 * common/config/mcore/mcore-common.c: Likewise.
5291 * common/config/microblaze/microblaze-common.c: Likewise.
5292 * common/config/mips/mips-common.c: Likewise.
5293 * common/config/mmix/mmix-common.c: Likewise.
5294 * common/config/mn10300/mn10300-common.c: Likewise.
5295 * common/config/nios2/nios2-common.c: Likewise.
5296 * common/config/pa/pa-common.c: Likewise.
5297 * common/config/pdp11/pdp11-common.c: Likewise.
5298 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
5299 * common/config/riscv/riscv-common.c: Likewise.
5300 * common/config/rs6000/rs6000-common.c: Likewise.
5301 * common/config/rx/rx-common.c: Likewise.
5302 * common/config/s390/s390-common.c: Likewise.
5303 * common/config/sh/sh-common.c: Likewise.
5304 * common/config/sparc/sparc-common.c: Likewise.
5305 * common/config/tilegx/tilegx-common.c: Likewise.
5306 * common/config/tilepro/tilepro-common.c: Likewise.
5307 * common/config/v850/v850-common.c: Likewise.
5308 * common/config/visium/visium-common.c: Likewise.
5309 * common/config/xstormy16/xstormy16-common.c: Likewise.
5310 * common/config/xtensa/xtensa-common.c: Likewise.
5311 * invoke.texi (-fomit-frame-pointer): Update documentation.
5313 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5315 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
5317 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5319 * cfg.c (scale_bbs_frequencies_int,
5320 cale_bbs_frequencies_gcov_type): Remove.
5321 * cfg.h (scale_bbs_frequencies_int,
5322 cale_bbs_frequencies_gcov_type): Remove.
5324 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5326 * tree-ssa-loop-manip.c
5327 (scale_dominated_blocks_in_loop): Update to profile counts.
5328 (tree_transform_and_unroll_loop): Likewise.
5330 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5332 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
5333 scale_bbs_frequencies_int.
5335 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5337 * final.c (compute_alignments): Use counts rather than frequencies.
5339 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5341 * cfgloopanal.c: Include sreal.h
5342 (average_num_loop_insns): Use counts and sreal for accounting.
5344 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5346 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
5349 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
5350 Kito Cheng <kito.cheng@gmail.com>
5352 * config/nds32/constraints.md: Provide more constraints.
5353 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
5354 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
5355 support constraints usage.
5357 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
5359 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
5361 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
5362 Kito Cheng <kito.cheng@gmail.com>
5364 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
5365 * config/nds32/nds32.opt: Refine the layout.
5366 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
5367 TARGET_EXT_STRING): Support new options.
5368 * config/nds32/nds32.h: Likewise.
5369 * config/nds32/nds32.md: Likewise.
5370 * config/nds32/nds32-predicates.c: Likewise.
5371 * config/nds32/constraints.md: Likewise.
5372 * common/config/nds32/nds32-common.c: Likewise.
5374 2017-11-16 Julia Koval <julia.koval@intel.com>
5377 * config/i386/gfniintrin.h: Add sse check.
5378 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
5380 2017-11-16 Julia Koval <julia.koval@intel.com>
5382 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
5383 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
5384 (ix86_handle_option): Handle -mavx512vbmi2.
5385 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
5386 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
5387 * config/i386/i386-c.c (__AVX512VBMI2__): New.
5388 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
5389 (ix86_valid_target_attribute_inner_p): Ditto.
5390 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
5391 * config/i386/i386.opt (mavx512vbmi2): New option.
5392 * doc/invoke.texi: Add new option.
5394 2017-11-16 Julia Koval <julia.koval@intel.com>
5396 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
5397 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
5398 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
5399 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
5400 _mm512_gf2p8mul_epi8): New intrinsics.
5401 * config/i386/i386-builtin-types.def
5402 (V64QI_FTYPE_V64QI_V64QI): New type.
5403 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
5404 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
5405 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
5406 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
5407 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
5408 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
5410 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
5412 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
5413 explicitly as a stream of bytes.
5415 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5417 * config/rs6000/altivec.h (vec_xst_be): New #define.
5418 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
5419 and externalize from *altivec_vperm_<mode>_internal.
5420 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
5422 (XL_BE_V8HI): Likewise.
5423 (XL_BE_V4SI): Likewise.
5424 (XL_BE_V4SI): Likewise.
5425 (XL_BE_V2DI): Likewise.
5426 (XL_BE_V4SF): Likewise.
5427 (XL_BE_V2DF): Likewise.
5428 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
5429 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
5430 all array entries with these keys: VSX_BUILTIN_VEC_XL,
5431 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
5432 VSX_BUILTIN_VEC_XST_BE.
5433 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
5434 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
5436 (altivec_init_builtins): Replace conditional calls to def_builtin
5437 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
5438 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
5439 with unconditional calls. Remove calls to def_builtin for
5440 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
5441 __builtin_vec_xst_be.
5442 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
5443 to define_expand, and add alternate RTL generation for P8.
5444 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
5445 vsx_ld_elemrev_v8hi.
5446 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
5447 add alternate RTL generation for P8.
5448 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
5449 vsx_ld_elemrev_v16qi.
5450 (vsx_st_elemrev_v8hi): Convert define_insn
5451 to define_expand, and add alternate RTL generation for P8.
5452 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
5453 vsx_st_elemrev_v8hi.
5454 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
5455 add alternate RTL generation for P8.
5456 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
5457 vsx_st_elemrev_v16qi.
5459 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
5462 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
5463 TARGET_AVX512ER check.
5464 (ix86_option_override_internal): Set MASK_VZEROUPPER if
5465 neither -mzeroupper nor -mno-zeroupper is used and
5466 TARGET_EMIT_VZEROUPPER is set.
5467 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
5468 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
5470 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
5472 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
5473 folding of vector compares.
5474 (fold_build_vec_cmp): New helper function.
5475 (fold_compare_helper): New helper function.
5476 (builtin_function_type): Add compare builtins to the list of functions
5477 having unsigned arguments. Cosmetic updates to comment indentation.
5478 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
5479 the not+eq combination.
5481 2017-11-15 Bin Cheng <bin.cheng@arm.com>
5483 PR tree-optimization/82726
5484 PR tree-optimization/70754
5485 * tree-predcom.c (order_drefs_by_pos): New function.
5486 (combine_chains): Move code setting has_max_use_after to...
5487 (try_combine_chains): ...here. New parameter. Sort combined chains
5488 according to position information.
5489 (tree_predictive_commoning_loop): Update call to above function.
5490 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
5492 2017-11-15 Bin Cheng <bin.cheng@arm.com>
5494 PR tree-optimization/82726
5496 2017-01-23 Bin Cheng <bin.cheng@arm.com>
5498 PR tree-optimization/70754
5499 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
5500 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
5501 combined stmt before it if not NULL.
5502 (combine_chains): Process refs reversely and compute dominance point
5506 2017-02-23 Bin Cheng <bin.cheng@arm.com>
5508 PR tree-optimization/79663
5509 * tree-predcom.c (combine_chains): Process refs in reverse order
5510 only for ZERO length chains, and add explaining comment.
5512 2017-11-15 Tamar Christina <tamar.christina@arm.com>
5514 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
5515 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
5516 * doc/invoke.texi (ARM Options): Add armv8.3-a.
5518 2017-11-15 Tamar Christina <tamar.christina@arm.com>
5520 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
5522 2017-11-15 Martin Liska <mliska@suse.cz>
5524 * tree-cfg.c (pass_warn_function_return::execute):
5525 Compare warn_return_type for greater than zero.
5527 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
5531 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
5532 to return true on Xeon and not on Xeon Phi.
5533 (ix86_check_avx256_register): Changed to ...
5534 (ix86_check_avx_upper_register): ... this. Add extra check for
5535 VALID_AVX512F_REG_OR_XI_MODE.
5536 (ix86_avx_u128_mode_needed): Changed
5537 ix86_check_avx256_register to ix86_check_avx_upper_register.
5538 (ix86_check_avx256_stores): Changed to ...
5539 (ix86_check_avx_upper_stores): ... this. Changed
5540 ix86_check_avx256_register to ix86_check_avx_upper_register.
5541 (ix86_avx_u128_mode_after): Changed
5542 avx_reg256_found to avx_upper_reg_found. Changed
5543 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
5544 (ix86_avx_u128_mode_entry): Changed
5545 ix86_check_avx256_register to ix86_check_avx_upper_register.
5546 (ix86_avx_u128_mode_exit): Ditto.
5547 * config/i386/i386.h: (host_detect_local_cpu): New define.
5549 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
5551 * config/arm/xgene1.md (xgene1): Split into automatons
5552 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
5553 (xgene1_f_load): Adjust reservations and/or types.
5554 (xgene1_f_store): Likewise.
5555 (xgene1_load_pair): Likewise.
5556 (xgene1_store_pair): Likewise.
5557 (xgene1_fp_load1): Likewise.
5558 (xgene1_load1): Likewise.
5559 (xgene1_store1): Likewise.
5560 (xgene1_move): Likewise.
5561 (xgene1_alu): Likewise.
5562 (xgene1_simd): Likewise.
5563 (xgene1_bfm): Likewise.
5564 (xgene1_neon_load1): Likewise.
5565 (xgene1_neon_store1): Likewise.
5566 (xgene1_neon_logic): Likewise.
5567 (xgene1_neon_st1): Likewise.
5568 (xgene1_neon_ld1r): Likewise.
5569 (xgene1_alu_cond): Added.
5570 (xgene1_shift_reg): Likwise.
5571 (xgene1_bfx): Likewise.
5572 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
5574 2017-11-15 Jakub Jelinek <jakub@redhat.com>
5577 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
5579 (can_widen_mult_without_libcall): New function.
5580 (expand_mul_overflow): If only checking unsigned mul overflow,
5581 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
5582 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
5583 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
5584 (expand_DIVMOD): Formatting fix.
5585 * expmed.h (expand_mult): Add NO_LIBCALL argument.
5586 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
5587 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
5589 PR tree-optimization/82977
5590 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
5591 constructed temporary to strlen_to_stridx.put.
5593 2017-11-15 Martin Liska <mliska@suse.cz>
5595 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
5596 * configure: Regenerate.
5598 2017-11-15 Martin Liska <mliska@suse.cz>
5601 * config/sh/sh-mem.cc: Use proper probability for
5604 2017-11-14 Jeff Law <law@redhat.com>
5606 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
5607 the red zone for stack_clash_protection_final_dynamic_probe targets
5608 when the total dynamic stack size is zero bytes.
5610 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
5611 blocks is post order.
5613 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
5615 * dumpfile.h (TDF_COMPARE_DEBUG): New.
5616 * final.c (rest_of_clean_state): Set it for the
5617 -fcompare-debug dump.
5618 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
5619 class when TDF_COMPARE_DEBUG is set.
5621 * dwarf2out.c (gen_producer_string): Discard
5624 2017-11-15 Joseph Myers <joseph@codesourcery.com>
5627 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
5628 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
5629 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
5630 (__TGMATH_REAL_2_3): Remove macros.
5631 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
5632 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
5634 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
5636 (remquo): Define using __TGMATH_REAL_3.
5638 2017-11-14 Jeff Law <law@redhat.com>
5640 * vr-values.c: New file with contents extracted from tree-vrp.c.
5641 * Makefile.in (OBJS): Add vr-values.o
5642 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
5643 (set_value_range, set_and_canonicalize_value_range): Likewise.
5644 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
5645 (value_range_constant_singleton, symbolic_range_p): Likewise.
5646 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
5647 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
5648 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
5649 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
5650 (find_case_label_range, find_case_label_index): Likewise.
5651 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
5652 (range_int_cst_singleton_p, value_inside_range): Likewise.
5653 (get_single_symbol): Likewise.
5654 (switch_update): Move structure definition here.
5655 (to_remove_edges, to_update_switch_stmts): Provide externs.
5656 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
5657 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
5658 (vrp_val_is_min, set_value_range): Likewise.
5659 (set_and_canonicalize_value_range, copy_value_range): Likewise.
5660 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
5661 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
5662 (range_is_nonnull, range_int_cst_p): Likewwise.
5663 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
5664 (get_single_symbol, operand_less_p): Likewise
5665 (compare_values_warnv, compare_values): Likewise.
5666 (value_inside_range, value_range_constant_singleton): Likewise.
5667 (zero_nonzero_bitgs_from_vr): Likewise.
5668 (extract_range_from_binary_expr_1): Likewise.
5669 (overflow_comparison_p): Likewise.
5670 (to_remove_edges, to_update_switch_stmts): Likewise.
5671 (find_case_label-index, find_case_label_range): Likewise.
5672 (switch_update, set_value_range_to_nonnegative): Remove.
5673 (set_value_range_to_truthvalue): Likewise.
5674 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
5675 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
5676 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
5677 (find_case_label_ranges, test_for_singularity): Likewise.
5678 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
5679 (x_vr_values): Move to its remaining use site.
5681 2017-11-10 Jeff Law <law@redhat.com>
5683 * vr-values.h (VR_INITIALIZER): Move #define here.
5684 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
5685 * Makefile.in (OBJS): Add tree-evrp.o
5686 * tree-vrp.h (assert_info): Move structure definition here.
5687 (set_value_range_to_varying): Prototype.
5688 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
5689 (infer_value_range, register_edge_assert_for): Likewise.
5690 (stmt_interesting_for_vrp): Likewise.
5691 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
5692 (set_value_range_to_varying): No longer static.
5693 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
5694 (infer_value_range, register_edge_assert_for): Likewise.
5696 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
5698 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
5699 generate the XXBRD instruction.
5701 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
5702 (rs6000_builtin_type_compatible): Treat _Float128 and long double
5703 as being compatible if -mabi=ieeelongdouble.
5704 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
5705 to setup float128 built-ins with hardware support.
5706 (BU_FLOAT128_HW_2): Likewise.
5707 (BU_FLOAT128_HW_3): Likewise.
5708 (BU_FLOAT128_HW_VSX_1): Likewise.
5709 (BU_FLOAT128_HW_VSX_2): Likewise.
5710 (scalar_extract_expq): Change float128 built-in functions to
5711 accommodate having both KFmode and TFmode functions. Use the
5712 KFmode variant as the default.
5713 (scalar_extract_sigq): Likewise.
5714 (scalar_test_neg_qp): Likewise.
5715 (scalar_insert_exp_q): Likewise.
5716 (scalar_insert_exp_qp): Likewise.
5717 (scalar_test_data_class_qp): Likewise.
5718 (sqrtf128_round_to_odd): Delete processing the round to odd
5719 built-in functions as special built-in functions, and define them
5720 as float128 built-ins. Use the KFmode variant as the default.
5721 (truncf128_round_to_odd): Likewise.
5722 (addf128_round_to_odd): Likewise.
5723 (subf128_round_to_odd): Likewise.
5724 (mulf128_round_to_odd): Likewise.
5725 (divf128_round_to_odd): Likewise.
5726 (fmaf128_round_to_odd): Likewise.
5727 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
5728 support for KFmode and TFmode xststdcqp calls.
5729 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
5730 point, switch the built-in handlers for the get/set float128
5731 exponent, get float128 mantissa, float128 test built-ins, and the
5732 float128 round to odd built-in functions. Eliminate creating the
5733 float128 round to odd built-in functions as special built-ins.
5734 (rs6000_init_builtins): Eliminate special creation of the float128
5735 round to odd built-in functions.
5736 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
5737 function insns to support both TFmode and KFmode variants.
5738 (xsxsigqp_<mode>): Likewise.
5739 (xsiexpqpf_<mode>): Likewise.
5740 (xsiexpqp_<mode>): Likewise.
5741 (xststdcqp_<mode>): Likewise.
5742 (xststdcnegqp_<mode>): Likewise.
5743 (xststdcqp_<mode>): Likewise.
5745 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5747 * tree-ssa-threadupdate.c (compute_path_counts): Remove
5748 unused path_in_freq_ptr parameter.
5749 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
5751 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5753 * ipa-inline.c (edge_badness): Dump sreal frequency.
5754 (compute_inlined_call_time): Match natural implementaiton ...
5755 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
5756 forgotten division by CGRAPH_FREQ_BASE.
5758 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5760 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
5761 Solaris 11. Update comment.
5762 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
5764 * config/sol2.h (STARTFILE_SPEC): Likewise.
5765 * configure: Regenerate.
5767 2017-11-14 Carl Love <cel@us.ibm.com>
5769 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
5770 le_ and be_ prefixes to swap* variables. Remove
5771 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
5774 2017-11-14 Jason Merrill <jason@redhat.com>
5776 Support GTY((cache)) on hash_map.
5777 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
5778 (ggc_cache_remove): Override it instead of ggc_mx.
5779 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
5780 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
5781 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
5782 (simple_cache_map_traits): Override maybe_mx.
5783 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
5784 (hash_map): Friend gt_cleare_cache.
5785 (gt_cleare_cache): New.
5786 * tree.h (tree_cache_traits): New hash_map traits class.
5787 (tree_cache_map): New typedef.
5789 2017-11-14 Richard Biener <rguenther@suse.de>
5791 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
5792 paramter and handling.
5793 (cleanup_control_flow_bb): Likewise.
5794 (cleanup_control_flow_pre): New helper performing a DFS walk
5795 to call cleanup_control_flow_bb in PRE order.
5796 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
5797 via cleanup_control_flow_pre.
5799 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
5801 * config/aarch64/aarch64-simd.md
5802 (aarch64_simd_bsl<mode>_internal): Remove DImode.
5803 (*aarch64_simd_bsl<mode>_alt): Likewise.
5804 (aarch64_simd_bsldi_internal): New.
5805 (aarch64_simd_bsldi_alt): Likewise.
5807 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5809 * tracer.c (better_p): Do not compare frequencies.
5810 * reg-stack.c (better_edge): Likewise.
5811 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
5814 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5816 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
5817 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
5818 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
5819 * lto-streamer-in.c (input_function): Use update_max_bb_count.
5820 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
5821 * predict.c (maybe_hot_frequency_p): Inline to ...
5822 (maybe_hot_count_p): ... here; rewrite to counts.
5823 (counts_to_freqs): Rename to ...
5824 (update_max_bb_count): ... this one.
5825 (expensive_function_p): Use counts.
5826 (estimate_bb_frequencies): Update.
5827 (rebuild_frequencies): Update.
5828 * predict.h (counts_to_freqs): Rename to ...
5829 (update_max_bb_count): ... this one.
5830 * profile.c (compute_branch_probabilities): Add debug info
5831 * tree-inline.c (expand_call_inline): Update debug info.
5832 (optimize_inline_calls): Use update_max_bb_count..
5833 (tree_function_versioning): Use update_max_bb_count..
5834 * value-prof.c (gimple_value_profile_transformations):
5835 Do not use update_max_bb_count.
5837 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5839 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
5840 always use frequencies.
5842 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5844 * bb-reorder.c: Remove frequencies from comments.
5845 (better_edge_p): Use profile counts.
5846 (find_traces): Dump profile counts.
5847 (rotate_loop): Use profile counts.
5848 (find_traces_1_round): Likewise.
5849 (connect_better_edge_p): Use counts instead of probabilities for
5851 (copy_bb_p): Drop early check for non-0 frequency.
5852 (sanitize_hot_paths): Update comments.
5854 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5856 * ipa-split.c (struct split_point): Add count.
5857 (consider_split): Do not compute incoming frequency; compute incoming
5858 count and store it to split_point.
5859 (split_function): Set count of the call to split part correctly.
5861 2017-11-13 Carl Love <cel@us.ibm.com>
5863 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
5865 2017-11-13 Tom Tromey <tom@tromey.com>
5867 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
5869 2017-11-13 Carl Love <cel@us.ibm.com>
5871 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
5872 Add support for builtins:
5873 unsigned int vec_first_{,miss}_match_{,or_eos}index,
5874 vector {un,}signed {char,int,short},
5875 vector {un,}signed {char,int,short}) arguments.
5876 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
5877 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
5878 Add BU_P9V_AV_2 expansions for the builtins.
5879 * config/rs6000/altivec.h (vec_first_match_index,
5880 vec_first_mismatch_index, vec_first_match_or_eos_index,
5881 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
5882 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
5883 new extern declaration.
5884 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
5885 * config/rs6000/vsx.md (first_match_index_<mode>,
5886 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
5887 first_mismatch_or_eos_index_<mode>): Add define expand.
5888 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
5889 * doc/extend.texi: Update the built-in documenation file for the new
5892 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
5894 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
5895 into the min/max operations for _Float<N> and _Float<N>X types.
5897 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
5900 * dwarf2out.c (do_eh_frame): New static variable.
5901 (dwarf2out_begin_prologue): Set it.
5902 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
5904 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5906 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
5908 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
5910 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
5913 * debug.h (dwarf2out_do_eh_frame): Declare.
5914 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
5915 (dwarf2out_do_frame): Use it.
5916 (dwarf2out_do_cfi_asm): Likewise.
5917 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
5918 (dwarf2out_assembly_start): Likewise.
5919 (dwarf2out_begin_prologue): Fix comment.
5920 * toplev.c (compile_file): Always call dwarf2out_frame_finish
5921 if the target needs either debug or unwind DWARF2 info.
5922 * lto-opts.c (lto_write_options): Do not save -fexceptions,
5923 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
5924 -ftrapping-math, -ftrapv and -fwrapv.
5926 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5928 * cgraph.c (cgraph_edge::sreal_frequency): New function.
5929 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
5930 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
5931 (estimate_edge_size_and_time): Likewise.
5932 (ipa_merge_fn_summary_after_inlining): Likewise.
5933 * ipa-inline.c (cgraph_freq_base_rec): Remove.
5934 (compute_uninlined_call_time): Use sreal_frequency.
5935 (compute_inlined_call_time): Likewise.
5936 (ipa_inline): Do not initialize cgraph_freq_base_rec.
5937 * profile-count.c: Include sreal.h.
5938 (profile_count::to_sreal_scale): New.
5939 * profile-count.h: Forward declare sreal.
5940 (profile_count::to_sreal_scale): Declare.
5942 2017-11-13 Nathan Sidwell <nathan@acm.org>
5944 * diagnostic.c (maybe_line_and_column): New.
5945 (diagnostic_get_location_text): Use it.
5946 (diagnostic_report_current_module): Likewise.
5947 (test_diagnostic_get_location_text): Add tests.
5949 2017-11-13 Luis Machado <luis.machado@linaro.org>
5951 * doc/md.texi (Specifying processor pipeline description): Fix
5952 incorrect latency for the div instruction example.
5954 2017-11-13 Jakub Jelinek <jakub@redhat.com>
5956 PR tree-optimization/78821
5957 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
5958 that bit_not_p is the same.
5959 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
5960 (split_group): Count precisely bit_not_p bits in each statement.
5961 (invert_op): New function.
5962 (imm_store_chain_info::output_merged_store): Use invert_op to
5963 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
5964 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
5966 2017-11-13 Martin Liska <mliska@suse.cz>
5968 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
5969 (struct source_info): Likewise.
5970 (add_branch_counts): Likewise.
5971 (add_line_counts): Likewise.
5972 (function_summary): Likewise.
5973 (output_intermediate_line): Likewise.
5974 (generate_results): Likewise.
5976 2017-11-13 Martin Liska <mliska@suse.cz>
5978 * gcov.c (struct block_info): Remove typedef for block_t.
5979 (struct line_info): Likewise.
5980 (line_info::has_block): Likewise.
5981 (EXIT_BLOCK): Likewise.
5982 (unblock): Likewise.
5983 (circuit): Likewise.
5984 (get_cycles_count): Likewise.
5985 (process_file): Likewise.
5986 (read_graph_file): Likewise.
5987 (solve_flow_graph): Likewise.
5988 (find_exception_blocks): Likewise.
5989 (add_line_counts): Likewise.
5990 (accumulate_line_info): Likewise.
5991 (output_line_details): Likewise.
5993 2017-11-13 Martin Liska <mliska@suse.cz>
5995 * gcov.c (struct arc_info): Remove typedef for arc_t.
5996 (struct line_info): Likewise.
5997 (add_branch_counts): Likewise.
5998 (output_branch_count): Likewise.
5999 (function_info::~function_info): Likewise.
6000 (circuit): Likewise.
6001 (output_intermediate_line): 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.
6008 (output_function_details): Likewise.
6010 2017-11-13 Martin Liska <mliska@suse.cz>
6012 * gcov.c (struct function_info): Remove typedef for function_t.
6013 (struct source_info): Likewise.
6014 (source_info::get_functions_at_location): Likewise.
6015 (solve_flow_graph): Likewise.
6016 (find_exception_blocks): Likewise.
6017 (add_line_counts): Likewise.
6018 (output_intermediate_file): Likewise.
6019 (process_file): Likewise.
6020 (generate_results): Likewise.
6021 (release_structures): Likewise.
6022 (read_graph_file): Likewise.
6023 (read_count_file): Likewise.
6024 (accumulate_line_counts): Likewise.
6025 (output_lines): Likewise.
6027 2017-11-13 Martin Liska <mliska@suse.cz>
6029 * gcov.c (function_info::function_info): Remove num_counts
6030 and add vector<gcov_type>.
6031 (function_info::~function_info): Use the vector.
6032 (process_file): Likewise.
6033 (read_graph_file): Likewise.
6034 (read_count_file): Likewise.
6035 (solve_flow_graph): Likewise.
6037 2017-11-13 Martin Liska <mliska@suse.cz>
6039 * gcov.c (function_info::is_artificial): New function.
6040 (process_file): Erase all artificial early.
6041 (generate_results): Skip as all artificial are already
6044 2017-11-13 Martin Liska <mliska@suse.cz>
6046 * gcov.c (read_graph_file): Store to global vector of functions.
6047 (read_count_file): Iterate the vector.
6048 (process_file): Likewise.
6049 (generate_results): Likewise.
6050 (release_structures): Likewise.
6052 2017-11-13 Jakub Jelinek <jakub@redhat.com>
6054 PR tree-optimization/82954
6055 * gimple-ssa-store-merging.c
6056 (imm_store_chain_info::coalesce_immediate_stores): If
6057 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
6059 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
6061 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
6062 Upddate call to ENDIAN_LANE_N.
6063 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
6064 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
6065 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
6066 and use aarch64_endian_lane_rtx.
6067 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
6069 2017-11-12 Tom de Vries <tom@codesourcery.com>
6071 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
6073 2017-11-12 Tom de Vries <tom@codesourcery.com>
6075 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
6077 2017-11-12 Tom de Vries <tom@codesourcery.com>
6079 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
6081 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
6082 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
6083 * defaults.h (ASM_OUTPUT_LABELREF): Same.
6085 2017-11-11 Martin Sebor <msebor@redhat.com>
6088 * doc/extend.texi (attribute nonstring): Remove spurious argument.
6091 * prefic.c (translate_name): Replace strncpy with memcpy to
6092 avoid -Wstringop-truncation.
6094 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6096 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
6098 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6100 * predict.c (maybe_hot_frequency_p): Do not use cfun.
6102 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6104 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
6107 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
6109 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
6110 enable generating XXBRH if the value is in a vector register.
6111 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
6112 value is in a vector register.
6113 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
6114 register to register bswap64's instead of doing the GPR sequence
6115 used on previous machines.
6116 (bswapdi2_xxbrd): New insn.
6117 (bswapdi2_reg): Disallow on ISA 3.0.
6118 (register to register bswap64 splitter): Do not split the insn on
6119 ISA 3.0 systems that use XXBRD.
6121 2017-11-10 Martin Sebor <msebor@redhat.com>
6124 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
6126 (find_subframework_file): Same.
6128 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6130 * auto-profile.c (afdo_indirect_call): Drop frequency.
6131 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
6132 (cgraph_node::create_edge): Drop frequency argument.
6133 (cgraph_node::create_indirect_edge): Drop frequency argument.
6134 (cgraph_edge::make_speculative): Drop frequency arguments.
6135 (cgraph_edge::resolve_speculation): Do not update frequencies
6136 (cgraph_edge::dump_edge_flags): Do not dump frequency.
6137 (cgraph_node::dump): Check consistency in IPA mode.
6138 (cgraph_edge::maybe_hot_p): Use IPA counter.
6139 (cgraph_edge::verify_count_and_frequency): Rename to ...
6140 (cgraph_edge::verify_count): ... this one; drop frequency checking.
6141 (cgraph_node::verify_node): Update.
6142 * cgraph.h (struct cgraph_edge): Drop frequency.
6143 (cgraph_edge::frequency): New function.
6144 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
6146 (cgraph_edge::rebuild_edges): Likewise.
6147 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
6148 (duplicate_thunk_for_node): Do not pass frequency.
6149 (cgraph_node::create_clone): Scale only counts.
6150 (cgraph_node::create_virtual_clone): Do not pass frequency.
6151 (cgraph_node::create_edge_including_clones): Do not pass frequency.
6152 (cgraph_node::create_version_clone): Do not pass frequency.
6153 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
6154 (cgraph_node::expand_thunk): Do not pass frequency.
6155 (cgraph_node::create_wrapper): Do not pass frequency.
6156 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
6158 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
6159 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
6160 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
6161 (gather_caller_stats): Use frequency function.
6162 (ipcp_cloning_candidate_p): Use frequency function.
6163 (ipcp_propagate_stage): Use frequency function.
6164 (get_info_about_necessary_edges): Use frequency function.
6165 (update_profiling_info): Update only IPA profile.
6166 (update_specialized_profile): Use frequency functoin.
6167 (perhaps_add_new_callers): Update only IPA profile.
6168 * ipa-devirt.c (ipa_devirt): Use IPA profile.
6169 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
6170 (dump_ipa_call_summary): Use frequency function.
6171 (estimate_edge_size_and_time): Use frequency function.
6172 (ipa_merge_fn_summary_after_inlining): Use frequency function.
6173 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
6174 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
6175 (update_noncloned_counts): ... ths one; scale counts only.
6176 (clone_inlined_nodes): Do not scale frequency.
6177 (inline_call): Do not pass frequency.
6178 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
6179 (compute_inlined_call_time): Use IPA profile.
6180 (want_inline_small_function_p): Use IPA profile.
6181 (want_inline_self_recursive_call_p): Use IPA profile.
6182 (edge_badness): Use IPA profile.
6183 (lookup_recursive_calls): Use IPA profile.
6184 (recursive_inlining): Do not pass frequency.
6185 (resolve_noninline_speculation): Do not update frequency.
6186 (inline_small_functions): Collect max of IPA profile.
6187 (dump_overall_stats): Dump IPA porfile.
6188 (dump_inline_stats): Dump IPA porfile.
6189 (ipa_inline): Collect IPA stats.
6190 * ipa-inline.h (clone_inlined_nodes): Update prototype.
6191 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
6192 (ipa_propagate_frequency): Use frequency function.
6193 (ipa_profile): Cleanup.
6194 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
6195 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
6196 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
6197 (input_node): Do not stream frequency.
6198 (input_edge): Do not stream frequency.
6199 (merge_profile_summaries): Scale only IPA profiles.
6200 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
6201 * predict.c (drop_profile): Do not recompute frequency.
6202 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
6203 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
6204 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
6205 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
6206 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
6207 * tree-inline.c (copy_bb): Do not scale frequency.
6208 (expand_call_inline): Do not scale frequency.
6209 (tree_function_versioning): Do not scale frequency.
6210 * ubsan.c (ubsan_create_edge): Do not pass frequency.
6212 2017-11-10 Julia Koval <julia.koval@intel.com>
6214 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
6215 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
6216 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
6217 (_mm256_maskz_gf2p8affine_epi64_epi8)
6218 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
6219 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
6220 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
6221 (__builtin_ia32_vgf2p8affineqb_v32qi)
6222 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
6223 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
6225 2017-11-10 Tamar Christina <tamar.christina@arm.com>
6229 (arm_option_override): Refactor.
6230 (arm_option_reconfigure_globals): New.
6231 (arm_options_perform_arch_sanity_checks): New.
6232 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
6234 (arm_options_perform_arch_sanity_checks): Likewise
6236 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
6238 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
6239 (power9-qpmul): New.
6240 * rs6000/rs6000.md ("type" attr): Add qmul.
6241 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
6242 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
6243 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
6245 2017-11-10 Martin Sebor <msebor@redhat.com>
6248 * builtins.c (compute_objsize): Handle arrays that
6249 compute_builtin_object_size likes to fail for. Make extern.
6250 * builtins.h (compute_objsize): Declare.
6251 (check_strncpy_sizes): New function.
6252 (expand_builtin_strncpy): Call check_strncpy_sizes.
6253 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
6254 -Wstringop-truncation.
6255 (gimple_fold_builtin_strncat): Same.
6256 * gimple.c (gimple_build_call_from_tree): Set call location.
6257 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
6258 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
6259 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
6260 (handle_builtin_strlen): Use strlen_to_stridx.
6261 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
6263 Use strlen_to_stridx.
6264 (pass_strlen::execute): Release strlen_to_stridx.
6265 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
6266 (-Wstringop-truncation): Document new option.
6268 2017-11-10 Martin Liska <mliska@suse.cz>
6270 PR gcov-profile/82702
6271 * gcov.c (main): Handle intermediate files in a different
6273 (get_gcov_intermediate_filename): New function.
6274 (output_gcov_file): Remove support of intermediate files.
6275 (generate_results): Allocate intermediate file.
6276 (release_structures): Clean-up properly fn_end.
6277 (output_intermediate_file): Start iterating with line 1.
6279 2017-11-10 Jakub Jelinek <jakub@redhat.com>
6281 PR tree-optimization/82929
6282 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
6283 ops_swapped_p non-static data member.
6284 (store_immediate_info::store_immediate_info): Clear it.
6285 (imm_store_chain_info::coalesce_immediate_stores): If swapping
6286 ops set ops_swapped_p.
6287 (count_multiple_uses): Handle ops_swapped_p.
6289 2017-11-10 Martin Liska <mliska@suse.cz>
6291 * coverage.c (coverage_init): Stream information about
6292 support of has_unexecuted_blocks.
6293 * doc/gcov.texi: Document that.
6294 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
6295 * gcov.c (read_graph_file): Likewise.
6296 (output_line_beginning): Fix a small issue with
6299 2017-11-10 Bin Cheng <bin.cheng@arm.com>
6301 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
6302 reference of trivial component.
6304 2017-11-10 Jakub Jelinek <jakub@redhat.com>
6307 * gimple-ssa-store-merging.c
6308 (pass_store_merging::terminate_all_aliasing_chains): For
6309 gimple_store_p stmts also call refs_output_dependent_p.
6311 PR rtl-optimization/82913
6312 * compare-elim.c (try_merge_compare): Punt if def_insn is not
6315 2017-11-09 Jeff Law <law@redhat.com>
6317 * vr-values.h: New file with vr_values class.
6318 * tree-vrp.c: Include vr-values.h
6319 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
6320 data objects into the vr_values class.
6321 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
6322 (get_value_range): Make it a member function within vr_values class.
6323 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
6324 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
6325 (op_with_constant_singleton_value_range): Likewise.
6326 (extract_range_for_var_from_comparison_expr): Likewise.
6327 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
6328 (extract_range_from_binary_expr): Likewise.
6329 (extract_range_from_unary_expr): Likewise.
6330 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
6331 (check_for_binary_op_overflow, extract_range_basic): Likewise.
6332 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
6333 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
6334 (compare_name_with_value, compare_names): Likewise.
6335 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
6336 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
6337 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
6338 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
6339 (extract_range_from_phi_node): Likewise.
6340 (simplify_truth_ops_using_ranges): Likewise.
6341 (simplify_div_or_mod_using_ranges): Likewise.
6342 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
6343 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
6344 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
6345 (simplify_float_conversion_using_ranges): Likewise.
6346 (simplify_internal_call_using_ranges): Likewise.
6347 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
6348 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
6349 poitner via x_vr_values for calls into gimple folder.
6350 (vrp_initialize_lattice): Make this the vr_values ctor.
6351 (vrp_free_lattice): Make this the vr_values dtor.
6352 (set_vr_value): New function.
6353 (class vrp_prop): Add vr_values data member. Add various member
6354 functions as well as member functions that delegate to vr_values.
6355 (check_array_ref): Make a member function within vrp_prop class.
6356 (search_for_addr_array, vrp_initialize): Likewise.
6357 (vrp_finalize): Likewise. Revamp to avoid direct access to
6358 vr_value, values_propagated, etc.
6359 (check_array_bounds): Extract vrp_prop class instance pointer from
6360 walk info structure. Use it to call member functions.
6361 (check_all_array_refs): Make a member function within vrp_prop class.
6362 Smuggle class instance pointer via walk info structure.
6363 (x_vr_values): New local static.
6364 (vrp_valueize): Use x_vr_values to get class instance.
6365 (vr_valueize_1): Likewise.
6366 (class vrp_folder): Add vr_values data member. Add various member
6367 functions as well as member functions that delegate to vr_values.
6368 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
6369 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
6370 class instance from vr_values. Use it to call member functions.
6371 (vrp_dom_walker): Add vr_values data member.
6372 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
6373 instance via x_vr_values.
6374 (identify_jump_threads): Accept vr_values as argument. Store
6375 it into the walker structure.
6376 (evrp_dom_walker): Add vr_values class data member. Add various
6378 (evrp_dom_walker::try_find_new_range): Use vr_values data
6379 member to access the memory allocator.
6380 (evrp_dom_walker::before_dom_children): Store vr_values class
6381 instance into the vrp_folder class.
6382 (evrp_dom_walker::push_value_range): Rework to avoid direct
6383 access to num_vr_values and vr_value.
6384 (evrp_dom_walker::pop_value_range): Likewise.
6385 (execute_early_vrp): Remove call to vrp_initialize_lattice.
6386 Use vr_values to get to dump_all_value_ranges member function.
6387 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
6388 and simplify_cond_using_ranges_2 via vrp_prop class instance.
6389 Pass vr_values class instance down to identify_jump_threads.
6390 Remove call to vrp_free_lattice.
6391 (debug_all_value_ranges): Remove.
6393 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
6394 (vrp_folder): Likewise.
6396 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
6397 Get it from the existing bitmap instead.
6398 (vrp_intersect_ranges_1): Likewise.
6400 2017-11-09 Jakub Jelinek <jakub@redhat.com>
6402 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
6404 (store_immediate_info::store_immediate_info): Add bitnotp argument,
6405 set bit_not_p to it.
6406 (imm_store_chain_info::coalesce_immediate_stores): Break group
6407 if bit_not_p is different.
6408 (count_multiple_uses, split_group,
6409 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
6410 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
6411 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
6412 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
6413 to store_immediate_info ctor.
6415 2017-11-09 Jim Wilson <jimw@sifive.com>
6417 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
6418 (scan_prog_file): Likewise.
6420 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
6422 * bb-reorder.c (max_entry_frequency): Remove.
6423 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
6424 connect_traces, push_to_next_round_p): Remove prototypes.
6425 (find_traces_1_round): Use counts only.
6426 (push_to_next_round_p): Likewise.
6427 (find_traces): Likewise.
6428 (rotate_loop): Likewise.
6429 (find_traces_1_round): Likewise.
6430 (connect_traces): Likewise.
6431 (edge_order): Likewise.
6433 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
6435 * config/arm/arm.c (output_return_instruction): Add comments to
6436 indicate requirement for cmse_nonsecure_entry return to account
6437 for the size of clearing instruction output here.
6438 (thumb_exit): Likewise.
6439 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
6440 return in hardfloat mode.
6442 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
6444 * config/rs6000/rs6000.c (machine_function): Add a bool,
6445 "toc_is_wrapped_separately".
6446 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
6447 if it wasn't explicitly set or unset, we are optimizing for speed, and
6448 doing separate shrink-wrapping.
6449 (rs6000_get_separate_components): Enable the TOC component if
6450 saving the TOC register in the prologue.
6451 (rs6000_components_for_bb): Handle the TOC component.
6452 (rs6000_emit_prologue_components): Store the TOC register where needed.
6453 (rs6000_set_handled_components): Mark TOC as handled, if handled.
6454 (rs6000_emit_prologue): Don't save the TOC if that is already done.
6456 2017-11-09 Martin Jambor <mjambor@suse.cz>
6458 * ipa-param-manipulation.c: New file.
6459 * ipa-param-manipulation.h: Likewise.
6460 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
6461 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
6462 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
6463 (ipa_parm_adjustment): Likewise.
6464 (ipa_parm_adjustment_vec): Likewise.
6465 (ipa_get_vector_of_formal_parms): Moved declaration to
6466 ipa-param-manipulation.h.
6467 (ipa_get_vector_of_formal_parm_types): Likewise.
6468 (ipa_modify_formal_parameters): Likewise.
6469 (ipa_modify_call_arguments): Likewise.
6470 (ipa_combine_adjustments): Likewise.
6471 (ipa_dump_param_adjustments): Likewise.
6472 (ipa_modify_expr): Likewise.
6473 (ipa_get_adjustment_candidate): Likewise.
6474 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
6475 ipa-param-manipulation.c.
6476 (ipa_get_vector_of_formal_parm_types): Likewise.
6477 (ipa_modify_formal_parameters): Likewise.
6478 (ipa_modify_call_arguments): Likewise.
6479 (ipa_modify_expr): Likewise.
6480 (get_ssa_base_param): Likewise.
6481 (ipa_get_adjustment_candidate): Likewise.
6482 (index_in_adjustments_multiple_times_p): Likewise.
6483 (ipa_combine_adjustments): Likewise.
6484 (ipa_dump_param_adjustments): Likewise.
6485 * tree-sra.c: Also include ipa-param-manipulation.h
6486 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
6489 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6490 Alan Hayward <alan.hayward@arm.com>
6491 David Sherwood <david.sherwood@arm.com>
6493 * doc/sourcebuild.texi (vect_masked_store): Document.
6495 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6496 Alan Hayward <alan.hayward@arm.com>
6497 David Sherwood <david.sherwood@arm.com>
6499 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
6501 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6502 Alan Hayward <alan.hayward@arm.com>
6503 David Sherwood <david.sherwood@arm.com>
6505 * doc/sourcebuild.texi (vect_variable_length): Document.
6507 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6508 Alan Hayward <alan.hayward@arm.com>
6509 David Sherwood <david.sherwood@arm.com>
6511 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
6513 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6514 Alan Hayward <alan.hayward@arm.com>
6515 David Sherwood <david.sherwood@arm.com>
6517 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
6519 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6520 Alan Hayward <alan.hayward@arm.com>
6521 David Sherwood <david.sherwood@arm.com>
6523 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
6524 previously undocumented selectors.
6525 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
6527 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6529 * doc/rtl.texi (const_vector): Say that elements can be
6530 const_wide_ints too.
6531 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
6532 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
6533 (gen_vec_duplicate): Use it instead of CONSTANT_P.
6534 * optabs.c (expand_vector_broadcast): Likewise.
6536 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6537 Alan Hayward <alan.hayward@arm.com>
6538 David Sherwood <david.sherwood@arm.com>
6540 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
6541 scaled index even if the unscaled address was invalid.
6542 Don't increase the complexity of using a scale in that case.
6544 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6545 Alan Hayward <alan.hayward@arm.com>
6546 David Sherwood <david.sherwood@arm.com>
6548 * doc/rtl.texi: Rewrite the subreg rules so that they partition
6549 the inner register into REGMODE_NATURAL_SIZE bytes rather than
6550 UNITS_PER_WORD bytes.
6551 * emit-rtl.c (validate_subreg): Divide subregs into blocks
6552 based on REGMODE_NATURAL_SIZE of the inner mode.
6553 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
6554 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
6555 * expmed.c (lowpart_bit_field_p): Divide the value up into
6556 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
6557 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
6558 whether something is likely to occupy more than one register.
6560 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
6563 * ipa-inline-transform.c (update_noncloned_frequencies): Use
6564 profile_count::adjust_for_ipa_scaling.
6565 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
6566 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
6568 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
6570 2017-11-09 Jakub Jelinek <jakub@redhat.com>
6572 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
6573 (split_group): Add total_orig and total_new arguments, estimate the
6574 number of statements related to the store group without store merging
6575 and with store merging.
6576 (imm_store_chain_info::output_merged_store): Adjust split_group
6577 callers, punt if estimated number of statements with store merging
6578 is not smaller than estimated number of statements without it.
6580 (handled_load): Remove has_single_use checks.
6581 (pass_store_merging::process_store): Likewise.
6583 2017-11-09 Richard Biener <rguenther@suse.de>
6585 PR tree-optimization/82902
6586 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
6588 2017-11-09 Martin Liska <mliska@suse.cz>
6591 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
6594 2017-11-09 Martin Liska <mliska@suse.cz>
6596 PR tree-optimization/82669
6597 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
6599 2017-11-09 Martin Liska <mliska@suse.cz>
6601 PR gcov-profile/48463
6602 * coverage.c (coverage_begin_function): Output also end locus
6603 of a function and information whether the function is
6605 * gcov-dump.c (tag_function): Parse and print the information.
6606 * gcov.c (INCLUDE_MAP): Add include.
6607 (INCLUDE_SET): Likewise.
6608 (struct line_info): Move earlier in the source file because
6609 of vector<line_info> in function_info structure.
6610 (line_info::line_info): Likewise.
6611 (line_info::has_block): Likewise.
6612 (struct source_info): Add new member index.
6613 (source_info::get_functions_at_location): New function.
6614 (function_info::group_line_p): New function.
6615 (output_intermediate_line): New function.
6616 (output_intermediate_file): Use the mentioned function.
6617 (struct function_start): New.
6618 (struct function_start_pair_hash): Likewise.
6619 (process_file): Add code that identifies group functions.
6620 Assign lines either to global or function scope.
6621 (generate_results): Skip artificial functions.
6622 (find_source): Assign index for each source file.
6623 (read_graph_file): Read new flag artificial and end_line.
6624 (add_line_counts): Assign it either to global of function scope.
6625 (accumulate_line_counts): Isolate core of the function to
6626 accumulate_line_info and call it for both function and global
6628 (accumulate_line_info): New function.
6629 (output_line_beginning): Fix GNU coding style.
6630 (print_source_line): New function.
6631 (output_line_details): Likewise.
6632 (output_function_details): Likewise.
6633 (output_lines): Iterate both source (global) scope and function
6635 (struct function_line_start_cmp): New class.
6636 * doc/gcov.texi: Reflect changes in documentation.
6638 2017-11-09 Jakub Jelinek <jakub@redhat.com>
6641 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
6642 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
6643 and similarly for not instead of neg.
6645 2017-11-08 Andi Kleen <ak@linux.intel.com>
6647 * config/i386/i386.opt: Add -mforce-indirect-call.
6648 * config/i386/predicates.md: Check for flag_force_indirect_call.
6649 * doc/invoke.texi: Document -mforce-indirect-call
6651 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
6653 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
6655 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
6656 config/riscv/riscv.c (predict.h): New include.
6657 (riscv_slow_unaligned_access_p): No longer static.
6658 (riscv_block_move_straight): Add require.
6659 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
6661 2017-11-08 Jakub Jelinek <jakub@redhat.com>
6664 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
6665 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
6666 nonimmediate_operand predicate for operand 1 instead of
6669 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6671 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
6673 * config/aarch64/constraints.md (Uml): New constraint.
6674 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
6677 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6679 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
6680 of two vec_duplicates into a vec_concat.
6682 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6684 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
6685 Simplify vec_merge of vec_duplicate and vec_concat.
6686 * config/aarch64/constraints.md (Utq): New constraint.
6687 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
6690 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6692 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
6693 Simplify vec_merge of vec_duplicate and const_vector.
6694 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
6696 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
6697 mode iterator. Update predicate on operand 1 to
6698 handle non-const_vec constants. Delete constraints.
6699 (*aarch64_combinez_be<mode>): Likewise for operand 2.
6701 2017-11-08 Jakub Jelinek <jakub@redhat.com>
6703 PR tree-optimization/78821
6704 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
6706 (store_operand_info::store_operand_info): Initialize it to false.
6707 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
6708 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
6709 store in the group, and if chain_info is non-NULL, to ignore altogether
6711 (compatible_load_p): Fail if bit_not_p does not match.
6712 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
6713 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
6714 (pass_store_merging::process_store): Adjust
6715 terminate_all_aliasing_chains calls to pass NULL in all current spots,
6716 call terminate_all_aliasing_chains newly when adding a store into
6717 a chain with non-NULL chain_info.
6719 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
6721 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
6723 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
6725 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
6727 (aarch64_layout_frame): Initialise emit_frame_chain.
6728 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
6729 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
6731 2017-11-08 Martin Liska <mliska@suse.cz>
6733 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
6734 of gimple_call_internal_p.
6736 2017-11-07 Tom de Vries <tom@codesourcery.com>
6738 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
6740 2017-11-07 Tom de Vries <tom@codesourcery.com>
6742 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
6745 2017-11-08 Martin Liska <mliska@suse.cz>
6748 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
6750 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
6752 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
6754 2017-11-07 Jakub Jelinek <jakub@redhat.com>
6757 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
6758 (*cmp<mode>_ccz_1): New insn with $k alternative.
6761 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
6762 RTX_COMM_COMPARE as commutative as well.
6763 (ix86_binary_operator_ok): Formatting fix.
6764 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
6765 *<code><mode>3<mask_name><round_saeonly_name>,
6766 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
6767 *<s>mul<mode>3_highpart<mask_name>,
6768 *vec_widen_umult_even_v16si<mask_name>,
6769 *vec_widen_umult_even_v8si<mask_name>,
6770 *vec_widen_umult_even_v4si<mask_name>,
6771 *vec_widen_smult_even_v16si<mask_name>,
6772 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
6773 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
6774 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
6775 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
6776 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
6777 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
6778 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
6779 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
6780 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
6781 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
6782 ix86_binary_operator_ok. Formatting fixes.
6783 (*<plusminus_insn><mode>3<mask_name><round_name>,
6784 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
6787 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
6789 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
6790 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
6791 eq and ne if TARGET_ISEL.
6792 (cmp): New code_iterator.
6793 (UNS, UNSU_, UNSIK): New code_attrs.
6794 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
6795 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
6797 ("eq<mode>3"): ... this.
6798 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
6800 ("ne<mode>3"): ... this.
6802 2017-11-07 Julia Koval <julia.koval@intel.com>
6805 * common/config/i386/i386-common.c
6806 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
6807 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
6808 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
6809 * config/i386/i386.opt: Ditto.
6810 * config/i386/i386.c (ix86_target_string): Ditto.
6811 (ix86_option_override_internal): Ditto.
6812 (ix86_init_mpx_builtins): Move MPX to args2.
6813 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
6814 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
6815 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
6816 __builtin_ia32_vgf2p8affineinvqb_v32qi,
6817 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
6818 __builtin_ia32_vgf2p8affineinvqb_v16qi,
6819 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
6821 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
6824 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
6825 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
6826 (zero-extendsidi peephole2): Remove peephole.
6828 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
6831 * stor-layout.c: Include attribs.h.
6832 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
6833 explicit lookup of "aligned" attribute.
6835 2017-11-07 Andrew Waterman <andrew@sifive.com>
6837 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
6838 (riscv_expand_block_move): Likewise.
6839 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
6841 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
6842 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
6843 * config/riscv/riscv.c (riscv_block_move_straight): New function.
6844 (riscv_adjust_block_mem): Likewise.
6845 (riscv_block_move_loop): Likewise.
6846 (riscv_expand_block_move): Likewise.
6847 * config/riscv/riscv.md (movmemsi): New pattern.
6849 2017-11-07 Michael Clark <michaeljclark@mac.com>
6851 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
6852 (MUSL_DYNAMIC_LINKER): Likewise.
6854 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
6856 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
6857 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
6860 2017-11-07 Richard Biener <rguenther@suse.de>
6862 * match.pd: Fix build.
6864 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
6865 Jackson Woodruff <jackson.woodruff@arm.com>
6867 PR tree-optimization/71026
6868 * match.pd: Canonicalize negate in division.
6870 2017-11-07 Sudakshina Das <sudi.das@arm.com>
6873 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
6875 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
6877 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
6878 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
6880 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
6882 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
6883 non-scalar integral types.
6884 * match.pd (negate_expr_p): Handle MINUS_EXPR.
6885 (-(A-B), -(~A)): New transformations.
6887 2017-11-07 Tom de Vries <tom@codesourcery.com>
6889 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
6890 semicolon after "do {} while (0)".
6891 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6892 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6893 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6894 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6895 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6897 2017-11-07 Tom de Vries <tom@codesourcery.com>
6899 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
6900 after "do {} while (0)".
6901 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6902 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6903 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6904 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6905 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6907 2017-11-07 Tom de Vries <tom@codesourcery.com>
6910 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
6912 (arm_rtx_costs_internal): Add missing semicolon after
6913 HANDLE_NARROW_SHIFT_ARITH call.
6915 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
6917 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
6918 disable isel if it was not set explicitly.
6920 2017-11-06 James Bowman <james.bowman@ftdichip.com>
6922 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
6923 (add_type_attribute) likewise.
6925 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
6927 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
6928 of struct ix86_frame.
6929 (ix86_initial_elimination_offset): Likewise.
6930 (ix86_expand_split_stack_prologue): Likewise.
6932 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
6934 * tree-vrp.h (enum value_range_type): Update stale comment.
6936 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
6937 Alan Hayward <alan.hayward@arm.com>
6938 David Sherwood <david.sherwood@arm.com>
6940 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
6941 (aarch64_expand_vec_perm_const): Take the number of units too.
6942 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
6943 (aarch64_expand_vec_perm_const): Likewise.
6944 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
6945 (vec_perm<mode>): Update accordingly.
6947 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
6948 Alan Hayward <alan.hayward@arm.com>
6949 David Sherwood <david.sherwood@arm.com>
6951 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
6952 Take the number of units too.
6953 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
6954 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
6955 but check for a vector mode before rather than after the call.
6956 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
6957 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
6958 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
6959 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
6960 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
6961 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
6962 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
6963 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
6964 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
6965 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
6966 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
6967 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
6968 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
6969 (aarch64_sqdmull2_n<mode>): Update accordingly.
6971 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
6972 Alan Hayward <alan.hayward@arm.com>
6973 David Sherwood <david.sherwood@arm.com>
6975 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
6976 the number of units too.
6977 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
6978 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
6979 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
6980 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
6981 (vec_store_lanesxi<mode>): Update accordingly.
6983 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
6984 Alan Hayward <alan.hayward@arm.com>
6985 David Sherwood <david.sherwood@arm.com>
6987 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
6988 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
6989 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
6990 of units rather than the mode.
6991 * config/aarch64/iterators.md (nunits): New mode attribute.
6992 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
6993 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
6994 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
6995 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
6996 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
6997 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
6998 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
6999 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
7000 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
7001 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
7002 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
7003 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
7004 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
7005 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
7006 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
7007 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
7008 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
7009 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
7010 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
7011 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
7012 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
7013 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
7014 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
7015 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
7016 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
7017 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
7018 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
7019 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
7020 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
7021 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
7022 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
7023 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
7024 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
7025 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
7026 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
7027 (aarch64_simd_vec_setv2di): Likewise.
7029 2017-11-06 Carl Love <cel@us.ibm.com>
7031 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
7033 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
7034 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
7036 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
7038 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
7039 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
7040 Add power 8 macro expansions.
7041 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
7042 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
7043 power 8 instructions. (VSX_XXBR): Add iterator.
7045 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
7047 * config/arm/arm.md (predicable_short_it): Change default to "no",
7048 improve documentation, remove uses that are identical to the default.
7049 (enabled_for_depr_it): Rename to enabled_for_short_it.
7050 * gcc/config/arm/arm-fixed.md (predicable_short_it):
7051 Remove default uses.
7052 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
7053 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
7054 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
7055 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
7057 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
7060 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
7061 float128 helper macros, which are no longer used after deleting
7062 the old 'q' built-in functions, and moving the round to odd
7063 built-in functions to being special built-in functions.
7064 (BU_FLOAT128_2): Likewise.
7065 (BU_FLOAT128_1_HW): Likewise.
7066 (BU_FLOAT128_2_HW): Likewise.
7067 (BU_FLOAT128_3_HW): Likewise.
7068 (FABSQ): Delete old 'q' built-in functions.
7069 (COPYSIGNQ): Likewise.
7070 (SQRTF128_ODD): Move round to odd built-in functions to be
7071 special built-in functions, so that we can handle
7072 -mabi=ieeelongdouble.
7073 (TRUNCF128_ODD): Likewise.
7074 (ADDF128_ODD): Likewise.
7075 (SUBF128_ODD): Likewise.
7076 (MULF128_ODD): Likewise.
7077 (DIVF128_ODD): Likewise.
7078 (FMAF128_ODD): Likewise.
7079 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
7080 built-in names to 'f128'.
7081 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
7082 old 'q' built-in functions, as the machine independent code for
7083 'f128' built-in functions handles this.
7084 (rs6000_expand_builtin): Add expansion for float128 round to odd
7085 functions, keying off on -mabi=ieeelongdouble of whether to use
7086 the KFmode or TFmode variant.
7087 (rs6000_init_builtins): Initialize the _Float128 round to odd
7089 * doc/extend.texi (PowerPC Built-in Functions): Document the old
7090 _Float128 'q' built-in functions are now mapped into the new
7091 'f128' built-in functions.
7093 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
7095 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
7096 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
7098 2017-11-06 David Malcolm <dmalcolm@redhat.com>
7101 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
7102 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
7103 * toplev.c: Include "ipa-fnsummary.h".
7104 (toplev::finalize): Call ipa_fnsummary_c_finalize.
7106 2017-11-06 Jakub Jelinek <jakub@redhat.com>
7108 PR tree-optimization/82838
7109 * gimple-ssa-store-merging.c
7110 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
7111 on a separate gimple_seq which is then appended to seq.
7113 2017-11-06 Jeff Law <law@redhat.com>
7116 * config/i386/i386.c (PROBE_INTERVAL): Remove.
7117 (get_probe_interval): New functions.
7118 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
7119 (ix86_adjust_stack_and_probe): Likewise.
7120 (output_adjust_stack_and_probe): Likewise.
7121 (ix86_emit_probe_stack_range): Likewise.
7122 (ix86_expand_prologue): Likewise.
7124 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7126 PR tree-optimization/82816
7127 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
7128 if the modes of the two types are the same.
7129 (convert_plusminus_to_widen): Likewise.
7131 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7133 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
7135 (usadv16qi): New define_expand.
7136 (usadv8hi): New define_expand.
7138 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
7141 * ipa-inline-transform.c (update_noncloned_frequencies): Always
7143 (inline_transform): Likewise.
7144 * predict.c (counts_to_freqs): Remove useless conditional.
7145 * profile-count.h (profile_count::apply_scale): Move sanity check.
7146 * tree-inline.c (copy_bb): Always scale.
7147 (copy_cfg_body): Likewise.
7149 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
7152 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
7154 (*cmp_ite0): Add enabled_for_depr_it attribute.
7155 (*cmp_ite1): Likewise.
7157 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
7159 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
7162 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7164 * tree-vrp.c (vrp_int_const_binop): Return true on success and
7165 return the value by pointer.
7166 (extract_range_from_multiplicative_op_1): Update accordingly.
7167 Return as soon as an operation fails.
7169 2017-11-05 Tom de Vries <tom@codesourcery.com>
7172 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
7173 (DEF_SANITIZER_BUILTIN): ... here.
7174 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
7175 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
7177 2017-11-05 Tom de Vries <tom@codesourcery.com>
7180 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
7182 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
7183 ASM_OUTPUT_BEFORE_CASE_LABEL call.
7184 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
7186 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
7187 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
7188 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
7190 2017-11-05 Tom de Vries <tom@codesourcery.com>
7193 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
7196 2017-11-04 Michael Clark <michaeljclark@mac.com>
7198 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
7199 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
7201 (*addsi3_extended): Likewise.
7202 (*addsi3_extended2): Likewise.
7203 (<optab>si3): Likewise.
7204 (<optab>di3): Likewise.
7205 (<optab><mode>3): Likewise.
7206 (<*optabe>si3_internal): Likewise.
7207 (zero_extendqi<SUPERQI:mode>2): Likewise.
7208 (*add<mode>hi3): Likewise.
7209 (*xor<mode>hi3): Likewise.
7210 (<optab>di3): Likewise.
7211 (*<optab>si3_extend): Likewise.
7212 (*sge<u>_<X:mode><GPR:mode>): Likewise.
7213 (*slt<u>_<X:mode><GPR:mode>): Likewise.
7214 (*sle<u>_<X:mode><GPR:mode>): Likewise.
7216 2017-11-04 Andrew Waterman <andrew@sifive.com>
7218 * config/riscv/riscv.c (riscv_option_override): Conditionally set
7219 TARGET_STRICT_ALIGN based upon -mtune argument.
7221 2017-11-04 Andrew Waterman <andrew@sifive.com>
7223 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
7225 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
7227 * config/i386/i386.c (choose_basereg): Use optional scratch
7228 register and add assertion.
7229 (x86_emit_outlined_ms2sysv_save): Use scratch register when
7230 needed, and don't allocate stack.
7231 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
7232 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
7233 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
7235 2017-11-03 Jeff Law <law@redhat.com>
7237 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
7238 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
7239 to probe at the start of a noreturn function.
7241 2017-11-03 Jakub Jelinek <jakub@redhat.com>
7243 PR tree-optimization/78821
7244 * gimple-ssa-store-merging.c: Update the file comment.
7245 (MAX_STORE_ALIAS_CHECKS): Define.
7246 (struct store_operand_info): New type.
7247 (store_operand_info::store_operand_info): New constructor.
7248 (struct store_immediate_info): Add rhs_code and ops data members.
7249 (store_immediate_info::store_immediate_info): Add rhscode, op0r
7250 and op1r arguments to the ctor, initialize corresponding data members.
7251 (struct merged_store_group): Add load_align_base and load_align
7253 (merged_store_group::merged_store_group): Initialize them.
7254 (merged_store_group::do_merge): Update them.
7255 (merged_store_group::apply_stores): Pick the constant for
7256 encode_tree_to_bitpos from one of the two operands, or skip
7257 encode_tree_to_bitpos if neither operand is a constant.
7258 (class pass_store_merging): Add process_store method decl. Remove
7259 bool argument from terminate_all_aliasing_chains method decl.
7260 (pass_store_merging::terminate_all_aliasing_chains): Remove
7261 var_offset_p argument and corresponding handling.
7262 (stmts_may_clobber_ref_p): New function.
7263 (compatible_load_p): New function.
7264 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
7265 if there is overlap and rhs_code is not INTEGER_CST. For
7266 non-overlapping stores terminate group if rhs is not mergeable.
7267 (get_alias_type_for_stmts): Change first argument from
7268 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
7269 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
7270 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
7272 (get_location_for_stmts): Change first argument from
7273 auto_vec<gimple *> & to vec<gimple *> &.
7274 (struct split_store): Remove orig_stmts data member, add orig_stores.
7275 (split_store::split_store): Create orig_stores rather than orig_stmts.
7276 (find_constituent_stmts): Renamed to ...
7277 (find_constituent_stores): ... this. Change second argument from
7278 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
7279 to info structures rather than the statements.
7280 (split_group): Rename ALLOW_UNALIGNED argument to
7281 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
7282 it. Adjust find_constituent_stores caller.
7283 (imm_store_chain_info::output_merged_store): Handle rhs_code other
7284 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
7285 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
7286 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
7287 (mem_valid_for_store_merging): New function.
7288 (handled_load): New function.
7289 (pass_store_merging::process_store): New method.
7290 (pass_store_merging::execute): Use process_store method. Adjust
7291 terminate_all_aliasing_chains caller.
7293 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
7295 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
7296 Return true for more constants, symbols and label references.
7297 (aarch64_valid_floating_const): Remove unused function.
7299 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
7302 * config/aarch64/aarch64.c (aarch64_layout_frame):
7303 Undo forcing of LR at bottom of frame.
7305 2017-11-03 Jeff Law <law@redhat.com>
7308 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
7309 for int_registers_saved.
7311 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
7312 extracted from tree-ssa-dom.c.
7313 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
7314 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
7315 (record_equivalences_from_incoming_edge): Add additional argument
7316 to single_pred_edge_ignoring_loop_edges call.
7317 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
7318 (uncprop_dom_walker::before_dom_children): Add additional argument
7319 to single_pred_edge_ignoring_loop_edges call.
7320 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
7321 single_pred_edge_ignoring_loop_edges rather than open coding.
7322 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
7324 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
7326 * match.pd (-(-A)): Rewrite.
7328 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
7330 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
7331 (rs6000_emit_int_cmove): New declaration.
7332 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
7333 (rs6000_emit_sISEL): Delete.
7334 (rs6000_emit_int_cmove): Make non-static.
7335 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
7336 instead of rs6000_emit_sISEL.
7338 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
7340 * asan.c (create_cond_insert_point): Maintain profile.
7341 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
7343 * basic-block.h (struct basic_block_def): Remove frequency.
7344 (EDGE_FREQUENCY): Use to_frequency
7345 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
7347 (find_traces): Update to use to_frequency.
7348 (find_traces_1_round): Likewise; use only IPA counts.
7349 (bb_to_key): Likewise.
7350 (connect_traces): Use IPA counts only.
7351 (copy_bb_p): Update to use to_frequency.
7352 (fix_up_crossing_landing_pad): Likewise.
7353 (sanitize_hot_paths): Likewise.
7354 * bt-load.c (basic_block_freq): Likewise.
7355 * cfg.c (init_flow): Set count_max to uninitialized.
7356 (check_bb_profile): Remove frequencies; check counts.
7357 (dump_bb_info): Do not dump frequencies.
7358 (update_bb_profile_for_threading): Update counts only.
7359 (scale_bbs_frequencies_int): Likewise.
7360 (MAX_SAFE_MULTIPLIER): Remove.
7361 (scale_bbs_frequencies_gcov_type): Update counts only.
7362 (scale_bbs_frequencies_profile_count): Update counts only.
7363 (scale_bbs_frequencies): Update counts only.
7364 * cfg.h (struct control_flow_graph): Add count-max.
7365 (update_bb_profile_for_threading): Update prototype.
7366 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
7367 (find_many_sub_basic_blocks): Likewise.
7368 * cfgcleanup.c (try_forward_edges): Likewise.
7369 (try_crossjump_to_edge): Likewise.
7370 * cfgexpand.c (expand_gimple_cond): Likewise.
7371 (expand_gimple_tailcall): Likewise.
7372 (construct_init_block): Likewise.
7373 (construct_exit_block): Likewise.
7374 * cfghooks.c (verify_flow_info): Check consistency of counts.
7375 (dump_bb_for_graph): Do not dump frequencies.
7376 (split_block_1): Do not update frequencies.
7377 (split_edge): Do not update frequencies.
7378 (make_forwarder_block): Do not update frequencies.
7379 (duplicate_block): Do not update frequencies.
7380 (account_profile_record): Do not update frequencies.
7381 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
7382 for global heuristics.
7383 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
7384 (expected_loop_iterations_unbounded): Use counts only.
7385 * cfgloopmanip.c (scale_loop_profile): Simplify.
7386 (create_empty_loop_on_edge): Simplify
7388 (duplicate_loop_to_header_edge): Simplify
7389 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
7390 (update_br_prob_note): Take care of removing note when profile
7392 (relink_block_chain): Do not dump frequency.
7393 (rtl_account_profile_record): Use to_frequency.
7394 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
7395 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
7396 (cgraph_update_edges_for_call_stmt_node): Likewise.
7397 (cgraph_edge::verify_count_and_frequency): Update.
7398 (cgraph_node::verify_node): Temporarily disable frequency verification.
7399 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
7400 to_cgraph_frequency.
7401 (cgraph_edge::rebuild_edges): Convert to ipa counts.
7402 * cgraphunit.c (init_lowered_empty_function): Do not initialize
7404 (cgraph_node::expand_thunk): Update profile.
7405 * except.c (dw2_build_landing_pads): Do not update frequency.
7406 * final.c (compute_alignments): Use to_frequency.
7407 (dump_basic_block_info): Do not dump frequency.
7408 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
7409 (dump_gimple_bb_header): Do not dump frequency.
7410 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
7412 * gimple-streamer-in.c (input_bb): Do not stream frequency.
7413 * gimple-streamer-out.c (output_bb): Do not stream frequency.
7414 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
7415 (init_before_recovery): Do not update frequency.
7416 (sched_create_recovery_edges): Do not update frequency.
7417 * hsa-gen.c (convert_switch_statements): Do not update frequency.
7418 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
7419 (ipa_cp_c_finalize): Set max_count to uninitialized.
7420 * ipa-fnsummary.c (get_minimal_bb): Use counts.
7421 (param_change_prob): Use counts.
7422 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
7424 * ipa-split.c (consider_split): Use to_frequency.
7425 (split_function): Use to_frequency.
7426 * ira-build.c (loop_compare_func): Likewise.
7427 (mark_loops_for_removal): Likewise.
7428 (mark_all_loops_for_removal): Likewise.
7429 * loop-doloop.c (doloop_modify): Do not update frequency.
7430 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
7432 * lto-streamer-in.c (input_function): Update count_max.
7433 * omp-expand.c (expand_omp_taskreg): Update count_max.
7434 * omp-simd-clone.c (simd_clone_adjust): Update profile.
7435 * predict.c (maybe_hot_frequency_p): Use to_frequency.
7436 (maybe_hot_count_p): Use ipa counts only.
7437 (maybe_hot_bb_p): Simplify.
7438 (maybe_hot_edge_p): Simplify.
7439 (probably_never_executed): Do not take frequency argument.
7440 (probably_never_executed_bb_p): Do not pass frequency.
7441 (probably_never_executed_edge_p): Likewise.
7442 (combine_predictions_for_bb): Check that profile is nonzero.
7443 (propagate_freq): Do not set frequency.
7444 (drop_profile): Simplify.
7445 (counts_to_freqs): Simplify.
7446 (expensive_function_p): Use to_frequency.
7447 (propagate_unlikely_bbs_forward): Simplify.
7448 (determine_unlikely_bbs): Simplify.
7449 (estimate_bb_frequencies): Add hack to silence graphite issues.
7450 (compute_function_frequency): Use ipa counts.
7451 (pass_profile::execute): Update.
7452 (rebuild_frequencies): Use counts only.
7453 (force_edge_cold): Use counts only.
7454 * profile-count.c (profile_count::dump): Dump new count types.
7455 (profile_count::differs_from_p): Check compatiblity.
7456 (profile_count::to_frequency): New function.
7457 (profile_count::to_cgraph_frequency): New function.
7458 * profile-count.h (struct function): Declare.
7459 (enum profile_quality): Add profile_guessed_local and
7460 profile_guessed_global0.
7461 (class profile_proability): Decrease number of bits to 29;
7462 update from_reg_br_prob_note and to_reg_br_prob_note.
7463 (class profile_count: Update comment; decrease number of bits
7464 to 61. Check compatibility.
7465 (profile_count::compatible_p): New private member function.
7466 (profile_count::ipa_p): New member function.
7467 (profile_count::operator<): Handle global zero correctly.
7468 (profile_count::operator>): Handle global zero correctly.
7469 (profile_count::operator<=): Handle global zero correctly.
7470 (profile_count::operator>=): Handle global zero correctly.
7471 (profile_count::nonzero_p): New member function.
7472 (profile_count::force_nonzero): New member function.
7473 (profile_count::max): New member function.
7474 (profile_count::apply_scale): Handle IPA scalling.
7475 (profile_count::guessed_local): New member function.
7476 (profile_count::global0): New member function.
7477 (profile_count::ipa): New member function.
7478 (profile_count::to_frequency): Declare.
7479 (profile_count::to_cgraph_frequency): Declare.
7480 * profile.c (OVERLAP_BASE): Delete.
7481 (compute_frequency_overlap): Delete.
7482 (compute_branch_probabilities): Do not use compute_frequency_overlap.
7483 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
7484 * sched-ebb.c (rank): Use counts only.
7485 * shrink-wrap.c (handle_simple_exit): Use counts only.
7486 (try_shrink_wrapping): Use counts only.
7487 (place_prologue_for_one_component): Use counts only.
7488 * tracer.c (find_best_predecessor): Use to_frequency.
7489 (find_trace): Use to_frequency.
7490 (tail_duplicate): Use to_frequency.
7491 * trans-mem.c (expand_transaction): Do not update frequency.
7492 * tree-call-cdce.c: Do not update frequency.
7493 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
7494 (gimple_merge_blocks): Likewise.
7495 (gimple_split_edge): Likewise.
7496 (gimple_duplicate_sese_region): Likewise.
7497 (gimple_duplicate_sese_tail): Likewise.
7498 (move_sese_region_to_fn): Likewise.
7499 (gimple_account_profile_record): Likewise.
7500 (insert_cond_bb): Likewise.
7501 * tree-complex.c (expand_complex_div_wide): Likewise.
7502 * tree-eh.c (lower_resx): Update profile.
7503 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
7505 (initialize_cfun): Do not initialize frequencies
7506 (freqs_to_counts): Delete.
7507 (copy_cfg_body): Ignore count parameter.
7508 (copy_body): Update.
7509 (expand_call_inline): Update count_max.
7510 (optimize_inline_calls): Update count_max.
7511 (tree_function_versioning): Update count_max.
7512 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
7513 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
7515 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
7516 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
7517 (try_peel_loop): Likewise.
7518 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
7520 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
7521 (tree_transform_and_unroll_loop): Do not use frequencies
7522 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
7523 Use reliable prediction only.
7524 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
7525 * tree-ssa-sink.c (select_best_block): Use to_frequency.
7526 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
7527 probability scaling.
7528 * tree-ssa-threadupdate.c (create_block_for_threading): Do
7529 not update frequency
7530 (any_remaining_duplicated_blocks): Likewise.
7531 (update_profile): Likewise.
7532 (estimated_freqs_path): Delete.
7533 (freqs_to_counts_path): Delete.
7534 (clear_counts_path): Delete.
7535 (ssa_fix_duplicate_block_edges): Likewise.
7536 (duplicate_thread_path): Likewise.
7537 * tree-switch-conversion.c (gen_inbound_check): Use counts.
7538 * tree-tailcall.c (decrease_profile): Do not update frequency.
7539 (eliminate_tail_call): Likewise.
7540 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
7541 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
7542 (optimize_mask_stores): Likewise.
7543 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
7544 * ubsan.c (ubsan_expand_null_ifn): Update profile.
7545 (ubsan_expand_ptr_ifn): Update profile.
7546 * value-prof.c (gimple_ic): Simplify.
7547 * value-prof.h (gimple_ic): Update prototype.
7548 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
7549 * ipa-inline.c (compute_uninlined_call_time): Be sure that
7551 (want_inline_self_recursive_call_p): Likewise.
7552 (resolve_noninline_speculation): Only cummulate defined counts.
7553 (inline_small_functions): Use nonzero_p.
7554 (ipa_inline): Do not access freed node.
7556 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
7558 * config/aarch64/aarch64.c (aarch64_override_options_internal):
7559 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
7561 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
7563 * config/riscv/riscv.c (riscv_legitimize_move): Handle
7564 non-legitimate address.
7566 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
7568 * config/rs6000/rs6000.md (*lt0_disi): Delete.
7569 (*lt0_<mode>di, *lt0_<mode>si): New.
7571 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
7573 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
7574 TARGET_PAIRED_FLOAT.
7576 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
7577 Jim Wilson <jim.wilson@linaro.org>
7579 * config/aarch64/aarch64-cores.def (saphira): New CPU.
7580 * config/aarch64/aarch64-tune.md: Regenerated.
7581 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
7582 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
7584 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
7586 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
7587 (arc_expand_prologue): Restore blink for millicode.
7588 * config/arc/linux.h (LINK_EH_SPEC): Defined.
7590 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
7593 * config/i386/i386.c (ix86_vector_duplicate_value): Use
7594 gen_vec_duplicate after forcing the scalar into a register.
7596 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
7598 * combine (try_combine): Print the insns input to try_combine to the
7601 2017-11-02 Steve Ellcey <sellcey@cavium.com>
7604 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
7605 Remove second argument from aarch64_process_target_attr call.
7606 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
7608 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
7610 (aarch64_handle_attr_arch): Remove second argument.
7611 (aarch64_handle_attr_cpu): Ditto.
7612 (aarch64_handle_attr_tune): Ditto.
7613 (aarch64_handle_attr_isa_flags): Ditto.
7614 (aarch64_process_one_target_attr): Ditto.
7615 (aarch64_process_target_attr): Ditto.
7616 (aarch64_option_valid_attribute_p): Remove second argument.
7617 on aarch64_process_target_attr call.
7619 2017-11-02 David Malcolm <dmalcolm@redhat.com>
7621 * diagnostic.c: Include "selftest-diagnostic.h".
7622 (selftest::assert_location_text): New function.
7623 (selftest::test_diagnostic_get_location_text): New function.
7624 (selftest::diagnostic_c_tests): Call it.
7626 2017-11-02 David Malcolm <dmalcolm@redhat.com>
7628 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
7629 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
7630 (class selftest::test_diagnostic_context): Move to...
7631 * selftest-diagnostic.c: New file.
7632 * selftest-diagnostic.h: New file.
7634 2017-11-02 James Bowman <james.bowman@ftdichip.com>
7636 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
7637 offset range for FT32B.
7638 * config/ft32/ft32.h: option "mcompress" enables relaxation.
7639 * config/ft32/ft32.md: Add TARGET_NOPM.
7640 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
7641 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
7643 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
7645 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
7647 2017-11-02 Jeff Law <law@redhat.com>
7649 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
7650 virtual keyword on FINAL OVERRIDE members.
7652 * tree-ssa-propagate.h (ssa_propagation_engine): Group
7653 virtuals together. Add virtual destructor.
7654 (substitute_and_fold_engine): Similarly.
7656 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
7658 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
7660 2017-11-02 Richard Biener <rguenther@suse.de>
7662 PR tree-optimization/82795
7663 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
7665 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7667 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
7669 (gcc_GAS_CHECK_FEATURE): Remove.
7670 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
7671 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
7672 * configure: Regenerate.
7674 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
7676 * config/arc/arc.c (hwloop_optimize): Account for empty
7679 2017-11-02 Richard Biener <rguenther@suse.de>
7682 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
7683 Truncate ARRAY_REF index and element size.
7685 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
7687 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
7689 2017-11-01 Jeff Law <law@redhat.com>
7691 * tree-ssa-ccp.c (ccp_folder): New class derived from
7692 substitute_and_fold_engine.
7693 (ccp_folder::get_value): New member function.
7694 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
7695 (ccp_fold_stmt): Remove prototype.
7696 (ccp_finalize): Call substitute_and_fold from the ccp_class.
7697 * tree-ssa-copy.c (copy_folder): New class derived from
7698 substitute_and_fold_engine.
7699 (copy_folder::get_value): Renamed from get_value.
7700 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
7701 * tree-vrp.c (vrp_folder): New class derived from
7702 substitute_and_fold_engine.
7703 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
7704 (vrp_folder::get_value): New member function.
7705 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
7706 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
7707 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
7708 provide a class interface to folder/substitute routines.
7709 (ssa_prop_fold_stmt_fn): Remove typedef.
7710 (ssa_prop_get_value_fn): Likewise.
7711 (subsitute_and_fold): Remove prototype.
7712 (replace_uses_in): Likewise.
7713 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
7714 Renamed from replace_uses_in. Call the virtual member function
7715 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
7716 (substitute_and_fold_dom_walker): Remove initialization of
7717 data member entries for calbacks. Add substitute_and_fold_engine
7718 member and initialize it.
7719 (substitute_and_fold_dom_walker::before_dom_children0: Use the
7720 member functions for get_value, replace_phi_args_in c
7721 replace_uses_in, and fold_stmt calls.
7722 (substitute_and_fold_engine::substitute_and_fold): Renamed from
7723 substitute_and_fold. Remove assert. Update ctor call.
7725 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
7726 (ssa_prop_visit_phi_fn): Likewise.
7727 (class ssa_propagation_engine): New class to provide an interface
7729 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
7731 (ssa_prop_visit_phi): Likewise.
7732 (ssa_propagation_engine::simulate_stmt): Moved into class.
7733 Call visit_phi/visit_stmt from the class rather than via
7734 file scoped static variables.
7735 (ssa_propagation_engine::simulate_block): Moved into class.
7736 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
7737 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
7738 set file scoped statics for the visit_stmt/visit_phi callbacks.
7739 * tree-complex.c (complex_propagate): New class derived from
7740 ssa_propagation_engine.
7741 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
7742 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
7743 (tree_lower_complex): Call ssa_propagate via the complex_propagate
7745 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
7746 ssa_propagation_engine.
7747 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
7748 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
7749 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
7750 * tree-ssa-copy.c (copy_prop): New class derived from
7751 ssa_propagation_engine.
7752 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
7753 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
7754 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
7755 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
7756 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
7757 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
7758 (execute_vrp): Call ssa_propagate from the vrp_prop class.
7760 2017-11-01 Jakub Jelinek <jakub@redhat.com>
7762 PR rtl-optimization/82778
7763 PR rtl-optimization/82597
7764 * compare-elim.c (struct comparison): Add in_a_setter field.
7765 (find_comparison_dom_walker::before_dom_children): Remove killed
7766 bitmap and df_simulate_find_defs call, instead walk the defs.
7767 Compute last_setter and initialize in_a_setter. Merge definitions
7768 with first initialization for a few variables.
7769 (try_validate_parallel): Use insn_invalid_p instead of
7770 recog_memoized. Return insn rather than just the pattern.
7771 (try_merge_compare): Fix up comment. Don't uselessly test if
7772 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
7774 (execute_compare_elim_after_reload): Remove df_chain_add_problem
7777 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7778 Alan Hayward <alan.hayward@arm.com>
7779 David Sherwood <david.sherwood@arm.com>
7781 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
7782 aarch64_hard_regno_nregs to get the number of registers
7785 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7786 Alan Hayward <alan.hayward@arm.com>
7787 David Sherwood <david.sherwood@arm.com>
7789 * config/aarch64/constraints.md (Upl): Rename to...
7791 * config/aarch64/aarch64.md
7792 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
7795 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7796 Alan Hayward <alan.hayward@arm.com>
7797 David Sherwood <david.sherwood@arm.com>
7799 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
7800 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
7803 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7804 Alan Hayward <alan.hayward@arm.com>
7805 David Sherwood <david.sherwood@arm.com>
7807 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
7808 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
7809 (aarch64_evpc_dup): Generate rtl direcly, rather than using
7811 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
7812 of a single element.
7813 * config/aarch64/iterators.md: Add a comment above the permute
7814 unspecs to say that they are generated directly by
7815 aarch64_expand_vec_perm_const.
7816 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
7818 2017-11-01 Nathan Sidwell <nathan@acm.org>
7820 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
7822 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
7824 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
7825 and medany code models, and describe what they do.
7827 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7829 Revert accidental duplicate:
7831 * combine.c (can_change_dest_mode): Reject changes in
7832 REGMODE_NATURAL_SIZE.
7834 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
7836 PR rtl-optimization/64682
7837 PR rtl-optimization/69567
7838 PR rtl-optimization/69737
7839 PR rtl-optimization/82683
7840 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
7841 register mentioned in the note, drop the note, unless it came from I3,
7842 in which case it should go to I3 again.
7844 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7846 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
7847 and return false if not.
7848 (clear_bytes_written_by, live_bytes_read): Update accordingly.
7850 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7852 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
7853 range is known to be empty.
7855 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7856 Alan Hayward <alan.hayward@arm.com>
7857 David Sherwood <david.sherwood@arm.com>
7859 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
7860 and CONST_VECTOR_NUNITS instead of computing the number of units from
7861 the byte sizes of the vector and element.
7862 (simplify_binary_operation_1): Likewise.
7863 (simplify_const_binary_operation): Likewise.
7864 (simplify_ternary_operation): Likewise.
7866 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7867 Alan Hayward <alan.hayward@arm.com>
7868 David Sherwood <david.sherwood@arm.com>
7870 * var-tracking.c (INT_MEM_OFFSET): Replace with...
7871 (int_mem_offset): ...this new function.
7872 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
7873 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
7874 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
7877 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7878 Alan Hayward <alan.hayward@arm.com>
7879 David Sherwood <david.sherwood@arm.com>
7881 * lower-subreg.c (interesting_mode_p): New function.
7882 (compute_costs, find_decomposable_subregs, decompose_register)
7883 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
7884 (resolve_clobber, dump_choices): Use it.
7886 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7887 Alan Hayward <alan.hayward@arm.com>
7888 David Sherwood <david.sherwood@arm.com>
7890 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
7892 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7893 Alan Hayward <alan.hayward@arm.com>
7894 David Sherwood <david.sherwood@arm.com>
7896 * alias.c (find_base_value, find_base_term): Only process integer
7897 truncations. Check the precision rather than the size.
7899 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7900 Alan Hayward <alan.hayward@arm.com>
7901 David Sherwood <david.sherwood@arm.com>
7903 * machmode.h (is_narrower_int_mode): New function
7904 * optabs.c (expand_float, expand_fix): Use it.
7905 * dwarf2out.c (rotate_loc_descriptor): Likewise.
7907 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7908 Alan Hayward <alan.hayward@arm.com>
7909 David Sherwood <david.sherwood@arm.com>
7911 * rtl.h (narrower_subreg_mode): New function.
7912 * ira-color.c (update_costs_from_allocno): Use it.
7914 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7915 Alan Hayward <alan.hayward@arm.com>
7916 David Sherwood <david.sherwood@arm.com>
7918 * optabs-query.h (convert_optab_p): New function, split out from...
7919 (convert_optab_handler): ...here.
7920 (widening_optab_handler): Delete.
7921 (find_widening_optab_handler): Remove permit_non_widening parameter.
7922 (find_widening_optab_handler_and_mode): Likewise. Provide an
7923 override that operates on mode class wrappers.
7924 * optabs-query.c (widening_optab_handler): Delete.
7925 (find_widening_optab_handler_and_mode): Remove permit_non_widening
7926 parameter. Assert that the two modes are the same class and that
7927 the "from" mode is narrower than the "to" mode. Use
7928 convert_optab_handler instead of widening_optab_handler.
7929 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
7930 instead of widening_optab_handler.
7931 * expr.c (expand_expr_real_2): Update calls to
7932 find_widening_optab_handler.
7933 * optabs.c (expand_widen_pattern_expr): Likewise.
7934 (expand_binop_directly): Take the insn_code as a parameter.
7935 (expand_binop): Only call find_widening_optab_handler for
7936 conversion optabs; use optab_handler otherwise. Update calls
7937 to find_widening_optab_handler and expand_binop_directly.
7938 Use convert_optab_handler instead of widening_optab_handler.
7939 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
7940 find_widening_optab_handler and use scalar_mode rather than
7942 (convert_plusminus_to_widen): Likewise.
7944 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7945 Alan Hayward <alan.hayward@arm.com>
7946 David Sherwood <david.sherwood@arm.com>
7948 * machmode.h (fixed_size_mode): New class.
7949 * rtl.h (get_pool_mode): Return fixed_size_mode.
7950 * gengtype.c (main): Add fixed_size_mode.
7951 * target.def (get_raw_result_mode): Return a fixed_size_mode.
7952 (get_raw_arg_mode): Likewise.
7953 * doc/tm.texi: Regenerate.
7954 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
7955 * targhooks.c (default_get_reg_raw_mode): Likewise.
7956 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
7957 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
7958 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
7959 (msp430_get_raw_result_mode): Likewise.
7960 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
7961 * dbxout.c (dbxout_parms): Require fixed-size modes.
7962 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
7963 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
7964 * omp-low.c (lower_oacc_reductions): Likewise.
7965 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
7966 (simplify_subreg): Update accordingly.
7967 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
7968 (force_const_mem): Update accordingly. Return NULL_RTX for modes
7969 that aren't fixed-size.
7970 (get_pool_mode): Return a fixed_size_mode.
7971 (output_constant_pool_2): Take a fixed_size_mode.
7973 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7974 Alan Hayward <alan.hayward@arm.com>
7975 David Sherwood <david.sherwood@arm.com>
7977 * doc/rtl.texi (vec_series): Document.
7978 (const): Say that the operand can be a vec_series.
7979 * rtl.def (VEC_SERIES): New rtx code.
7980 * rtl.h (const_vec_series_p_1): Declare.
7981 (const_vec_series_p): New function.
7982 * emit-rtl.h (gen_const_vec_series): Declare.
7983 (gen_vec_series): Likewise.
7984 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
7985 (gen_vec_series): Likewise.
7986 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
7987 * simplify-rtx.c (simplify_unary_operation): Handle negations
7989 (simplify_binary_operation_series): New function.
7990 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
7991 (test_vector_ops_series): New function.
7992 (test_vector_ops): Call it.
7993 * config/powerpcspe/altivec.md (altivec_lvsl): Use
7994 gen_const_vec_series.
7995 (altivec_lvsr): Likewise.
7996 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
7998 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7999 Alan Hayward <alan.hayward@arm.com>
8000 David Sherwood <david.sherwood@arm.com>
8002 * doc/rtl.texi (const): Update description of address constants.
8003 Say that vector constants are allowed too.
8004 * common.md (E, F): Use CONSTANT_P instead of checking for
8006 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
8007 checking for CONST_VECTOR.
8008 * expmed.c (make_tree): Use build_vector_from_val for a CONST
8010 * expr.c (expand_expr_real_2): Check for vector modes instead
8011 of checking for CONST_VECTOR.
8012 * rtl.h (const_vec_p): New function.
8013 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
8014 (unwrap_const_vec_duplicate): Handle them here too.
8016 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8017 David Malcolm <dmalcolm@redhat.com>
8018 Alan Hayward <alan.hayward@arm.com>
8019 David Sherwood <david.sherwood@arm.com>
8021 * rtl.h (vec_duplicate_p): New function.
8022 * selftest-rtl.c (assert_rtx_eq_at): New function.
8023 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
8024 (assert_rtx_eq_at): Declare.
8025 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
8026 * selftest-run-tests.c (selftest::run_tests): Call it.
8027 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
8028 (simplify_unary_operation_1): Recursively handle vector duplicates.
8029 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
8031 (simplify_subreg): Handle subregs of vector duplicates.
8032 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
8033 (selftest::simplify_rtx_c_tests): New functions.
8035 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8036 Alan Hayward <alan.hayward@arm.com>
8037 David Sherwood <david.sherwood@arm.com>
8039 * emit-rtl.h (gen_const_vec_duplicate): Declare.
8040 (gen_vec_duplicate): Likewise.
8041 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
8043 (gen_const_vector): ...here.
8044 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
8045 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
8046 whose elements are all equal.
8047 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
8048 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
8049 (simplify_relational_operation): Likewise.
8050 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
8052 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
8053 (aarch64_expand_vector_init): Likewise.
8054 * config/arm/arm.c (neon_vdup_constant): Likewise.
8055 (neon_expand_vector_init): Likewise.
8056 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
8057 (arm_block_set_unaligned_vect): Likewise.
8058 (arm_block_set_aligned_vect): Likewise.
8059 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
8060 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
8061 (expand_vec_perm_even_odd_pack): Likewise.
8062 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
8063 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
8064 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
8065 gen_const_vec_duplicate.
8066 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
8067 * config/mips/mips.c (mips_gen_const_int_vector): Use
8068 gen_const_vec_duplicate.
8069 (mips_expand_vector_init): Use CONST0_RTX.
8070 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
8071 (define_split): Use gen_const_vec_duplicate.
8072 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
8073 (define_split): Use gen_const_vec_duplicate.
8074 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
8075 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
8076 * config/spu/spu.c (spu_const): Likewise.
8078 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8079 Alan Hayward <alan.hayward@arm.com>
8080 David Sherwood <david.sherwood@arm.com>
8082 * combine.c (can_change_dest_mode): Reject changes in
8083 REGMODE_NATURAL_SIZE.
8085 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
8087 * configure.ac (--enable-libssp): New.
8088 (gcc_cv_libc_provides_ssp): Check for explicit setting before
8089 trying to determine target-specific default. Adjust indentation.
8090 * configure: Regenerated.
8091 * doc/install.texi (Configuration): Expand --disable-libssp
8094 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
8096 config/i386/i386.c (ix86_expand_epilogue): Correct stack
8099 2017-10-31 Martin Jambor <mjambor@suse.cz>
8102 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
8104 2017-10-31 David Malcolm <dmalcolm@redhat.com>
8106 * auto-profile.c (autofdo_source_profile::read): Use
8107 UNKNOWN_LOCATION rather than 0.
8108 * diagnostic-core.h (warning_at_rich_loc): Rename to...
8109 (warning_at): ...this overload.
8110 (warning_at_rich_loc_n): Rename to...
8111 (warning_n): ...this overload.
8112 (error_at_rich_loc): Rename to...
8113 (error_at): ...this overload.
8114 (pedwarn_at_rich_loc): Rename to...
8115 (pedwarn): ...this overload.
8116 (permerror_at_rich_loc): Rename to...
8117 (permerror): ...this overload.
8118 (inform_at_rich_loc): Rename to...
8119 (inform): ...this overload.
8120 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
8121 (diagnostic_n_impl_richloc): Rename to...
8122 (diagnostic_n_impl): ...this rich_location *-based decl.
8123 (inform_at_rich_loc): Rename to...
8124 (inform): ...this, and add an assertion.
8125 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
8126 (warning_at_rich_loc): Rename to...
8127 (warning_at): ...this, and add an assertion.
8128 (warning_at_rich_loc_n): Rename to...
8129 (warning_n): ...this, and add an assertion.
8130 (warning_n): Update location_t-based implementation for removal of
8131 location_t-based diagnostic_n_impl.
8132 (pedwarn_at_rich_loc): Rename to...
8133 (pedwarn): ...this, and add an assertion.
8134 (permerror_at_rich_loc): Rename to...
8135 (permerror): ...this, and add an assertion.
8136 (error_n): Update for removal of location_t-based diagnostic_n_impl.
8137 (error_at_rich_loc): Rename to...
8138 (error_at): ...this, and add an assertion.
8139 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
8140 (driver::do_spec_on_infiles): Likewise.
8141 * substring-locations.c (format_warning_va): Update for renaming
8142 of inform_at_rich_loc.
8144 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
8146 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
8147 _Float<N>X built-in functions so that the variant without the
8148 "__builtin_" prefix is only enabled for the GNU C and Objective C
8149 languages when they are in non-strict ANSI/ISO mode.
8150 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
8151 * target.def (floatn_builtin_p): Add a target hook to control
8152 whether _Float<N> and _Float<N>X built-in functions without the
8153 "__builtin_" prefix are enabled, and return true for C and
8154 Objective C in the default hook. Include langhooks.h in
8156 * targhooks.h (default_floatn_builtin_p): Likewise.
8157 * targhooks.c (default_floatn_builtin_p): Likewise.
8158 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
8159 floatn_builtin_p target hook.
8160 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
8162 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
8163 Eric Botcazou <ebotcazou@adacore.com>
8165 PR rtl-optimization/81803
8166 * lra-constraints.c (curr_insn_transform): Also reload the whole
8167 register for a strict subreg no wider than a word if this is for
8168 a WORD_REGISTER_OPERATIONS target.
8170 2017-10-31 Jason Merrill <jason@redhat.com>
8172 * gdbinit.in: Skip over inlines from timevar.h.
8174 2017-10-31 Martin Liska <mliska@suse.cz>
8176 * doc/gcov.texi: Document new option.
8177 * gcov.c (print_usage): Likewise print it.
8178 (process_args): Support the argument.
8179 (format_count): New function.
8180 (format_gcov): Use the function.
8182 2017-10-31 Martin Liska <mliska@suse.cz>
8184 * gcov.c (struct name_map): do not use typedef.
8185 Define operator== and operator<.
8186 (name_search): Remove.
8187 (name_sort): Remove.
8188 (main): Do not allocate names.
8189 (process_file): Add vertical space.
8190 (generate_results): Use std::find.
8191 (release_structures): Do not release memory.
8192 (find_source): Use std::find.
8194 2017-10-31 Martin Liska <mliska@suse.cz>
8196 * gcov.c (struct line_info): Remove it's typedef.
8197 (line_info::line_info): Add proper ctor.
8198 (line_info::has_block): Do not use a typedef.
8199 (struct source_info): Do not use typedef.
8200 (circuit): Likewise.
8201 (get_cycles_count): Likewise.
8202 (output_intermediate_file): Iterate via vector iterator.
8203 (add_line_counts): Use std::vector methods.
8204 (accumulate_line_counts): Likewise.
8205 (output_lines): Likewise.
8207 2017-10-31 Martin Liska <mliska@suse.cz>
8209 * gcov.c (struct source_info): Remove typedef.
8210 (source_info::source_info): Add proper ctor.
8211 (accumulate_line_counts): Use struct, not it's typedef.
8212 (output_gcov_file): Likewise.
8213 (output_lines): Likewise.
8214 (main): Do not allocate an array.
8215 (output_intermediate_file): Use size of vector container.
8216 (process_file): Resize the vector.
8217 (generate_results): Do not preallocate, use newly added vector
8219 (release_structures): Do not release sources.
8220 (find_source): Use vector methods.
8221 (add_line_counts): Do not use typedef.
8223 2017-10-31 Martin Liska <mliska@suse.cz>
8225 * doc/gcov.texi: Document that.
8226 * gcov.c (add_line_counts): Mark lines with a non-executed
8228 (output_line_beginning): Handle such lines.
8229 (output_lines): Pass new argument.
8230 (output_intermediate_file): Print it in intermediate format.
8232 2017-10-31 Martin Liska <mliska@suse.cz>
8234 * color-macros.h: New file.
8235 * diagnostic-color.c: Factor out color related to macros to
8237 * doc/gcov.texi: Document -k option.
8238 * gcov.c (INCLUDE_STRING): Include string.h.
8239 (print_usage): Add -k option.
8240 (process_args): Parse it.
8241 (pad_count_string): New function.
8242 (output_line_beginning): Likewise.
8243 (DEFAULT_LINE_START): New macro.
8244 (output_lines): Support color output.
8246 2017-10-31 Martin Liska <mliska@suse.cz>
8248 PR gcov-profile/82633
8249 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
8250 their interaction with GCOV infrastructure.
8251 * configure.ac: Add -fkeep-{inline,static}-functions to
8253 * configure: Regenerate.
8255 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
8258 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
8260 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
8263 * config/rs6000/rs6000.md (allocate_stack): Force update interval
8264 into a register if it does not fit into an immediate offset field.
8266 2017-10-31 Olivier Hainque <hainque@adacore.com>
8268 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
8270 2017-10-31 Julia Koval <julia.koval@intel.com>
8272 * config.gcc: Add gfniintrin.h.
8273 * config/i386/gfniintrin.h: New.
8274 * config/i386/i386-builtin-types.def
8275 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
8276 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
8277 __builtin_ia32_vgf2p8affineinvqb_v32qi,
8278 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
8279 __builtin_ia32_vgf2p8affineinvqb_v16qi,
8280 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
8281 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
8282 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
8283 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
8284 V64QI_FTYPE_V64QI_V64QI_INT): New types.
8285 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
8286 * config/i386/immintrin.h: Include gfniintrin.h.
8287 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
8289 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
8291 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
8293 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
8295 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
8296 (arm_ashldi3_1bit): Remove pattern.
8297 (ashrdi3): Remove shift by 1 expansion.
8298 (arm_ashrdi3_1bit): Remove pattern.
8299 (lshrdi3): Remove shift by 1 expansion.
8300 (arm_lshrdi3_1bit): Remove pattern.
8301 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
8302 cost of ashldi3 by 1.
8303 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
8304 (<shift>di3_neon): Likewise.
8306 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
8308 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
8309 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
8310 and (*aarch64_simd_mov<VQ:mode>).
8311 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
8312 pattern alternative.
8313 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
8314 attributes to match pattern alternative.
8316 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
8318 * config.gcc (powerpc*-*-*): Add emmintrin.h.
8319 * config/rs6000/emmintrin.h: New file.
8320 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
8322 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
8324 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
8325 * (movdi_vfp_cortexa8): Remove pattern.
8327 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8329 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
8331 (Specific, alpha*-dec-osf5.1): Remove.
8332 (Specific, mips-sgi-irix5): Remove.
8333 (Specific, mips-sgi-irix6): Remove.
8335 2017-10-30 Jakub Jelinek <jakub@redhat.com>
8338 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
8339 arguments to clear_bit_region_be.
8341 2017-10-30 Jim Wilson <wilson@tuliptree.org>
8343 * gimplify.c: Include memmodel.h.
8345 2017-10-30 Martin Jambor <mjambor@suse.cz>
8347 * omp-grid.c (grid_attempt_target_gridification): Also insert a
8348 condition whether loop should be executed at all.
8350 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
8352 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
8353 gimple folding of vec_madd() intrinsics.
8354 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
8355 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
8356 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
8358 2017-10-30 Richard Biener <rguenther@suse.de>
8360 PR tree-optimization/82762
8362 2017-10-23 Richard Biener <rguenther@suse.de>
8364 PR tree-optimization/82129
8366 2017-08-01 Richard Biener <rguenther@suse.de>
8368 PR tree-optimization/81181
8369 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
8370 (compute_antic): ... end of iteration here.
8372 2017-10-30 Joseph Myers <joseph@codesourcery.com>
8374 * doc/invoke.texi (C Dialect Options): Document -std=c17,
8375 -std=iso9899:2017 and -std=gnu17.
8376 * doc/standards.texi (C Language): Document C17 support.
8377 * doc/cpp.texi (Overview): Mention -std=c17.
8378 (Standard Predefined Macros): Document C11 and C17 values of
8379 __STDC_VERSION__. Do not refer to C99 support as incomplete.
8380 * doc/extend.texi (Inline): Do not list individual options for
8381 standards newer than C99.
8382 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
8384 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
8387 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
8389 * asan.c (asan_finish_file): Align asan globals array by shadow
8392 2017-10-30 Jakub Jelinek <jakub@redhat.com>
8395 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
8396 (struct store_immediate_info): Add bitregion_start and bitregion_end
8398 (store_immediate_info::store_immediate_info): Add brs and bre
8399 arguments and initialize bitregion_{start,end} from those.
8400 (struct merged_store_group): Add bitregion_start, bitregion_end,
8401 align_base and mask fields. Drop unnecessary struct keyword from
8402 struct store_immediate_info. Add do_merge method.
8403 (clear_bit_region_be): Use memset instead of loop storing zeros.
8404 (merged_store_group::do_merge): New method.
8405 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
8406 stores as long as the surrounding bitregions have no gaps.
8407 (merged_store_group::merge_overlapping): Use do_merge.
8408 (merged_store_group::apply_stores): Test that bitregion_{start,end}
8409 is byte aligned, rather than requiring that start and width are
8410 byte aligned. Drop unnecessary struct keyword from
8411 struct store_immediate_info. Allocate and populate also mask array.
8412 Make start of the arrays relative to bitregion_start rather than
8413 start and size them according to bitregion_{end,start} difference.
8414 (struct imm_store_chain_info): Drop unnecessary struct keyword from
8415 struct store_immediate_info.
8416 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
8417 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
8418 struct keyword from struct store_immediate_info.
8419 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
8420 between stores as long as the surrounding bitregions have no gaps.
8422 (struct split_store): Add orig non-static data member.
8423 (split_store::split_store): Initialize orig to false.
8424 (find_constituent_stmts): Return store_immediate_info *, non-NULL
8425 if there is exactly a single original stmt. Change stmts argument
8426 to pointer from reference, if NULL, don't push anything to it. Add
8427 first argument, use it to optimize skipping over orig stmts that
8428 are known to be before bitpos already. Simplify.
8429 (split_group): Return unsigned int count how many stores are or
8430 would be needed rather than a bool. Add allow_unaligned argument.
8431 Change split_stores argument from reference to pointer, if NULL,
8432 only do a dry run computing how many stores would be produced.
8433 Rewritten algorithm to use both alignment and misalign if
8434 !allow_unaligned and handle bitfield stores with gaps.
8435 (imm_store_chain_info::output_merged_store): Set start_byte_pos
8436 from bitregion_start instead of start. Compute allow_unaligned
8437 here, if true, do 2 split_group dry runs to compute which one
8438 produces fewer stores and prefer aligned if equal. Punt if
8439 new count is bigger or equal than original before emitting any
8440 statements, rather than during that. Remove no longer needed
8441 new_ssa_names tracking. Replace num_stmts with
8442 split_stores.length (). Use 32-bit stack allocated entries
8443 in split_stores auto_vec. Try to reuse original store lhs/rhs1
8444 if possible. Handle bitfields with gaps.
8445 (pass_store_merging::execute): Ignore bitsize == 0 stores.
8446 Compute bitregion_{start,end} for the stores and construct
8447 store_immediate_info with that. Formatting fixes.
8449 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
8452 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
8453 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
8455 2017-10-29 Jim Wilson <wilson@tuliptree.org>
8457 * gimplify.c: Include tm_p.h.
8459 * common.opt (gcoff): Re-add as ignored option.
8460 (gcoff1, gcoff2, gcoff3): Likewise.
8462 * Makefile.in (OBJS): Delete sdbout.o.
8463 (GTFILES): Delete $(srcdir)/sdbout.c.
8464 * debug.h: Delete sdb_debug_hooks.
8465 * final.c: Delete sdbout.h include.
8466 (final_scan_insn): Delete SDB_DEBUG check.
8467 (rest_of_clean_state): Likewise.
8468 * output.h: Delete sdb_begin_function_line.
8471 * toplev.c: Delete sdbout.h include.
8472 (process_options): Delete SDB_DEBUG check.
8473 * tree-core.h (tree_type_common): Delete pointer field of
8475 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
8476 TYPE_SYMTAB_POINTER.
8477 * tree.h (TYPE_SYMTAB_POINTER): Delete.
8478 (TYPE_SYMTAB_IS_POINTER): Delete.
8479 (TYPE_SYMTAB_IS_DIE): Renumber.
8480 * xcoffout.c: Refer to former sdbout.c file.
8481 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
8483 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
8484 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
8485 (Debugging Options): Delete -gcoff.
8486 (-gstabs): Delete SDB reference.
8488 (-gcoff@var{level}): Delete.
8489 * doc/passes.texi (Debugging information output): Delete SDB and
8490 sdbout.c references.
8491 * doc/tm.texi: Regenerate.
8492 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
8493 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
8495 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
8496 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
8497 (SDB_DEBUGGING_INFO): Delete.
8498 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
8499 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
8500 * target.def (output_source_filename): Delete COFF reference.
8502 * common.opt (gcoff): Delete.
8503 (gxcoff+): Update Negative chain.
8504 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
8506 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
8507 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
8508 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
8510 (expand_function_start): Change sdb reference to past tense.
8511 (expand_function_end): Change sdb reference to past tense.
8512 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
8513 * opts.c (debug_type_names): Delete coff entry.
8514 (common_handle_option): Delete OPT_gcoff case.
8515 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
8517 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
8518 * config/cris/cris.h: Delete SDB reference in comment.
8519 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
8520 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
8521 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
8522 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
8524 (ix86_expand_prologue): Likewise.
8525 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
8526 * config/ia64/ia64.h: Likewise.
8527 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
8528 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
8530 * config/mmix/mmix.h: Likewise.
8531 * config/nds32/nds32.c: Likewise.
8532 * config/stormy/storym16.h: Likewise.
8533 * config/visium/visium.h: Likewise.
8534 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
8536 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
8538 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
8539 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
8540 FRAME_POINTER_REGNUM point at high end of local var area.
8542 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
8544 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
8545 Move comment around. Do not reset best_edge for a copiable
8546 destination if the copy would cause a partition change.
8547 (better_edge_p): Remove redundant check.
8549 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
8551 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
8553 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
8555 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
8556 for math functions that have _Float<N> and _Float<N>X variants.
8557 (mathfn_built_in_2): Add support for math functions that have
8558 _Float<N> and _Float<N>X variants.
8559 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
8560 (expand_builtin_mathfn_ternary): Add support for fma with
8561 _Float<N> and _Float<N>X variants.
8562 (expand_builtin): Likewise.
8563 (fold_builtin_3): Likewise.
8564 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
8565 create math function _Float<N> and _Float<N>X variants as external
8567 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
8568 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
8569 the __builtin_ prefix and if not strict ansi, without the prefix.
8570 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
8571 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
8572 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
8573 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
8574 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
8575 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
8576 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
8577 function signatures for fma _Float<N> and _Float<N>X variants.
8578 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
8579 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
8580 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
8581 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
8582 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
8583 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
8584 * gencfn-macros.c (print_case_cfn): Add support for math functions
8585 that have _Float<N> and _Float<N>X variants.
8586 (print_define_operator_list): Likewise.
8587 (fltfn_suffixes): Likewise.
8589 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
8590 for math functions that have _Float<N> and _Float<N>X variants.
8591 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
8592 and _Float<N>X variants.
8593 (COPYSIGN): Likewise.
8596 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
8597 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
8599 (integer_valued_read_call_p): Likewise.
8600 * fold-const-call.c (fold_const_call_ss): Likewise.
8601 (fold_const_call_sss): Add support for copysign, fmin, and fmax
8602 _Float<N> and _Float<N>X variants.
8603 (fold_const_call_ssss): Add support for fma _Float<N> and
8604 _Float<N>X variants.
8605 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
8606 support for copysign and fma _Float<N> and _Float<N>X variants.
8607 (backprop::process_builtin_call_use): Likewise.
8608 * tree-call-cdce.c (can_test_argument_range); Add support for
8609 sqrt _Float<N> and _Float<N>X variants.
8610 (edom_only_function): Likewise.
8611 (get_no_error_domain): Likewise.
8612 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
8613 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
8614 copysign _Float<N> and _Float<N>X variants.
8615 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
8616 handled by machine independent code.
8617 (FMAF128): Likewise.
8618 * doc/cpp.texi (Common Predefined Macros): Document defining
8619 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
8620 fma _Float<N> and _Float<N>X variants.
8622 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
8625 * config/i386/i386-modes.def (CCFPU): Remove definition.
8626 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
8627 (ix86_cc_modes_compatible): Ditto.
8628 (ix86_expand_carry_flag_compare): Ditto.
8629 (ix86_expand_int_movcc): Ditto.
8630 (ix86_expand_int_addcc): Ditto.
8631 (ix86_reverse_condition): Ditto.
8632 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
8633 Return true/false for unordered/ordered fp comparisons.
8634 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
8635 (ix86_prepare_fp_compare_args): Update for rename.
8636 (ix86_expand_fp_compare): Update for rename. Generate unordered
8637 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
8638 (ix86_expand_sse_compare_and_jump): Ditto.
8639 * config/i386/predicates.md (fcmov_comparison_operator):
8640 Remove CCFPU mode handling.
8641 (ix86_comparison_operator): Ditto.
8642 (ix86_carry_flag_operator): Ditto.
8643 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
8644 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
8645 (*cmpu<mode>_cc_i387): Ditto.
8646 (FPCMP): Remove mode iterator.
8647 (unord): Remove mode attribute.
8648 (unord_subst): New define_subst transformation
8649 (unord): New define_subst attribute.
8651 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
8652 (*cmpi<unord>xf_i387): Ditto.
8653 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
8654 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
8655 using unord_subst transformation.
8656 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
8657 (round_saeonly): Also handle CCFP mode.
8658 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
8659 Remove UNSPEC_SAHF unspec handling.
8661 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
8663 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
8665 2017-10-27 Jeff Law <law@redhat.com>
8667 * gimple-ssa-sprintf.c: Include domwalk.h.
8668 (class sprintf_dom_walker): New class, derived from dom_walker.
8669 (sprintf_dom_walker::before_dom_children): New function.
8670 (struct call_info): Moved into sprintf_dom_walker class
8671 (compute_formath_length, handle_gimple_call): Likewise.
8672 (sprintf_length::execute): Call the dominator walker rather
8673 than walking the statements.
8675 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
8676 gimple statement locations.
8677 (check_array_bounds): Corresponding changes. Get the statement's
8678 location directly from wi->stmt.
8680 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
8683 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
8685 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
8687 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
8688 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
8690 2017-10-27 Jakub Jelinek <jakub@redhat.com>
8693 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
8694 * config/i386/i386.c (maybe_get_pool_constant): Removed.
8695 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
8696 maybe_get_pool_constant.
8697 * config/i386/predicates.md (zero_extended_scalar_load_operand):
8700 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8702 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
8703 2.26 caveat. Update gas and gld versions.
8704 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
8707 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
8709 * cgraph.h (set_malloc_flag): Declare.
8710 * cgraph.c (set_malloc_flag_1): New function.
8711 (set_malloc_flag): Likewise.
8712 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
8713 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
8715 (read_ipa_call_summary): Add support for reading is_return_callee.
8716 (write_ipa_call_summary): Stream is_return_callee.
8717 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
8718 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
8719 ipa-prop.h, ipa-fnsummary.h.
8720 (pure_const_names): Change to static.
8721 (malloc_state_e): Define.
8722 (malloc_state_names): Define.
8723 (funct_state_d): Add field malloc_state.
8724 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
8725 (check_retval_uses): New function.
8726 (malloc_candidate_p): Likewise.
8727 (analyze_function): Add support for malloc attribute.
8728 (pure_const_write_summary): Stream malloc_state.
8729 (pure_const_read_summary): Add support for reading malloc_state.
8730 (dump_malloc_lattice): New function.
8731 (propagate_malloc): New function.
8732 (warn_function_malloc): New function.
8733 (ipa_pure_const::execute): Call propagate_malloc and
8734 ipa_free_fn_summary.
8735 (pass_local_pure_const::execute): Add support for malloc attribute.
8736 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
8737 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
8739 2017-10-27 Martin Liska <mliska@suse.cz>
8741 PR gcov-profile/82457
8742 * doc/invoke.texi: Document that one needs a non-strict ISO mode
8743 for fork-like functions to be properly instrumented.
8745 2017-10-27 Richard Biener <rguenther@suse.de>
8748 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
8749 info when we redirected EH.
8751 2017-10-26 Michael Collison <michael.collison@arm.com>
8753 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
8755 (<optab>_trunchf<GPI:mode>2: New pattern.
8756 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
8757 * config/aarch64/iterators.md (wv): New mode attribute.
8758 (vf, VF): New mode attributes.
8759 (vgp, VGP): New mode attributes.
8760 (s): Update attribute with SImode and DImode prefixes.
8762 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
8764 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
8765 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
8766 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
8767 (nios2_option_overide): Initialize it. Don't allow R0-relative
8768 addressing with PIC.
8769 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
8770 (nios2_symbolic_constant_p): Likewise.
8771 (nios2_legitimate_address_p): Likewise.
8772 (nios2_r0rel_section_name_p): New.
8773 (nios2_symbol_ref_in_r0rel_data_p): New.
8774 (nios2_emit_move_sequence): Handle r0rel_constant_p.
8775 (r0rel_constant_p): New.
8776 (nios2_print_operand_address): Handle r0rel_constant_p.
8777 (nios2_cdx_narrow_form_p): Likewise.
8778 * config/nios2/nios2.opt (mr0rel-sec=): New option.
8779 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
8780 (Nios II Options): Document -mr0rel-sec.
8782 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
8784 * config/nios2/nios2.c: Include xregex.h.
8785 (nios2_gprel_sec_regex): New.
8786 (nios2_option_overide): Initialize it. Don't allow GP-relative
8787 addressing with PIC.
8788 (nios2_small_section_name_p): Check for regex match.
8789 * config/nios2/nios2.opt (mgprel-sec=): New option.
8790 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
8791 (Nios II Options): Document -mgprel-sec.
8793 2017-10-26 Jim Wilson <wilson@tuliptree.org>
8795 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
8797 2017-10-26 Tom de Vries <tom@codesourcery.com>
8799 PR tree-optimization/82707
8800 * gimple.c (gimple_copy): Fix unsharing of
8801 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
8803 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
8805 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
8806 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
8807 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
8808 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
8809 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
8810 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
8811 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
8812 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
8813 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
8814 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
8815 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
8816 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
8817 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
8818 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
8819 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
8820 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
8821 _mm512_mask_cmpunord_ps_mask): New intrinsics.
8823 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
8825 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
8827 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
8828 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
8829 warning to rs6000.c. Remove the Undocumented flag, since it has
8831 (-mabi=ibmlongdouble): Likewise.
8832 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
8833 already set, set the default format for long double.
8834 (rs6000_debug_reg_global): Print whether long double is IBM or
8836 (rs6000_option_override_internal): Rework setting long double
8837 format. Only warn if the user is changing the long double default
8838 and they did not use -Wno-psabi.
8839 * doc/invoke.texi (PowerPC options): Update the documentation for
8840 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
8842 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
8843 Alan Hayward <alan.hayward@arm.com>
8844 David Sherwood <david.sherwood@arm.com>
8846 * rtl.h (wider_subreg_mode): New function.
8847 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
8848 rather than an unsigned int *.
8849 * ira-color.c (regno_max_ref_width): Replace with...
8850 (regno_max_ref_mode): ...this new variable.
8851 (coalesced_pseudo_reg_slot_compare): Update accordingly.
8852 Use wider_subreg_mode.
8853 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
8854 rather than an unsigned int *.
8855 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
8856 (process_alt_operands): Likewise.
8857 (invariant_p): Likewise.
8858 * lra-spills.c (assign_mem_slot): Likewise.
8859 (add_pseudo_to_slot): Likewise.
8860 * lra.c (collect_non_operand_hard_regs): Likewise.
8861 (add_regs_to_insn_regno_info): Likewise.
8862 * reload1.c (regno_max_ref_width): Replace with...
8863 (regno_max_ref_mode): ...this new variable.
8864 (reload): Update accordingly. Update call to
8865 ira_sort_regnos_for_alter_reg.
8866 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
8867 (init_eliminable_invariants): Update to use regno_max_ref_mode.
8868 (scan_paradoxical_subregs): Likewise.
8870 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
8872 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
8873 (aarch64_frame): Add emit_frame_chain boolean.
8874 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
8875 Move eh_return case to aarch64_layout_frame.
8876 (aarch64_layout_frame): Initialize emit_frame_chain.
8877 (aarch64_expand_prologue): Use emit_frame_chain.
8879 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
8881 * config/aarch64/aarch64.c (aarch64_layout_frame):
8882 Ensure LR is always stored at the bottom of the callee-saves.
8883 Remove rarely used frame layout which saves callee-saves at top of
8884 frame, so the store of LR can be used as a valid probe in all cases.
8886 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
8888 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
8889 Improve unaligned TImode/TFmode base/offset split.
8891 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
8892 Alan Hayward <alan.hayward@arm.com>
8893 David Sherwood <david.sherwood@arm.com>
8895 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
8896 * combine.c (find_single_use_1): Likewise.
8897 (expand_field_assignment): Likewise.
8898 (move_deaths): Likewise.
8899 * lra-constraints.c (simplify_operand_subreg): Likewise.
8900 (curr_insn_transform): Likewise.
8901 * lra.c (collect_non_operand_hard_regs): Likewise.
8902 (add_regs_to_insn_regno_info): Likewise.
8903 * rtlanal.c (reg_referenced_p): Likewise.
8904 (covers_regno_no_parallel_p): Likewise.
8906 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
8908 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
8909 Don't print any bits outside the precision of the value.
8910 * wide-int.cc (test_printing): Add some new tests.
8912 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8914 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
8915 supports -xbrace_comment option.
8916 * configure: Regenerate.
8917 * config.in: Regenerate.
8918 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
8919 (ASM_CPU_SPEC): Use it.
8921 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
8923 * target.def (static_rtx_alignment): New hook.
8924 * targhooks.h (default_static_rtx_alignment): Declare.
8925 * targhooks.c (default_static_rtx_alignment): New function.
8926 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
8927 * doc/tm.texi: Regenerate.
8928 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
8929 instead of targetm.constant_alignment. Remove call to
8931 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
8932 (cris_preferred_mininum_alignment): New function, split out from...
8933 (cris_constant_alignment): ...here.
8934 (cris_static_rtx_alignment): New function.
8935 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
8937 (ix86_constant_alignment): ...here.
8938 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
8939 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
8940 (mmix_static_rtx_alignment): New function.
8941 * config/spu/spu.c (spu_static_rtx_alignment): New function.
8942 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
8944 2017-10-26 Tamar Christina <tamar.christina@arm.com>
8947 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
8948 Add flag_trapping_math and flag_fp_int_builtin_inexact.
8950 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
8952 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
8953 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
8954 mark as a sign-extending load.
8955 (local_pic_load_u): Define.
8957 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
8960 * fold-const.c (operand_equal_for_comparison_p): Also return true
8961 if ARG0 is a simple variant of ARG1 with narrower precision.
8962 (fold_ternary_loc): Always pass unstripped operands to the predicate.
8964 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
8966 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
8968 * i386.h (processor_costs): Add gather_static, gather_per_elt,
8969 scatter_static, scatter_per_elt.
8970 * x86-tune-costs.h: Add new cost entries.
8972 2017-10-25 Richard Biener <rguenther@suse.de>
8974 * tree-ssa-sccvn.h (vn_eliminate): Declare.
8975 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
8976 class pass_fre): Move to ...
8977 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
8978 class pass_fre): ... here and adjust for statistics.
8980 2017-10-25 Jakub Jelinek <jakub@redhat.com>
8983 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
8984 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
8985 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
8986 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
8989 2017-10-25 Richard Biener <rguenther@suse.de>
8991 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
8992 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
8993 eliminate_push_avail, eliminate_insert): Move inside...
8994 (class eliminate_dom_walker): ... this class in preparation
8996 (fini_eliminate): Remove by merging with ...
8997 (eliminate): ... this function. Adjust for class changes.
8998 (pass_pre::execute): Remove fini_eliminate call.
8999 (pass_fre::execute): Likewise.
9001 2017-10-24 Jakub Jelinek <jakub@redhat.com>
9004 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
9005 (VPERMI2, VPERMI2I): New mode iterators.
9006 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
9007 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
9009 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
9010 mode iterator. Remove 3 old define_insn patterns.
9011 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
9012 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
9013 VPERMI2 mode iterator, remove the other two expanders.
9014 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
9015 to use VPERMI2 mode iterator, add another alternative for vpermi2*
9016 instructions, remove the other two patterns.
9017 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
9018 mode iterator, remove the other two patterns.
9019 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
9020 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
9021 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
9022 and adjust argument order accordingly.
9023 (ix86_expand_vec_perm): Adjust caller.
9024 (expand_vec_perm_1): Likewise.
9025 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
9026 (expand_vec_perm_vpermt2_vpshub2): ... this.
9027 (ix86_expand_vec_perm_const_1): Adjust caller.
9028 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
9031 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
9032 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
9033 (vec_shl_<mode>): Remove unused expander.
9034 (avx512bw_<shift_insn><mode>3): New define_insn.
9035 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
9036 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
9038 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
9041 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
9044 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
9046 PR rtl-optimization/82396
9047 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
9048 (autopref_multipass_init): Simplify initialization.
9049 (autopref_rank_data): Simplify sort order.
9050 * gcc/sched-int.h (autopref_multipass_data_): Remove
9051 multi_mem_insn_p, min_offset and max_offset.
9053 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
9056 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
9057 Check special value of flag_omit_frame_pointer.
9058 (aarch64_can_eliminate): Likewise.
9059 (aarch64_override_options_after_change_1): Simplify handling of
9060 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
9062 2017-10-24 Richard Biener <rguenther@suse.de>
9064 PR tree-optimization/82697
9065 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
9066 zero for conditional load and unconditional store.
9068 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
9070 * doc/install.texi: Document bootstrap-cet.
9072 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
9075 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
9076 ENDBR instruction at function entrance if function is only
9079 2017-10-24 Jakub Jelinek <jakub@redhat.com>
9082 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
9083 patterns to better describe from which operation the CF is computed.
9084 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
9085 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
9086 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
9087 is 0, use _0 suffixed expanders instead of emitting a comparison
9090 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
9092 * config/i386/i386.md(*movsf_internal, *movdf_internal):
9093 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
9095 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
9098 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
9099 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
9100 * loop-iv.c (iv_get_reaching_def): Likewise.
9101 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
9102 variable is promoted and the partition contains undefined values.
9104 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9106 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
9108 (nios2_address_cost): Define.
9109 (nios2_legitimize_address): Recognize (exp + constant) directly.
9110 (TARGET_ADDRESS_COST): Define.
9112 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9114 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
9115 (nios2_symbolic_memory_operand_p): Declare.
9116 (nios2_split_large_constant): Declare.
9117 (nios2_split_symbolic_memory_operand): Declare.
9118 * config/nios2/nios2.c: Adjust includes.
9119 (nios2_symbolic_constant_allowed): New.
9120 (nios2_symbolic_constant_p): New.
9121 (nios2_plus_symbolic_constant_p): New.
9122 (nios2_valid_addr_expr_p): Recognize addresses involving
9124 (nios2_legitimate_address_p): Likewise, also LO_SUM.
9125 (nios2_symbolic_memory_operand_p): New.
9126 (nios2_large_constant_p): New.
9127 (nios2_split_large_constant): New.
9128 (nios2_split_plus_large_constant): New.
9129 (nios2_split_symbolic_memory_operand): New.
9130 (nios2_legitimize_address): Code refactoring. Handle addresses
9131 involving symbolic constants.
9132 (nios2_emit_move_sequence): Likewise.
9133 (nios2_print_operand): Improve error output.
9134 (nios2_print_operand_address): Handle LO_SUM.
9135 (nios2_cdx_narrow_form_p): Likewise.
9136 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
9137 operands involving symbolic constants.
9138 (movhi_internal, movsi_internal): Likewise.
9139 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
9140 (extendhisi2, extendqi<mode>2): Likewise.
9142 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9144 * tree-pass.h (PROP_rtl_split_insns): Define.
9145 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
9147 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9149 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
9151 2017-10-23 Jakub Jelinek <jakub@redhat.com>
9154 * target.def (const_not_ok_for_debug_p): Default to
9155 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
9156 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
9157 * targhooks.c (default_const_not_ok_for_debug_p): New function.
9158 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
9159 which targetm.const_not_ok_for_debug_p returned true.
9160 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
9162 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
9164 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
9165 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
9166 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
9168 (ix86_const_not_ok_for_debug_p): New function.
9169 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
9170 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
9172 2017-10-23 David Malcolm <dmalcolm@redhat.com>
9175 * system.h: Conditionally include "unique-ptr.h" if
9176 INCLUDE_UNIQUE_PTR is defined.
9177 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
9178 of defining INCLUDE_UNIQUE_PTR before including "system.h".
9180 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
9182 * config/rl78/rl78.md: New define_expand "subdi3".
9184 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
9187 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
9188 DF_REF_INSN if DF_REF_INSN_INFO is false.
9190 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
9192 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
9193 xmm_move instead of sse_move.
9194 (sse_store_index): New function.
9195 (ix86_register_move_cost): Be more sensible about mismatch stall;
9196 model AVX moves correctly; make difference between sse->integer and
9198 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
9199 moves; make difference between SSE and AVX.
9200 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
9201 and zmm_move. Increase size of sse load and store tables;
9202 add unaligned load and store tables; add ssemmx_to_integer.
9203 * x86-tune-costs.h: Update all entries according to real
9204 move latencies from Agner Fog's manual and chip documentation.
9206 2017-10-23 Jakub Jelinek <jakub@redhat.com>
9209 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
9210 * config/i386/constraints.md (Wf): New constraint.
9211 * config/i386/i386.md (UNSPEC_SBB): New unspec.
9212 (cmp<dwi>_doubleword): Removed.
9213 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
9214 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
9215 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
9216 expand with cmp<dwi>_doubleword. For LTU and GEU use
9217 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
9219 * common.opt (gcolumn-info): Enable by default.
9220 * doc/invoke.texi (gcolumn-info): Document new default.
9222 2017-10-23 Richard Biener <rguenther@suse.de>
9224 PR tree-optimization/82672
9225 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
9226 Fold the stmt if we propagated into it.
9228 2017-10-23 Richard Biener <rguenther@suse.de>
9230 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
9231 (bitmap_remove_expr_from_set): ... this. All callers call this
9232 for non-constant values.
9233 (bitmap_set_subtract): Rename to...
9234 (bitmap_set_subtract_expressions): ... this. Adjust and
9236 (bitmap_set_contains_value): Remove superfluous check.
9237 (bitmap_set_replace_value): Inline into single caller ...
9238 (bitmap_value_replace_in_set): ... here and simplify.
9239 (dependent_clean): Merge into ...
9240 (clean): ... this using an overload. Adjust.
9241 (prune_clobbered_mems): Adjust.
9242 (compute_antic_aux): Likewise.
9243 (compute_partial_antic_aux): Likewise.
9245 2017-10-23 Richard Biener <rguenther@suse.de>
9247 PR tree-optimization/82129
9249 2017-08-01 Richard Biener <rguenther@suse.de>
9251 PR tree-optimization/81181
9252 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
9253 (compute_antic): ... end of iteration here.
9255 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9257 * target.def (starting_frame_offset): New hook.
9258 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
9259 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
9260 * doc/tm.texi.in: Regenerate.
9261 * hooks.h (hook_hwi_void_0): Declare.
9262 * hooks.c (hook_hwi_void_0): New function.
9263 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
9264 STARTING_FRAME_OFFSET.
9265 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
9266 * reload1.c (reload): Likewise.
9267 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
9268 instead of STARTING_FRAME_OFFSET.
9269 * function.c (try_fit_stack_local): Likewise.
9270 (assign_stack_local_1): Likewise
9271 (instantiate_virtual_regs): Likewise.
9272 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
9273 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
9274 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
9275 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
9276 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
9277 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
9278 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
9279 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
9280 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
9281 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
9282 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
9283 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
9284 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
9285 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
9286 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
9287 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
9288 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
9289 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
9290 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
9291 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
9292 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
9293 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
9294 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
9295 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
9296 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
9297 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
9298 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
9299 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
9300 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
9301 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
9302 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
9303 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
9304 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
9305 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
9306 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
9307 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
9308 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
9309 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
9310 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
9311 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
9312 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
9313 * config/avr/avr.c (avr_starting_frame_offset): Make static and
9314 return a HOST_WIDE_INT.
9315 (avr_builtin_setjmp_frame_value): Use it instead of
9316 STARTING_FRAME_OFFSET.
9317 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9318 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
9319 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
9321 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9322 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
9323 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
9324 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9325 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
9326 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
9327 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9328 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
9329 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
9330 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9331 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
9332 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
9334 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9335 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
9336 * config/mips/mips.c (mips_compute_frame_info): Refer to
9337 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
9338 (mips_starting_frame_offset): New function.
9339 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9340 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
9341 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
9342 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
9343 and return a HOST_WIDE_INT.
9344 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9345 (mmix_initial_elimination_offset): Refer to
9346 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
9347 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
9348 * config/pa/pa.c (pa_starting_frame_offset): New function.
9349 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
9350 (pa_expand_prologue): Likewise.
9351 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9352 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
9353 !FRAME_GROWS_DOWNWARD handling to...
9354 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9355 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
9356 !FRAME_GROWS_DOWNWARD handling to...
9357 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9358 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
9359 !FRAME_GROWS_DOWNWARD handling to...
9360 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9361 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
9363 (rs6000_starting_frame_offset): New function.
9364 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
9365 !FRAME_GROWS_DOWNWARD handling to...
9366 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9367 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
9368 !FRAME_GROWS_DOWNWARD handling to...
9369 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9370 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
9371 !FRAME_GROWS_DOWNWARD handling to...
9372 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9373 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
9374 (rs6000_starting_frame_offset): New function.
9375 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
9376 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
9377 * config/vax/vax.c (vax_starting_frame_offset): New function.
9378 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
9379 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9380 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
9381 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
9382 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9383 * system.h (STARTING_FRAME_OFFSET): Poison.
9385 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9387 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
9388 SCALAR_TYPE_MODE instead of TYPE_MODE.
9390 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9391 Alan Hayward <alan.hayward@arm.com>
9392 David Sherwood <david.sherwood@arm.com>
9394 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
9396 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9397 Alan Hayward <alan.hayward@arm.com>
9398 David Sherwood <david.sherwood@arm.com>
9400 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
9402 2017-10-23 Richard Biener <rguenther@suse.de>
9404 PR tree-optimization/82129
9405 * tree-ssa-pre.c (bitmap_set_and): Remove.
9406 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
9407 canonicalizing expressions in the set to those with lowest
9408 ID rather than taking that from the first edge.
9410 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9412 * combine.c (rtx_equal_for_field_assignment_p): Use
9413 byte_lowpart_offset.
9415 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9416 Alan Hayward <alan.hayward@arm.com>
9417 David Sherwood <david.sherwood@arm.com>
9419 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
9420 to a SUBREG_PROMOTED_VAR.
9422 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9423 Alan Hayward <alan.hayward@arm.com>
9424 David Sherwood <david.sherwood@arm.com>
9426 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
9427 (expand_debug_source_expr): Likewise.
9428 * combine.c (combine_simplify_rtx): Likewise.
9429 * cse.c (fold_rtx): Likewise.
9430 * optabs.c (expand_float): Likewise.
9431 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9432 (simplify_binary_operation_1): Likewise.
9434 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9435 Alan Hayward <alan.hayward@arm.com>
9436 David Sherwood <david.sherwood@arm.com>
9438 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
9439 (record_promoted_value): Likewise.
9440 * expr.c (expand_expr_real_2): Likewise.
9441 * ree.c (update_reg_equal_equiv_notes): Likewise.
9442 (combine_set_extension): Likewise.
9443 * rtlanal.c (low_bitmask_len): Likewise.
9444 * simplify-rtx.c (neg_const_int): Likewise.
9445 (simplify_binary_operation_1): Likewise.
9447 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9448 Alan Hayward <alan.hayward@arm.com>
9449 David Sherwood <david.sherwood@arm.com>
9451 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
9452 * regcprop.c (maybe_mode_change): Likewise.
9453 * reload1.c (alter_reg): Likewise.
9455 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9457 * inchash.h (inchash::hash::add_wide_int): New function.
9458 * lto-streamer-out.c (hash_tree): Use it.
9460 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9462 * inchash.h (inchash::hash::add_wide_int): Rename to...
9463 (inchash::hash::add_hwi): ...this.
9464 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
9465 (polymorphic_call_target_hasher::hash): Likewise.
9466 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
9467 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
9468 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
9469 * lto-streamer-out.c (hash_tree): Likewise.
9470 * optc-save-gen.awk: Likewise.
9471 * tree.c (add_expr): Likewise.
9473 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
9476 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
9477 for ordered inequality comparisons even with TARGET_IEEE_FP.
9479 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
9482 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
9483 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
9484 Expand with cmp<dwi>_doubleword.
9486 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
9488 * extend.texi: Add x86 specific to 'nocf_check' attribute.
9489 List CET intrinsics.
9490 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
9491 specific to -fcf-protection option.
9493 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
9495 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
9496 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
9497 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
9498 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
9499 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
9500 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
9501 (extra_objs): Add cet.o for Linux/x86 targets.
9502 (tmake_file): Add i386/t-cet for Linux/x86 targets.
9503 * config/i386/cet.c: New file.
9504 * config/i386/cetintrin.h: Likewise.
9505 * config/i386/t-cet: Likewise.
9506 * config/i386/cpuid.h (bit_SHSTK): New.
9507 (bit_IBT): Likewise.
9508 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
9509 pass IBT and SHSTK bits.
9510 * config/i386/i386-builtin-types.def
9511 (VOID_FTYPE_UNSIGNED_PVOID): New.
9512 (VOID_FTYPE_UINT64_PVOID): Likewise.
9513 * config/i386/i386-builtin.def: Add CET intrinsics.
9514 * config/i386/i386-c.c (ix86_target_macros_internal): Add
9515 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
9516 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
9517 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
9519 * config/i386/i386.c (rest_of_insert_endbranch): New.
9520 (pass_data_insert_endbranch): Likewise.
9521 (pass_insert_endbranch): Likewise.
9522 (make_pass_insert_endbranch): Likewise.
9523 (ix86_notrack_prefixed_insn_p): Likewise.
9524 (ix86_target_string): Add -mibt, -mshstk flags.
9525 (ix86_option_override_internal): Add flag_cf_protection
9527 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
9528 (ix86_print_operand): Add 'notrack' prefix output.
9529 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
9530 (ix86_expand_builtin): Expand CET intrinsics.
9531 (x86_output_mi_thunk): Add 'endbranch' instruction.
9532 * config/i386/i386.h (TARGET_IBT): New.
9533 (TARGET_IBT_P): Likewise.
9534 (TARGET_SHSTK): Likewise.
9535 (TARGET_SHSTK_P): Likewise.
9536 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
9537 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
9538 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
9539 (builtin_setjmp_setup): New pattern.
9540 (builtin_longjmp): Likewise.
9541 (rdssp<mode>): Likewise.
9542 (incssp<mode>): Likewise.
9543 (saveprevssp): Likewise.
9544 (rstorssp): Likewise.
9545 (wrss<mode>): Likewise.
9546 (wruss<mode>): Likewise.
9547 (setssbsy): Likewise.
9548 (clrssbsy): Likewise.
9549 (nop_endbr): Likewise.
9550 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
9552 * config/i386/immintrin.h: Include <cetintrin.h>.
9553 * config/i386/linux-common.h
9554 (file_end_indicate_exec_stack_and_cet): New prototype.
9555 (TARGET_ASM_FILE_END): New.
9557 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
9559 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
9560 latencies instead of having separate table; make difference between
9561 integer and float costs.
9562 * i386.h (processor_costs): Remove scalar_stmt_cost,
9563 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
9564 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
9566 * x86-tune-costs.h: Remove entries which has been removed in
9567 procesor_costs from all tables; make cond_taken_branch_cost
9568 and cond_not_taken_branch_cost COST_N_INSNS based.
9570 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
9572 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
9574 2017-10-20 Jakub Jelinek <jakub@redhat.com>
9576 * config/i386/i386.md (isa): Remove fma_avx512f.
9577 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
9578 <avx512>_fmadd_<mode>_mask3<round_name>,
9579 <avx512>_fmsub_<mode>_mask<round_name>,
9580 <avx512>_fmsub_<mode>_mask3<round_name>,
9581 <avx512>_fnmadd_<mode>_mask<round_name>,
9582 <avx512>_fnmadd_<mode>_mask3<round_name>,
9583 <avx512>_fnmsub_<mode>_mask<round_name>,
9584 <avx512>_fnmsub_<mode>_mask3<round_name>,
9585 <avx512>_fmaddsub_<mode>_mask<round_name>,
9586 <avx512>_fmaddsub_<mode>_mask3<round_name>,
9587 <avx512>_fmsubadd_<mode>_mask<round_name>,
9588 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
9589 (*vec_widen_umult_even_v16si<mask_name>,
9590 *vec_widen_smult_even_v16si<mask_name>): Likewise.
9591 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
9593 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
9595 * extend.texi: Add 'nocf_check' documentation.
9596 * gimple.texi: Add second parameter to
9597 gimple_build_call_from_tree.
9598 * invoke.texi: Add -fcf-protection documentation.
9599 * rtl.texi: Add REG_CALL_NOTRACK documenation.
9601 2017-10-20 Richard Biener <rguenther@suse.de>
9603 PR tree-optimization/82473
9604 * tree-vect-loop.c (vectorizable_reduction): Properly get at
9605 the largest input type.
9607 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
9609 * c-attribs.c (handle_nocf_check_attribute): New function.
9610 (c_common_attribute_table): Add 'nocf_check' handling.
9611 * gimple-parser.c: Add second argument NULL to
9612 gimple_build_call_from_tree.
9613 * attrib.c (comp_type_attributes): Check nocf_check attribute.
9614 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
9616 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
9617 * common.opt: Add fcf-protection flag.
9618 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
9619 * flag-types.h: Add enum cf_protection_level.
9620 * gimple.c (gimple_build_call_from_tree): Add second parameter.
9621 Add 'nocf_check' attribute propagation to gimple call.
9622 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
9623 (gimple_build_call_from_tree): Update prototype.
9624 (gimple_call_nocf_check_p): New function.
9625 (gimple_call_set_nocf_check): Likewise.
9626 * gimplify.c: Add second argument to gimple_build_call_from_tree.
9627 * ipa-icf.c: Add nocf_check attribute in statement hash.
9628 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
9629 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
9630 * toplev.c (process_options): Add flag_cf_protection handling.
9632 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
9634 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
9636 2017-10-20 Richard Biener <rguenther@suse.de>
9638 PR tree-optimization/82603
9639 * tree-if-conv.c (predicate_mem_writes): Make sure to only
9640 remove false predicated stores.
9642 2017-10-20 Richard Biener <rguenther@suse.de>
9644 * graphite-isl-ast-to-gimple.c
9645 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
9646 Remove return value and simplify, dump copied stmt after lhs
9648 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
9649 Reduce dump verbosity.
9650 (gsi_insert_earliest): Likewise.
9651 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
9652 * graphite.c (print_global_statistics): Adjust dumping.
9653 (print_graphite_scop_statistics): Likewise.
9654 (print_graphite_statistics): Do not dump loops here.
9655 (graphite_transform_loops): But here.
9657 2017-10-20 Nicolas Roche <roche@adacore.com>
9659 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
9660 * configure: Regenerate.
9662 2017-10-20 Jakub Jelinek <jakub@redhat.com>
9665 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
9666 functions when optimizing replace GIMPLE_RETURN stmts with
9667 calls to __builtin_unreachable ().
9670 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
9671 for -fsanitize=thread link of executables.
9672 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
9673 link of executables.
9676 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
9678 (<shift_insn><mode>3<mask_name>): Change the last of the 3
9679 define_insns for logical vector shifts to use VI248_AVX512BW
9680 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
9681 condition, useless isa and prefix attributes. Change the first
9682 2 of these define_insns to ...
9683 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
9684 define_insn for avx512vl.
9685 (<shift_insn><mode>3): ... and this, new define_insn without
9686 masking for non-avx512vl.
9689 * config/i386/sse.md (*andnot<mode>3,
9690 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
9691 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
9692 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
9693 not applied use empty suffix even for TARGET_AVX512VL.
9694 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
9695 is applied, supply evex,evex or evex,evex,evex instead of just
9698 2017-10-20 Julia Koval <julia.koval@intel.com>
9700 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
9701 (OPTION_MASK_ISA_GFNI_UNSET): New.
9702 (ix86_handle_option): Handle OPT_mgfni.
9703 * config/i386/cpuid.h (bit_GFNI): New.
9704 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
9705 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
9706 * config/i386/i386.c (ix86_target_string): Add -mgfni.
9707 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
9708 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
9709 * config/i386/i386.opt: Add mgfni.
9711 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
9713 * config/msp430/msp430.c (msp430_option_override): Disable
9714 -fdelete-null-pointer-checks.
9715 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
9717 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
9719 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
9720 of x87 and SSE instructions.
9722 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
9724 * asan.c (create_cond_insert_point): Do not update edge count.
9725 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
9726 (afdo_propagate_circuit): Likewise.
9727 (afdo_calculate_branch_prob): Likewise.
9728 (afdo_annotate_cfg): Likewise.
9729 * basic-block.h (struct edge_def): Remove count.
9730 (edge_def::count): New accessor.
9731 * bb-reorder.c (rotate_loop): Update.
9732 (find_traces_1_round): Update.
9733 (connect_traces): Update.
9734 (sanitize_hot_paths): Update.
9735 * cfg.c (unchecked_make_edge): Update.
9736 (make_single_succ_edge): Update.
9737 (check_bb_profile): Update.
9738 (dump_edge_info): Update.
9739 (update_bb_profile_for_threading): Update.
9740 (scale_bbs_frequencies_int): Update.
9741 (scale_bbs_frequencies_gcov_type): Update.
9742 (scale_bbs_frequencies_profile_count): Update.
9743 (scale_bbs_frequencies): Update.
9744 * cfganal.c (connect_infinite_loops_to_exit): Update.
9745 * cfgbuild.c (compute_outgoing_frequencies): Update.
9746 (find_many_sub_basic_blocks): Update.
9747 * cfgcleanup.c (try_forward_edges): Update.
9748 (try_crossjump_to_edge): Update
9749 * cfgexpand.c (expand_gimple_cond): Update
9750 (expand_gimple_tailcall): Update
9751 (construct_exit_block): Update
9752 * cfghooks.c (verify_flow_info): Update
9753 (redirect_edge_succ_nodup): Update
9754 (split_edge): Update
9755 (make_forwarder_block): Update
9756 (duplicate_block): Update
9757 (account_profile_record): Update
9758 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
9759 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
9760 * cfgloopmanip.c (scale_loop_profile): Update.
9762 (lv_adjust_loop_entry_edge): Update.
9763 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
9764 (force_nonfallthru_and_redirect): Update.
9765 (purge_dead_edges): Update.
9766 (rtl_flow_call_edges_add): Update.
9767 * cgraphunit.c (init_lowered_empty_function): Update.
9768 (cgraph_node::expand_thunk): Update.
9769 * gimple-pretty-print.c (dump_probability): Update.
9770 (dump_edge_probability): Update.
9771 * gimple-ssa-isolate-paths.c (isolate_path): Update.
9772 * haifa-sched.c (sched_create_recovery_edges): Update.
9773 * hsa-gen.c (convert_switch_statements): Update.
9774 * ifcvt.c (dead_or_predicable): Update.
9775 * ipa-inline-transform.c (inline_transform): Update.
9776 * ipa-split.c (split_function): Update.
9777 * ipa-utils.c (ipa_merge_profiles): Update.
9778 * loop-doloop.c (add_test): Update.
9779 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
9780 * lto-streamer-in.c (input_cfg): Update.
9781 (input_function): Update.
9782 * lto-streamer-out.c (output_cfg): Update.
9783 * modulo-sched.c (sms_schedule): Update.
9784 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
9785 * predict.c (maybe_hot_edge_p): Update.
9786 (unlikely_executed_edge_p): Update.
9787 (probably_never_executed_edge_p): Update.
9788 (dump_prediction): Update.
9789 (drop_profile): Update.
9790 (propagate_unlikely_bbs_forward): Update.
9791 (determine_unlikely_bbs): Update.
9792 (force_edge_cold): Update.
9793 * profile.c (compute_branch_probabilities): Update.
9794 * reg-stack.c (better_edge): Update.
9795 * shrink-wrap.c (handle_simple_exit): Update.
9796 * tracer.c (better_p): Update.
9797 * trans-mem.c (expand_transaction): Update.
9798 (split_bb_make_tm_edge): Update.
9799 * tree-call-cdce.c: Update.
9800 * tree-cfg.c (gimple_find_sub_bbs): Update.
9801 (gimple_split_edge): Update.
9802 (gimple_duplicate_sese_region): Update.
9803 (gimple_duplicate_sese_tail): Update.
9804 (gimple_flow_call_edges_add): Update.
9805 (insert_cond_bb): Update.
9806 (execute_fixup_cfg): Update.
9807 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
9808 * tree-complex.c (expand_complex_div_wide): Update.
9809 * tree-eh.c (lower_resx): Update.
9810 (unsplit_eh): Update.
9811 (cleanup_empty_eh_move_lp): Update.
9812 * tree-inline.c (copy_edges_for_bb): Update.
9813 (freqs_to_counts): Update.
9814 (copy_cfg_body): Update.
9815 * tree-ssa-dce.c (remove_dead_stmt): Update.
9816 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
9817 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
9818 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
9819 (unloop_loops): Update.
9820 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
9821 * tree-ssa-loop-split.c (connect_loops): Update.
9822 (split_loop): Update.
9823 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
9824 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
9825 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
9826 * tree-ssa-reassoc.c (branch_fixup): Update.
9827 * tree-ssa-tail-merge.c (replace_block_by): Update.
9828 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
9829 (compute_path_counts): Update.
9830 (update_profile): Update.
9831 (recompute_probabilities): Update.
9832 (update_joiner_offpath_counts): Update.
9833 (estimated_freqs_path): Update.
9834 (freqs_to_counts_path): Update.
9835 (clear_counts_path): Update.
9836 (ssa_fix_duplicate_block_edges): Update.
9837 (duplicate_thread_path): Update.
9838 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
9839 (case_bit_test_cmp): Update.
9840 (collect_switch_conv_info): Update.
9841 (gen_inbound_check): Update.
9842 (do_jump_if_equal): Update.
9843 (emit_cmp_and_jump_insns): Update.
9844 * tree-tailcall.c (decrease_profile): Update.
9845 (eliminate_tail_call): Update.
9846 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
9847 (vect_do_peeling): Update.
9848 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
9849 * ubsan.c (ubsan_expand_null_ifn): Update.
9850 (ubsan_expand_ptr_ifn): Update.
9851 * value-prof.c (gimple_divmod_fixed_value): Update.
9852 (gimple_mod_pow2): Update.
9853 (gimple_mod_subtract): Update.
9854 (gimple_ic): Update.
9855 (gimple_stringop_fixed_value): Update.
9857 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
9860 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
9862 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
9864 PR rtl-optimization/82395
9865 * ira-color.c (allocno_priority_compare_func): Fix comparison step
9866 based on non_spilled_static_chain_regno_p.
9868 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
9870 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
9871 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
9872 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
9874 2017-10-19 Martin Sebor <msebor@redhat.com>
9876 PR tree-optimization/82596
9877 * tree.c (array_at_struct_end_p): Handle STRING_CST.
9879 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
9881 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
9882 (get_mem_refs_of_builtin_call): Likewise.
9883 * builtins.c (expand_builtin_apply): Adjust call to
9884 allocate_dynamic_stack_space.
9885 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
9886 the third argument to allocate_dynamic_stack_space, otherwise -1.
9887 (expand_builtin): Deal with all alloca variants.
9888 (is_inexpensive_builtin): Likewise.
9889 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
9890 * calls.c (special_function_p): Deal with all alloca variants.
9891 (initialize_argument_information): Adjust call to
9892 allocate_dynamic_stack_space.
9893 (expand_call): Likewise.
9894 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
9895 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
9896 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
9897 use it for the stack usage computation.
9898 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
9899 * function.c (gimplify_parameters): Call build_alloca_call_expr.
9900 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
9901 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
9902 (in_loop_p): Remove first argument and useless check.
9903 (pass_walloca::execute): Remove useless test and adjust call to above.
9904 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
9905 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
9906 (gimplify_call_expr): Deal with all alloca variants.
9907 * hsa-gen.c (gen_hsa_alloca): Likewise.
9908 (gen_hsa_insns_for_call): Likewise.
9909 * ipa-pure-const.c (special_builtin_state): Likewise.
9910 * tree-chkp.c (chkp_build_returned_bound): Likewise.
9911 * tree-object-size.c (alloc_object_size): Likewise.
9912 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
9913 (call_may_clobber_ref_p_1): Likewise.
9914 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
9915 (ccp_fold_stmt): Likewise.
9916 (optimize_stack_restore): Likewise.
9917 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
9918 (mark_all_reaching_defs_necessary_1): Likewise.
9919 (propagate_necessity): Likewise.
9920 (eliminate_unnecessary_stmts): Likewise.
9921 * tree.c (build_common_builtin_nodes): Build
9922 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
9923 (build_alloca_call_expr): New function.
9924 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
9925 (CASE_BUILT_IN_ALLOCA): Likewise.
9926 (build_alloca_call_expr): Declare.
9927 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
9929 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
9932 * dwarf2out.c (new_die_raw): New static inline function.
9933 (new_die): Use it to create the DIE.
9934 (add_AT_external_die_ref): Likewise.
9935 (clone_die): Likewise.
9936 (clone_as_declaration): Likewise.
9937 (dwarf2out_vms_debug_main_pointer): Likewise.
9938 (base_type_die): Likewise. Remove early return for corner cases.
9939 Do not call add_pubtype on the DIE here.
9940 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
9941 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
9942 typedefs for base types with DW_AT_endianity. Make sure a DIE with
9943 native order exists for base types, attach the DIE manually and call
9944 add_pubtype on it. Do not equate a reverse order DIE to the type.
9946 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
9948 * config/arm/arm.c (align_ok_ldrd_strd): New function.
9949 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
9951 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
9953 2017-10-19 Jakub Jelinek <jakub@redhat.com>
9955 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
9956 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
9957 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
9958 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
9959 * opts.c (sanitizer_opts): Add builtin.
9960 * ubsan.c (instrument_builtin): New function.
9961 (pass_ubsan::execute): Call it.
9962 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
9963 * doc/invoke.texi: Document -fsanitize=builtin.
9965 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
9966 builtins, store max (log2 (align), 0) into uchar field instead of
9967 align into uptr field.
9968 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
9969 store uchar 0 field instead of uptr 0 field.
9970 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
9971 instead of passing one address of struct with 2 locations pass
9972 two addresses of structs with 1 location each.
9973 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
9974 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
9975 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
9976 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
9977 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
9978 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
9979 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
9980 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
9982 2017-10-19 Martin Liska <mliska@suse.cz>
9985 * file-find.c (remove_prefix): Remove.
9986 * file-find.h (remove_prefix): Likewise.
9987 * gcc-ar.c: Remove smartness of lookup.
9989 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
9991 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
9992 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
9993 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
9995 2017-10-19 Jakub Jelinek <jakub@redhat.com>
9998 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
9999 (setcc + and to xor + setcc): New peephole2.
10001 2017-10-19 Tom de Vries <tom@codesourcery.com>
10003 * doc/sourcebuild.texi (Test Directives, Variants of
10004 dg-require-support): Add dg-require-stack-size.
10006 2017-10-19 Martin Liska <mliska@suse.cz>
10009 * gimplify.c (gimplify_decl_expr): Do not instrument variables
10010 that have a large alignment.
10011 (gimplify_target_expr): Likewise.
10013 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
10015 PR rtl-optimization/82602
10016 * ira.c (rtx_moveable_p): Return false for volatile asm.
10018 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
10021 * config/i386/i386-modes.def (CCGZ): New CC mode.
10022 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
10023 * config/i386/predicates.md (ix86_comparison_operator):
10025 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
10026 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
10027 with double-word subtraction.
10028 (put_condition_code): Handle CCGZmode.
10030 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
10032 * wide-int.cc (debug (const wide_int &)): New.
10033 (debug (const wide_int *)): New.
10034 (debug (const widest_int &)): New.
10035 (debug (const widest_int *)): New.
10037 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
10039 PR middle-end/82556
10040 * lra-constraints.c (curr_insn_transform): Use non-input operand
10041 instead of output one for matched reload.
10043 2017-10-18 Bin Cheng <bin.cheng@arm.com>
10045 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
10046 (tree-ssa-loop-ivopts.h): New header file.
10047 (struct builtin_info): New fields.
10048 (classify_builtin_1): Compute and record base and offset parts for
10049 memset builtin partition by calling strip_offset.
10050 (offset_cmp, fuse_memset_builtins): New functions.
10051 (finalize_partitions): Fuse adjacent memset partitions by calling
10053 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
10054 Expose the interface.
10055 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
10057 2017-10-18 Bin Cheng <bin.cheng@arm.com>
10059 PR tree-optimization/82574
10060 * tree-loop-distribution.c (find_single_drs): New parameter. Check
10061 that data reference must be executed exactly once per iteration
10062 against the outermost loop in nest.
10063 (classify_partition): Update call to above function.
10065 2017-10-18 Richard Biener <rguenther@suse.de>
10067 PR tree-optimization/82591
10068 * graphite.c (graphite_transform_loops): Move code gen message
10070 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
10071 Here. Handle scop_to_isl_ast failing.
10072 (scop_to_isl_ast): Limit the number of ISL operations.
10074 2017-10-18 Richard Biener <rguenther@suse.de>
10076 * graphite-isl-ast-to-gimple.c
10077 (translate_isl_ast_to_gimple::set_rename): Simplify.
10078 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
10079 (graphite_copy_stmts_from_block): ... here.
10080 (copy_bb_and_scalar_dependences): Simplify.
10081 (add_parameters_to_ivs_params): Canonicalize.
10082 (generate_entry_out_of_ssa_copies): Simplify.
10083 * graphite-sese-to-poly.c (extract_affine_name): Simplify
10084 by passing in ISL dimension.
10085 (parameter_index_in_region_1): Rename to ...
10086 (parameter_index_in_region): ... this.
10087 (extract_affine): Adjust assert, pass down parameter index.
10088 (add_param_constraints): Use range-info when available.
10089 (build_scop_context): Adjust.
10090 * sese.c (new_sese_info): Adjust.
10091 (free_sese_info): Likewise.
10092 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
10093 Remove unused typedefs.
10094 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
10096 2017-10-18 Martin Liska <mliska@suse.cz>
10098 * combine.c (simplify_compare_const): Add gcc_fallthrough.
10100 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
10102 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
10103 (s390_sched_init): Do not reset s390_sched_state if we entered the
10104 current basic block via a fallthru edge and all others are unlikely.
10106 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
10108 * config/s390/s390.c (NUM_SIDES): New variable.
10109 (LONGRUNNING_THRESHOLD): New variable.
10110 (LATENCY_FACTOR): New variable.
10111 (s390_sched_score): Decrease score for long-running instructions on
10113 (s390_sched_variable_issue): Perform bookkeeping for long-running
10116 2017-10-18 Richard Biener <rguenther@suse.de>
10118 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
10119 Simplify with removal of the parameter rename map.
10120 (set_rename): Likewise.
10121 (should_copy_to_new_region): Likewise.
10122 (graphite_copy_stmts_from_block): Likewise.
10123 (copy_bb_and_scalar_dependences): Remove initialization of
10124 unused copied_bb_map.
10125 (copy_def): Remove.
10126 (copy_internal_parameters): Likewise.
10127 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
10128 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
10129 Use INTEGRAL_TYPE_P.
10130 (parameter_index_in_region_1): Rename to ...
10131 (assign_parameter_index_in_region): ... this. Assert we have
10132 a parameter we handle.
10133 (scan_tree_for_params): Adjust.
10134 * sese.h (parameter_rename_map_t): Remove.
10135 (struct sese_info_t): Remove unused parameter_rename_map and
10136 copied_bb_map members.
10137 * sese.c (new_sese_info): Adjust.
10138 (free_sese_info): Likewise.
10140 2017-10-18 Martin Liska <mliska@suse.cz>
10143 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
10144 on an abnormal edge.
10146 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
10148 * doc/invoke.texi (ffunction-sections and fdata-sections):
10151 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
10153 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
10154 the use statement can throw internally.
10156 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
10158 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
10159 any RTX present on the RHS of a SET.
10160 * compare-elim.c (try_eliminate_compare): Restore comment.
10162 2017-10-17 Jakub Jelinek <jakub@redhat.com>
10164 * langhooks.h (struct lang_hooks): Document that tree_size langhook
10165 may be also called on tcc_type nodes.
10166 * langhooks.c (lhd_tree_size): Likewise.
10168 2017-10-17 David Malcolm <dmalcolm@redhat.com>
10170 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
10171 format_warning_at_substring.
10172 (maybe_warn): Convert source_range * param to a location_t. Pass
10173 UNKNOWN_LOCATION rather than NULL to fmtwarn.
10174 (format_directive): Remove code to extract source_ranges and
10175 source_range * in favor of just a location_t.
10176 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
10178 * substring-locations.c (format_warning_va): Convert
10179 source_range * param to a location_t.
10180 (format_warning_at_substring): Likewise.
10181 * substring-locations.h (format_warning_va): Likewise.
10182 (format_warning_at_substring): Likewise.
10184 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
10186 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
10188 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
10189 and scatter/gather ops.
10191 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
10192 vec_gather_load and vec_scatter_store.
10193 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
10194 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
10195 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
10196 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
10197 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
10198 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
10200 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
10202 * reg-stack.c (compare_for_stack_reg): Add bool argument.
10203 Detect FTST instruction and handle its register pops. Only pop
10204 second operand if can_pop_second_op is true.
10205 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
10206 set can_pop_second_op to false in the compare_for_stack_reg call.
10208 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
10209 output_fp_compare for stack register operands.
10210 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
10211 instructions here. Do not emit stack register pops here. Assert
10212 that FCOMPP pops next to top stack register. Rewrite function.
10214 2017-10-17 Nathan Sidwell <nathan@acm.org>
10216 PR middle-end/82577
10217 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
10218 use DECL_ASSEMBLER_NAME_RAW.
10220 PR middle-end/82546
10221 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
10224 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
10225 Wilco Dijkstra <wilco.dijkstra@arm.com>
10227 * builtins.c (expand_builtin_update_setjmp_buf): Add a
10228 converstion to Pmode from the buf_addr.
10230 2017-10-17 Richard Biener <rguenther@suse.de>
10232 * graphite-dependences.c (scop_get_reads_and_writes): Change
10233 output parameters to references.
10235 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
10237 PR 71026/tree-optimization
10238 * fold-const.c (distribute_real_division): Removed.
10239 (fold_binary_loc): Remove calls to distribute_real_divison.
10241 2017-10-17 Richard Biener <rguenther@suse.de>
10243 * graphite-scop-detection.c
10244 (scop_detection::stmt_has_simple_data_refs_p): Always use
10245 the full nest as region.
10246 (try_generate_gimple_bb): Likewise.
10247 * sese.c (scalar_evolution_in_region): Simplify now that
10248 SCEV can handle instantiation in regions.
10249 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
10250 in the non-loop part of a function if requested.
10252 2017-10-17 Richard Biener <rguenther@suse.de>
10254 PR tree-optimization/82563
10255 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
10257 (graphite_regenerate_ast_isl): Call it.
10258 * graphite-scop-detection.c (build_scops): Remove entry edge split.
10260 2017-10-17 Jakub Jelinek <jakub@redhat.com>
10262 PR tree-optimization/82549
10263 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
10264 Formatting fixes. Instead of calling make_bit_field_ref with negative
10267 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
10269 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
10270 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
10271 _mm_maskz_reduce_ss): New.
10272 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
10273 __builtin_ia32_reducess_mask): Ditto..
10274 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
10275 * config/i386/sse.md (reduces<mode>): Renamed to ...
10276 (reduces<mode><mask_scalar_name>): ... this.
10277 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
10279 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
10280 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
10282 2017-10-16 David Malcolm <dmalcolm@redhat.com>
10284 * Makefile.in (OBJS): Add unique-ptr-tests.o.
10285 * selftest-run-tests.c (selftest::run_tests): Call
10286 selftest::unique_ptr_tests_cc_tests.
10287 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
10288 * unique-ptr-tests.cc: New file.
10290 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
10293 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
10295 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
10296 (make_hard_regno_born, make_hard_regno_dead): Update
10297 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
10299 2017-10-16 Jeff Law <law@redhat.com>
10301 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
10303 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
10305 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
10307 2017-10-16 Olivier Hainque <hainque@adacore.com>
10309 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
10310 with_cpu if we were configured for an e500v2 target cpu name.
10312 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
10314 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
10315 * doc/invoke.texi: Document +nodsp as a valid extension for
10318 2017-10-16 Martin Liska <mliska@suse.cz>
10320 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
10321 (test_set_range): Likewise.
10322 (test_range_functions): Rename to ...
10323 (test_bit_in_range): ... this.
10324 (sbitmap_c_tests): Add new test.
10326 2017-10-16 Tamar Christina <tamar.christina@arm.com>
10328 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
10330 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
10331 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
10333 2017-10-16 Tamar Christina <tamar.christina@arm.com>
10335 * config/aarch64/aarch64-builtins.c
10336 (aarch64_types_quadopu_lane_qualifiers): New.
10337 (TYPES_QUADOPU_LANE): New.
10338 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
10339 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
10340 (aarch64_<sur>dot_laneq<vsi2qi>): New.
10341 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
10342 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
10343 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
10344 (Vdottype, DOTPROD): New.
10345 (sur): Add SDOT and UDOT.
10347 2017-10-16 Tamar Christina <tamar.christina@arm.com>
10349 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
10350 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
10351 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
10352 Add TARGET_DOTPROD.
10353 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
10354 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
10355 Enable TARGET_DOTPROD.
10356 (cortex-a75.cortex-a55): Likewise.
10357 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
10359 2017-10-16 Tamar Christina <tamar.christina@arm.com>
10361 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
10362 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
10364 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
10366 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
10367 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
10368 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
10369 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
10370 * config/arm/types.md (neon_dot, neon_dot_q): New.
10371 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
10373 2017-10-16 Tamar Christina <tamar.christina@arm.com>
10375 * config/arm/arm.h (TARGET_DOTPROD): New.
10376 * config/arm/arm.c (arm_arch_dotprod): New.
10377 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
10378 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
10379 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
10380 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
10381 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
10382 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
10383 * doc/invoke.texi (armv8.2-a): Document dotprod
10385 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
10387 * i386.c (ix86_vec_cost): New function.
10388 (ix86_rtx_costs): Handle vector operations better.
10389 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
10390 * x86-tune-costs.h: Add new costs to all tables.
10392 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
10394 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
10396 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
10397 divsd, sqrtss and sqrtsd
10398 * x86-tune-costs.h: Add new entries to all costs.
10399 (znver1_cost): Fix to match real instruction latencies.
10401 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10402 Michael Collison <michael.collison@arm.com>
10404 * compare-elim.c: Include emit-rtl.h.
10405 (can_merge_compare_into_arith): New function.
10406 (try_validate_parallel): Likewise.
10407 (try_merge_compare): Likewise.
10408 (try_eliminate_compare): Call the above when no previous clobber
10410 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
10413 2017-10-14 Jakub Jelinek <jakub@redhat.com>
10415 PR middle-end/62263
10416 PR middle-end/82498
10417 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
10418 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
10420 PR middle-end/62263
10421 PR middle-end/82498
10422 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
10423 to be any operand_equal_p operands. For & (B - 1) require
10424 B to be power of 2. Recognize
10425 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
10427 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
10430 * optabs.c (expand_memory_blockage): Fix call of
10431 targetm.have_memory_blockage.
10433 2017-10-14 Jakub Jelinek <jakub@redhat.com>
10436 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
10437 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
10438 objects to extra_objs instead of overwriting it.
10440 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
10442 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
10443 Use any_fp_register_operand as operand[3] predicate. Simplify
10444 equality test for operands[2] and operands[4] memory location.
10445 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
10446 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
10447 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
10448 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
10449 any_fp_register_operand as operand[1] predicate. Simplify
10450 equality test for operands[0] and operands[3] memory location.
10451 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
10452 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
10453 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
10455 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
10457 * target-insns.def: Add memory_blockage.
10458 * optabs.c (expand_memory_blockage): New function.
10459 (expand_asm_memory_barrier): Rename ...
10460 (expand_asm_memory_blockage): ... to this.
10461 (expand_mem_thread_fence): Call expand_memory_blockage
10462 instead of expand_asm_memory_barrier.
10463 (expand_mem_singnal_fence): Ditto.
10464 (expand_atomic_load): Ditto.
10465 (expand_atomic_store): Ditto.
10466 * doc/md.texi (Standard Pattern Names For Generation):
10467 Document memory_blockage instruction pattern.
10469 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
10471 * config/rl78/rl78.c (rl78_emit_libcall): New function.
10472 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
10473 * config/rl78/rl78.md: New define_expand "adddi3".
10475 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
10477 * cfghooks.c (verify_flow_info): Disable check that all probabilities
10480 2017-10-13 Jeff Law <law@redhat.com>
10482 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
10484 2017-10-13 Jakub Jelinek <jakub@redhat.com>
10487 * internal-fn.c (expand_mul_overflow): If both operands have
10488 the same highpart of -1 or 0 and the topmost bit of lowpart
10489 is different, overflow is if res <= 0 rather than res < 0.
10491 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
10493 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
10494 TARGET_P9_VECTOR code for unaligned_load case.
10496 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
10498 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
10500 2017-10-13 Nathan Sidwell <nathan@acm.org>
10502 * tree-core.h (tree_contains_struct): Make bool.
10503 * tree.c (tree_contains_struct): Likewise.
10504 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
10505 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
10506 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
10507 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
10508 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
10510 2017-10-13 Richard Biener <rguenther@suse.de>
10512 * graphite-isl-ast-to-gimple.c
10513 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
10514 parameters and dominance check.
10515 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
10516 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
10517 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
10518 Do not update SSA form here or do intermediate IL verification.
10519 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
10520 (graphite_initialize): Remove check on the number of loops in
10521 the function and inline into graphite_transform_loops.
10522 (graphite_finalize): Inline into graphite_transform_loops.
10523 (graphite_transform_loops): Perform SSA update and IL verification
10525 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
10527 2017-10-13 Richard Biener <rguenther@suse.de>
10529 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
10530 graphite_expression_type_precision): Avoid global constructor
10532 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
10533 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
10534 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
10535 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
10536 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
10537 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
10539 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
10542 * config/i386/i386.h (ix86_red_zone_size): New.
10543 * config/i386/i386.md (push peephole2s): Replace
10544 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
10546 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
10547 Alan Hayward <alan.hayward@arm.com>
10548 David Sherwood <david.sherwood@arm.com>
10550 * combine.c (can_change_dest_mode): Reject changes in
10551 REGMODE_NATURAL_SIZE.
10553 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
10554 Alan Hayward <alan.hayward@arm.com>
10555 David Sherwood <david.sherwood@arm.com>
10557 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
10558 (expand_debug_source_expr): Likewise.
10559 * combine.c (combine_simplify_rtx): Likewise.
10560 * cse.c (fold_rtx): Likewise.
10561 * fwprop.c (canonicalize_address): Likewise.
10562 * targhooks.c (default_shift_truncation_mask): Likewise.
10564 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
10565 Alan Hayward <alan.hayward@arm.com>
10566 David Sherwood <david.sherwood@arm.com>
10568 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
10569 (widened_mode): Likewise.
10570 (expand_unop): Likewise.
10571 * ree.c (transform_ifelse): Likewise.
10572 (merge_def_and_ext): Likewise.
10573 (combine_reaching_defs): Likewise.
10574 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10576 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
10577 Alan Hayward <alan.hayward@arm.com>
10578 David Sherwood <david.sherwood@arm.com>
10580 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
10581 * combine.c (gen_lowpart_for_combine): Likewise.
10582 * dwarf2out.c (rtl_for_decl_location): Likewise.
10583 * final.c (alter_subreg): Likewise.
10584 * rtlhooks.c (gen_lowpart_general): Likewise.
10585 (gen_lowpart_if_possible): Likewise.
10587 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
10588 Alan Hayward <alan.hayward@arm.com>
10589 David Sherwood <david.sherwood@arm.com>
10591 * calls.c (expand_call): Use subreg_lowpart_offset.
10592 * cse.c (cse_insn): Likewise.
10593 * regcprop.c (copy_value): Likewise.
10594 (copyprop_hardreg_forward_1): Likewise.
10596 2017-10-13 Jakub Jelinek <jakub@redhat.com>
10599 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
10600 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
10601 =Q constraints to +Q and into insn condition add check
10602 that operands[0] and operands[1] are equal.
10603 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
10604 =Q constraints to +Q and into insn condition add check
10605 that operands[0] is equal to either operands[1] or operands[2].
10608 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
10609 instead of handling MINUS_EXPR twice (once for each argument),
10610 canonicalize operand order and handle just once, use rtype where
10611 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
10614 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
10615 any values of __C while still being pattern recognizable as a simple
10616 rotate instruction.
10618 2017-10-13 Richard Biener <rguenther@suse.de>
10620 PR tree-optimization/82451
10622 2017-10-02 Richard Biener <rguenther@suse.de>
10624 PR tree-optimization/82355
10625 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
10626 a mapping for the enclosing loop but avoid generating one for
10627 the loop tree root.
10628 (copy_bb_and_scalar_dependences): Remove premature codegen
10629 error on PHIs in blocks duplicated into multiple places.
10630 * graphite-scop-detection.c
10631 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
10632 in the region use it as loop and nest to analyze the DR in.
10633 (try_generate_gimple_bb): Likewise.
10634 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
10635 (add_loop_constraints): For blocks in a loop not in the region
10636 create a dimension with a single iteration.
10637 * sese.h (gbb_loop_at_index): Remove assert.
10639 * cfgloop.c (loop_preheader_edge): For the loop tree root
10640 return the single successor of the entry block.
10641 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
10642 Reset the SCEV hashtable and niters.
10643 * graphite-scop-detection.c
10644 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
10645 assert that we only have POLYNOMIAL_CHREC that vary in loops
10646 contained in the region.
10647 (scop_detection::graphite_can_represent_expr): Adjust.
10648 (scop_detection::stmt_has_simple_data_refs_p): For loops
10649 not in the region set loop to NULL. The nest is now the
10650 entry edge to the region.
10651 (try_generate_gimple_bb): Likewise.
10652 * sese.c (scalar_evolution_in_region): Adjust for
10653 instantiate_scev change.
10654 * tree-data-ref.h (graphite_find_data_references_in_stmt):
10655 Make nest parameter the edge into the region.
10656 (create_data_ref): Likewise.
10657 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
10658 entry edge into a region and adjust instantiate_scev calls.
10659 (create_data_ref): Likewise.
10660 (graphite_find_data_references_in_stmt): Likewise.
10661 (find_data_references_in_stmt): Pass the loop preheader edge
10662 from the nest argument.
10663 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
10664 parameter the edge into the region.
10665 (instantiate_parameters): Use the loop preheader edge as entry.
10666 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
10668 (get_instantiated_value_entry): Make instantiate_below parameter
10669 the edge into the region.
10670 (instantiate_scev_name): Likewise. Adjust dominance checks,
10671 when we cannot use loop-based instantiation instantiate by
10672 walking use-def chains.
10673 (instantiate_scev_poly): Adjust.
10674 (instantiate_scev_binary): Likewise.
10675 (instantiate_scev_convert): Likewise.
10676 (instantiate_scev_not): Likewise.
10677 (instantiate_array_ref): Remove.
10678 (instantiate_scev_3): Likewise.
10679 (instantiate_scev_2): Likewise.
10680 (instantiate_scev_1): Likewise.
10681 (instantiate_scev_r): Do not blindly handle N-operand trees.
10682 Do not instantiate array-refs. Handle all constants and invariants.
10683 (instantiate_scev): Make instantiate_below parameter
10684 the edge into the region.
10685 (resolve_mixers): Use the loop preheader edge for the region
10686 parameter to instantiate_scev_r.
10687 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
10689 2017-10-13 Richard Biener <rguenther@suse.de>
10691 PR tree-optimization/82525
10692 * graphite-isl-ast-to-gimple.c
10693 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
10695 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
10696 Fail code generation when we cannot represent the isl integer.
10697 (binary_op_to_tree): Elide modulo operations that are no-ops
10698 in the type we code generate. Remove now superfluous code
10701 2017-10-13 Richard Biener <rguenther@suse.de>
10703 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
10704 (scop_detection::harmful_loop_in_region): Remove premature
10705 IV type restriction.
10706 (scop_detection::graphite_can_represent_scev): We can handle
10707 pointer IVs just fine.
10709 2017-10-13 Alan Modra <amodra@gmail.com>
10711 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
10712 "Clobbers and Scratch Registers". Add paragraph on
10713 alternative to clobbers for scratch registers and OpenBLAS
10716 2017-10-13 Alan Modra <amodra@gmail.com>
10718 * doc/extend.texi (Clobbers): Correct vax example. Delete old
10719 example of a memory input for a string of known length. Move
10720 commentary out of table. Add a number of new examples
10721 covering array memory inputs.
10723 2017-10-12 Martin Liska <mliska@suse.cz>
10725 PR tree-optimization/82493
10726 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
10727 (test_range_functions): New function.
10728 (sbitmap_c_tests): Likewise.
10729 * selftest-run-tests.c (selftest::run_tests): Run new tests.
10730 * selftest.h (sbitmap_c_tests): New function.
10732 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
10734 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
10736 * config/rs6000/amo.h: Fix spacing issue.
10738 2017-10-12 Jakub Jelinek <jakub@redhat.com>
10741 * config/i386/i386.md (*ashl<mode>3_mask_1,
10742 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
10743 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
10746 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
10748 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
10749 (profile_probability): Set max_probability
10750 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
10752 * profile-count.c (profile_probability::differs_from_p): Do not
10753 rely on max_probaiblity == 10000
10755 2017-10-12 Jeff Law <law@redhat.com>
10757 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
10760 2017-10-12 Martin Sebor <msebor@redhat.com>
10764 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
10765 (handle_alias_pairs): Call it.
10766 * common.opt (-Wattribute-alias): New option.
10767 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
10768 * doc/invoke.texi (-Wattribute-alias): Document.
10770 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
10773 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
10775 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
10777 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
10778 (make_hard_regno_born, make_hard_regno_dead): Update
10779 bb_killed_pseudos and bb_gen_pseudos.
10781 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
10783 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
10785 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
10787 * config/alpha/alpha.c (alpha_split_conditional_move):
10788 Use std::swap instead of manually swapping.
10789 (alpha_stdarg_optimize_hook): Ditto.
10790 (alpha_canonicalize_comparison): Ditto.
10792 2017-10-12 Bin Cheng <bin.cheng@arm.com>
10794 * tree-loop-distribution.c (struct builtin_info): New struct.
10795 (struct partition): Refactor fields into struct builtin_info.
10796 (partition_free): Free struct builtin_info.
10797 (build_size_arg_loc, build_addr_arg_loc): Delete.
10798 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
10799 information from struct builtin_info.
10800 (find_single_drs): New function refactored from classify_partition.
10801 Also moved builtin validity checks to this function.
10802 (compute_access_range, alloc_builtin): New functions.
10803 (classify_builtin_st, classify_builtin_ldst): New functions.
10804 (classify_partition): Refactor code into functions find_single_drs,
10805 classify_builtin_st and classify_builtin_ldst.
10806 (distribute_loop): Don't do runtime alias check when distributing
10808 (find_seed_stmts_for_distribution): New function.
10809 (pass_loop_distribution::execute): Refactor code finding seed
10810 stmts into above function. Support distribution for the innermost
10811 two-level loop nest. Adjust dump information.
10813 2017-10-12 Bin Cheng <bin.cheng@arm.com>
10815 * tree-loop-distribution.c: Adjust the general comment.
10816 (NUM_PARTITION_THRESHOLD): New macro.
10817 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
10818 (classify_partition): Skip builtin pattern of loop nest's inner loop.
10819 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
10820 in call to build_partition_graph.
10821 (finalize_partitions): New parameter. Make loop distribution more
10822 conservative by fusing more partitions.
10823 (distribute_loop): Don't do runtime alias check in case of loop nest
10825 (find_seed_stmts_for_distribution): New function.
10826 (prepare_perfect_loop_nest): New function.
10827 (pass_loop_distribution::execute): Refactor code finding seed stmts
10828 and loop nest into above functions. Support loop nest distribution.
10829 Adjust dump information accordingly.
10831 2017-10-12 Bin Cheng <bin.cheng@arm.com>
10833 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
10834 and set PTYPE_SEQUENTIAL for merged partition.
10836 2017-10-12 Richard Biener <rguenther@suse.de>
10838 PR tree-optimization/69728
10840 2017-09-19 Richard Biener <rguenther@suse.de>
10842 PR tree-optimization/69728
10843 * graphite-sese-to-poly.c (schedule_error): New global.
10844 (add_loop_schedule): Handle empty domain by failing the
10846 (build_original_schedule): Handle schedule_error.
10848 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
10849 domain by returning an unchanged schedule.
10851 2017-10-12 Jakub Jelinek <jakub@redhat.com>
10853 * genrecog.c (validate_pattern): For VEC_SELECT verify that
10854 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
10856 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
10858 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
10861 2017-10-12 Jakub Jelinek <jakub@redhat.com>
10864 * expr.c (store_field): Don't optimize away bitsize == 0 store
10865 from CALL_EXPR with addressable return type.
10867 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
10869 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
10870 * config/rs6000/rs6000.md (sel): Delete mode attribute.
10871 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
10872 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
10873 TARGET_ISEL instead of TARGET_ISEL<sel>.
10875 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
10877 * config/rs6000/rs6000.c
10878 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
10880 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
10882 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
10884 (reg_or_cint_operand): Fix comment.
10885 (reg_or_zero_operand): New predicate.
10886 * config/rs6000/rs6000-protos.h (output_isel): Delete.
10887 * config/rs6000/rs6000.c (output_isel): Delete.
10888 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
10889 instead of reg_or_cint_operand. Output instruction directly (not via
10891 (isel_unsigned_<mode>): Ditto.
10892 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
10893 gpc_reg_operand. Add an instruction alternative for this. Output
10894 instruction directly.
10895 (*isel_reversed_unsigned_<mode>): Ditto.
10897 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
10899 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
10900 (TARGET_CANONICALIZE_COMPARISON): Define.
10902 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
10905 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
10906 Check whether the dest is REG before adding REG_EQUIV note.
10908 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
10911 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
10913 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
10914 (make_hard_regno_born, make_hard_regno_dead): Update
10915 bb_killed_pseudos and bb_gen_pseudos.
10917 2017-10-11 Nathan Sidwell <nathan@acm.org>
10919 * incpath.h (enum incpath_kind): Name enum, prefix values.
10920 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
10921 * incpath.c (heads, tails): Use INC_MAX.
10922 (add_env_var_paths, add_standard_paths): Use incpath_kind.
10923 (merge_include_chains, split_quote_chain,
10924 register_include_chains): Update incpath_kind names.
10925 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
10926 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
10928 (add_framework_path, darwin_register_objc_includes): Likewise.
10929 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
10931 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
10933 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
10934 Do not use float_operator operator predicate.
10935 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
10936 * config/i386/predicates.md (float_operator): Remove predicate.
10938 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
10940 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
10941 (*jccxf_i387): Ditto.
10942 (*jcc<mode>_i387): Ditto.
10943 (*jccu<mode>_i387): Ditto.
10944 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
10945 (*jcc_*_i387 splitters): Remove.
10946 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
10947 * config/i386/i386.c (ix86_split_fp_branch): Remove.
10948 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
10951 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
10953 * profile-count.h (slow_safe_scale_64bit): New function.
10954 (safe_scale_64bit): New inline.
10955 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
10956 * profile-count.c: Include wide-int.h
10957 (slow_safe_scale_64bit): New.
10959 2017-10-11 Nathan Sidwell <nathan@acm.org>
10961 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
10962 HAS_DECL_ASSEMBLER_NAME_P.
10963 * gimple-expr.c (gimple_decl_printable_name: Check
10964 HAS_DECL_ASSEMBLER_NAME_P too.
10965 * ipa-utils.h (type_in_anonymous_namespace_p): Check
10966 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
10967 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
10968 * passes.c (rest_of_decl_compilation): Check
10969 HAS_DECL_ASSEMBLER_NAME_P too.
10970 * recog.c (verify_changes): Likewise.
10971 * tree-pretty-print.c (dump_decl_name): Likewise.
10972 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
10974 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
10975 (SET_DECL_ASSEMBLER_NAME): Use it.
10976 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
10977 (COPY_DECL_ASSEMBLER_NAME): Likewise.
10978 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
10980 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
10982 * config.gcc (i386, x86_64): Add extra objects.
10983 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
10984 (ix86_min_insn_size): Declare.
10985 (ix86_issue_rate): Declare.
10986 (ix86_adjust_cost): Declare.
10987 (ia32_multipass_dfa_lookahead): Declare.
10988 (ix86_macro_fusion_p): Declare.
10989 (ix86_macro_fusion_pair_p): Declare.
10990 (ix86_bd_has_dispatch): Declare.
10991 (ix86_bd_do_dispatch): Declare.
10992 (ix86_core2i7_init_hooks): Declare.
10993 (ix86_atom_sched_reorder): Declare.
10994 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
10995 (COSTS_N_BYTES): Move to x86-tune-costs.h.
10996 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
10997 (rip_relative_addr_p): Rename to ...
10998 (ix86_rip_relative_addr_p): ... this one; export.
10999 (memory_address_length): Update.
11000 (ix86_issue_rate): Move to x86-tune-sched.c.
11001 (ix86_flags_dependent): Move to x86-tune-sched.c.
11002 (ix86_agi_dependent): Move to x86-tune-sched.c.
11003 (exact_dependency_1): Move to x86-tune-sched.c.
11004 (exact_store_load_dependency): Move to x86-tune-sched.c.
11005 (ix86_adjust_cost): Move to x86-tune-sched.c.
11006 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
11007 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
11008 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
11009 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
11010 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
11011 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
11012 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
11013 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
11014 (min_insn_size): Rename to ...
11015 (ix86_min_insn_size): ... this one; export.
11016 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
11017 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
11018 (core2i7_first_cycle_multipass_backtrack): Move to
11019 x86-tune-sched-core.c.
11020 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
11021 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
11022 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
11023 (ix86_avoid_jump_mispredicts): Update.
11024 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
11025 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
11026 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
11027 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
11028 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
11029 (MAX_INSN): Move to ix86-tune-sched-bd.c.
11030 (MAX_IMM): Move to ix86-tune-sched-bd.c.
11031 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
11032 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
11033 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
11034 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
11035 (MAX_STORE): Move to ix86-tune-sched-bd.c.
11036 (BIG): Move to ix86-tune-sched-bd.c.
11037 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
11038 (enum insn_path): Move to ix86-tune-sched-bd.c.
11039 (get_mem_group): Move to ix86-tune-sched-bd.c.
11040 (is_cmp): Move to ix86-tune-sched-bd.c.
11041 (dispatch_violation): Move to ix86-tune-sched-bd.c.
11042 (is_branch): Move to ix86-tune-sched-bd.c.
11043 (is_prefetch): Move to ix86-tune-sched-bd.c.
11044 (init_window): Move to ix86-tune-sched-bd.c.
11045 (allocate_window): Move to ix86-tune-sched-bd.c.
11046 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
11047 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
11048 (process_end_window): Move to ix86-tune-sched-bd.c.
11049 (allocate_next_window): Move to ix86-tune-sched-bd.c.
11050 (find_constant): Move to ix86-tune-sched-bd.c.
11051 (get_num_immediates): Move to ix86-tune-sched-bd.c.
11052 (has_immediate): Move to ix86-tune-sched-bd.c.
11053 (get_insn_path): Move to ix86-tune-sched-bd.c.
11054 (get_insn_group): Move to ix86-tune-sched-bd.c.
11055 (count_num_restricted): Move to ix86-tune-sched-bd.c.
11056 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
11057 (add_insn_window): Move to ix86-tune-sched-bd.c.
11058 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
11059 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
11060 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
11061 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
11062 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
11063 (do_dispatch): Move to ix86-tune-sched-bd.c.
11064 (has_dispatch): Move to ix86-tune-sched-bd.c.
11065 * config/i386/t-i386: Add new object files.
11066 * config/i386/x86-tune-costs.h: New file.
11067 * config/i386/x86-tune-sched-atom.c: New file.
11068 * config/i386/x86-tune-sched-bd.c: New file.
11069 * config/i386/x86-tune-sched-core.c: New file.
11070 * config/i386/x86-tune-sched.c: New file.
11072 2017-10-11 Liu Hao <lh_mouse@126.com>
11074 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
11075 the generic version below instead.
11076 (should_colorize): Recognize Windows consoles as terminals
11078 * pretty-print.c [__MINGW32__] (write_all): New function.
11079 [__MINGW32__] (find_esc_head): Likewise.
11080 [__MINGW32__] (find_esc_terminator): Likewise.
11081 [__MINGW32__] (eat_esc_sequence): Likewise.
11082 [__MINGW32__] (mingw_ansi_fputs): New function that handles
11084 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
11087 2017-10-11 Richard Biener <rguenther@suse.de>
11089 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
11090 Properly call analyze_scalar_evolution with the loop of the stmt.
11092 2017-10-11 Richard Biener <rguenther@suse.de>
11094 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
11095 * tree-core.h (tree_base): Add chrec_var union member.
11096 * tree.h (CHREC_VAR): Remove.
11097 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
11098 * tree-chrec.h (build_polynomial_chrec): Adjust.
11099 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
11100 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
11102 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
11104 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
11105 * match.pd: ... here.
11106 ((T) X == (T) Y): Relax condition.
11108 2017-10-11 Bin Cheng <bin.cheng@arm.com>
11110 PR tree-optimization/82472
11111 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
11113 (break_alias_scc_partitions): Update postorder number.
11115 2017-10-11 Martin Liska <mliska@suse.cz>
11118 * opts.c (parse_no_sanitize_attribute): Do not use error_value
11120 * opts.h (parse_no_sanitize_attribute): Remove last argument.
11122 2017-10-11 Martin Liska <mliska@suse.cz>
11124 * print-rtl.c (print_insn): Move declaration of idbuf
11125 to same scope as name.
11127 2017-10-11 Martin Liska <mliska@suse.cz>
11132 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
11133 volatile arguments.
11135 2017-10-11 Martin Liska <mliska@suse.cz>
11138 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
11139 volatile arguments.
11141 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
11143 * config.gcc (default_gnu_indirect_function): Default to yes for
11144 arm*-*-linux* with glibc.
11146 2017-10-11 Richard Biener <rguenther@suse.de>
11148 * tree-scalar-evolution.c (get_scalar_evolution): Handle
11149 default-defs and types we do not want to analyze.
11150 (interpret_loop_phi): Replace unreachable code with an assert.
11151 (compute_scalar_evolution_in_loop): Remove and inline ...
11152 (analyze_scalar_evolution_1): ... here, replacing condition with
11153 what makes the intent clearer. Remove handling of cases
11154 get_scalar_evolution now handles.
11156 2017-10-10 Jim Wilson <wilson@tuliptree.org>
11158 PR rtl-optimization/81434
11159 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
11160 comment for main loop. In sched_group_found if, also add checks for
11161 pass and min_cost_group.
11163 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
11165 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
11166 (rs6000_insn_cost): New function.
11167 * config/rs6000/rs6000.md (cost): New attribute.
11169 2017-10-10 Jakub Jelinek <jakub@redhat.com>
11170 H.J. Lu <hongjiu.lu@intel.com>
11174 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
11175 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
11176 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
11177 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
11178 to OPTION_MASK_ISA_AVX512VL - builtins that have both
11179 OPTION_MASK_ISA_MMX and some other bit set require both
11180 mmx and the ISAs without the mmx bit.
11181 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
11182 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
11183 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
11184 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
11185 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
11186 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
11187 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
11188 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
11189 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
11190 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
11191 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
11192 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
11193 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
11194 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
11195 Add OPTION_MASK_ISA_MMX.
11197 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
11199 * config.gcc (armv7*-*-freebsd*): New target.
11200 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
11202 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
11204 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
11205 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
11208 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
11210 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
11211 a template parameter.
11212 (WIDE_INT_REF_FOR): Update accordingly.
11213 * tree.h (wi::int_traits <const_tree>): Delete.
11214 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
11215 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
11216 (wi::tree_to_wide_ref): New typedef.
11217 (wi::to_wide): New function.
11218 * calls.c (get_size_range): Use wi::to_wide when operating on
11219 trees as wide_ints.
11220 * cgraph.c (cgraph_node::create_thunk): Likewise.
11221 * config/i386/i386.c (ix86_data_alignment): Likewise.
11222 (ix86_local_alignment): Likewise.
11223 * dbxout.c (stabstr_O): Likewise.
11224 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
11225 * expr.c (const_vector_from_tree): Likewise.
11226 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
11227 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
11228 (fold_negate_expr_1, int_const_binop_1, const_binop)
11229 (fold_convert_const_int_from_real, optimize_bit_field_compare)
11230 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
11231 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
11232 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
11233 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
11234 (fold_not_const, round_up_loc): Likewise.
11235 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
11236 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
11237 (alloca_call_type): Likewise.
11238 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
11239 * godump.c (go_output_typedef): Likewise.
11240 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
11241 * internal-fn.c (get_min_precision): Likewise.
11242 * ipa-cp.c (ipcp_store_vr_results): Likewise.
11243 * ipa-polymorphic-call.c
11244 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
11245 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
11246 (ipa_modify_call_arguments): Likewise.
11247 * match.pd: Likewise.
11248 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
11249 * print-tree.c (print_node_brief, print_node): Likewise.
11250 * stmt.c (expand_case): Likewise.
11251 * stor-layout.c (layout_type): Likewise.
11252 * tree-affine.c (tree_to_aff_combination): Likewise.
11253 * tree-cfg.c (group_case_labels_stmt): Likewise.
11254 * tree-data-ref.c (dr_analyze_indices): Likewise.
11255 (prune_runtime_alias_test_list): Likewise.
11256 * tree-dump.c (dequeue_and_dump): Likewise.
11257 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
11258 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
11259 * tree-pretty-print.c (dump_generic_node): Likewise.
11260 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
11261 (simple_iv_with_niters): Likewise.
11262 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
11263 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
11264 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
11265 * tree-ssa-loop-niter.c (split_to_var_and_offset)
11266 (refine_value_range_using_guard, number_of_iterations_ne_max)
11267 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
11268 (get_cst_init_from_scev, record_nonwrapping_iv)
11269 (scev_var_range_cant_overflow): Likewise.
11270 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
11271 * tree-ssa-pre.c (compute_avail): Likewise.
11272 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
11273 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
11274 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
11275 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
11276 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
11277 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
11278 (dump_case_nodes, try_switch_expansion): Likewise.
11279 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
11280 (vect_do_peeling): Likewise.
11281 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
11282 * tree-vect-stmts.c (vectorizable_load): Likewise.
11283 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
11284 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
11285 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
11286 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
11287 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
11288 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
11289 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
11290 (evrp_dom_walker::before_dom_children): Likewise.
11291 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
11292 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
11293 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
11294 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
11295 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
11296 (get_range_pos_neg): Likewise.
11297 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
11298 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
11299 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
11300 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
11301 * config/avr/avr.c (avr_fold_builtin): Likewise.
11302 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
11303 * config/msp430/msp430.c (msp430_attr): Likewise.
11304 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
11305 * config/powerpcspe/powerpcspe-c.c
11306 (altivec_resolve_overloaded_builtin): Likewise.
11307 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
11308 (rs6000_expand_ternop_builtin): Likewise.
11309 * config/rs6000/rs6000-c.c
11310 (altivec_resolve_overloaded_builtin): Likewise.
11311 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
11312 (rs6000_expand_ternop_builtin): Likewise.
11313 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
11315 2017-10-10 Bin Cheng <bin.cheng@arm.com>
11317 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
11318 when copying loop nest with only one inner loop.
11320 2017-10-10 Richard Biener <rguenther@suse.de>
11322 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
11323 blocks if SCEV is active.
11324 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
11326 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
11327 (scev_initialize): Assert we are not yet initialized.
11329 2017-10-10 Bin Cheng <bin.cheng@arm.com>
11331 * tree-loop-distribution.c (generate_loops_for_partition): Remove
11332 inner loop's exit stmt by making it always exit the loop, otherwise
11333 we would generate an infinite empty loop.
11335 2017-10-10 Bin Cheng <bin.cheng@arm.com>
11337 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
11338 renaming variables in new preheader if it's deleted.
11340 2017-10-10 Bin Cheng <bin.cheng@arm.com>
11342 * tree-loop-distribution.c (struct partition): Remove unused field
11343 loops of the structure.
11344 (partition_alloc, partition_free): Ditto.
11345 (build_rdg_partition_for_vertex): Ditto.
11347 2017-10-09 Jeff Law <law@redhat.com>
11349 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
11350 return type to match prototype and documentation.
11352 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
11354 * config/rs6000/rs6000.c (processor_costs): Move to ...
11355 * config/rs6000/rs6000.h: ... here.
11356 (rs6000_cost): Declare.
11358 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
11360 * except.c (setjmp_fn): New global variable.
11361 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
11362 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
11363 if DONT_USE_BUILTIN_SETJMP is defined.
11365 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
11367 * target.def (insn_cost): New hook.
11368 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
11369 * doc/tm.texi: Regenerate.
11370 * rtlanal.c (insn_cost): Use the new hook.
11372 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
11374 * combine.c (combine_validate_cost): Compute the new insn_cost,
11375 not just pattern_cost.
11376 (try_combine): Adjust comment.
11378 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
11380 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
11382 * combine.c (uid_insn_cost): Adjust comment.
11383 (combine_validate_cost): Adjust comment. Use pattern_cost instead
11385 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
11386 * dse.c (find_shift_sequence): Ditto.
11387 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
11388 (bb_valid_for_noce_process_p): Use pattern_cost.
11389 * rtl.h (insn_rtx_cost): Delete.
11390 (pattern_cost): New prototype.
11391 (insn_cost): New prototype.
11392 * rtlanal.c (insn_rtx_cost): Rename to...
11393 (pattern_cost): ... this.
11396 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
11398 * config/i386/i386.md (*jcc_2): Remove insn pattern.
11399 (*jcc<mode>_0_r_i387): Ditto.
11400 (*jccxf_r_i387): Ditto.
11401 (*jcc<mode>_r_i387): Ditto.
11402 (*jccu<mode>_r_i387): Ditto.
11403 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
11404 (*jcc): Rename from *jcc_1.
11406 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11408 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
11409 deferred rescans after the lvx/stvx recombination pre-pass.
11411 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
11413 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
11414 memory operation instruction support.
11415 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
11416 (rs6000-ibm-aix[789]*): Likewise.
11417 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
11418 Document new functions.
11420 2017-10-09 Richard Biener <rguenther@suse.de>
11422 PR tree-optimization/82397
11423 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
11424 equality only for semantically equal trees.
11426 2017-10-09 Richard Biener <rguenther@suse.de>
11428 PR tree-optimization/82449
11429 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
11430 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
11431 allow constant addresses.
11432 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
11435 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11437 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
11440 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11443 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
11446 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11449 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
11451 2017-10-09 Jakub Jelinek <jakub@redhat.com>
11454 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
11455 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
11457 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
11459 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
11460 (WI_BINARY_PREDICATE_RESULT): Likewise.
11461 (wi::binary_traits::operator_result): New type.
11462 (wi::binary_traits::predicate_result): Likewise.
11463 (generic_wide_int::operator~, unary generic_wide_int::operator-)
11464 (generic_wide_int::operator==, generic_wide_int::operator!=)
11465 (generic_wide_int::operator&, generic_wide_int::and_not)
11466 (generic_wide_int::operator|, generic_wide_int::or_not)
11467 (generic_wide_int::operator^, generic_wide_int::operator+
11468 (binary generic_wide_int::operator-, generic_wide_int::operator*):
11470 (operator~, unary operator-, operator==, operator!=, operator&)
11471 (operator|, operator^, operator+, binary operator-, operator*): New
11473 * expr.c (get_inner_reference): Use wi::bit_and_not.
11474 * fold-const.c (fold_binary_loc): Likewise.
11475 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
11476 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
11477 (bit_value_binop): Likewise.
11478 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
11479 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
11480 (extract_range_from_binary_expr_1): Likewise.
11481 (masked_increment): Likewise.
11482 (simplify_bit_ops_using_ranges): Likewise.
11484 2017-10-09 Martin Jambor <mjambor@suse.cz>
11487 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
11488 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
11489 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
11491 (hsa_fixup_mov_insn_type): New function.
11492 (hsa_op_with_type::get_in_type): Use it.
11493 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
11494 immediates in an assert.
11495 (hsa_op_with_type::extend_int_to_32bit): New method.
11496 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
11497 types. Convert to dest type if necessary.
11498 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
11499 (reg_for_gimple_ssa): Pass false as min32int to
11500 hsa_type_for_scalar_tree_type.
11501 (gen_hsa_addr): Fixup type when creating addresable temporary.
11502 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
11503 (gen_hsa_unary_operation): Extend operands and convert to dest type if
11504 necessary. Call hsa_fixup_mov_insn_type.
11505 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
11506 extend operands and convert to dest type if necessary.
11507 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
11508 to dest type if necessary.
11509 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
11510 if conversion nt necessary and size matches.
11511 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
11512 to dest type if necessary.
11513 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
11514 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
11516 (gen_hsa_clrsb): Likewise.
11517 (gen_hsa_ffs): Likewise.
11518 (gen_hsa_divmod): Extend operands and convert to dest type if
11520 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
11522 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
11524 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
11525 Remove empty default arguments. Use a brace block as output
11527 (conditional return): Ditto.
11529 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
11530 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
11532 (group_ending_nop): Ditto.
11533 (doloop_end): Ditto.
11534 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
11535 (splitters for those): Ditto.
11537 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
11539 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
11540 a conditional jump (and the compare for it) so that pc_rtx is the
11542 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
11543 for the deleted and renamed ctr<mode>_internal[234] patterns.
11544 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
11545 Delete second conditional return pattern.
11546 (ctr<mode>_internal2): Delete this second bdnz pattern.
11547 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
11548 (ctr<mode>_internal4): Delete this second bdz pattern.
11550 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
11552 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
11553 (always_initialized_rtx_for_ssa_name_p): New predicate.
11554 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
11555 (finish_out_of_ssa): Free new field of SA.
11556 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
11557 * tree-ssa-coalesce.c: Include tree-ssa.h.
11558 (get_parm_default_def_partitions): Remove extern keyword.
11559 (get_undefined_value_partitions): New function.
11560 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
11561 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
11562 uninitialized bits.
11563 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
11565 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
11567 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
11569 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
11571 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
11572 for targets that preffer 128bit.
11574 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
11576 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
11578 2017-10-08 Olivier Hainque <hainque@adacore.com>
11580 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
11581 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
11583 (thumb_set_return_address): Likewise.
11585 2017-10-08 Olivier Hainque <hainque@adacore.com>
11587 * common/config/arm/arm-common.c (arm_except_unwind_info):
11588 Handle DWARF2_UNWIND_INFO.
11590 2017-10-07 Michael Collison <michael.collison@arm.com>
11592 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
11595 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
11597 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
11598 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
11599 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
11600 defined, force the creation of a new block for a dispatch label.
11602 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
11604 * invoke.texi (Wsuggest-attribute=cold): Document.
11605 * common.opt (Wsuggest-attribute=cold): New
11606 * ipa-pure-const.c (warn_function_cold): New function.
11607 * predict.c (compute_function_frequency): Use it.
11608 * predict.h (warn_function_cold): Declare.
11610 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
11612 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
11615 2017-10-06 Martin Liska <mliska@suse.cz>
11617 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
11618 keyword for member functions.
11619 (struct sanopt_tree_couple): New struct.
11620 (struct sanopt_tree_couple_hash): New function.
11621 (struct sanopt_ctx): Add new hash_map.
11622 (has_dominating_ubsan_ptr_check): New function.
11623 (record_ubsan_ptr_check_stmt): Likewise.
11624 (maybe_optimize_ubsan_ptr_ifn): Likewise.
11625 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
11626 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
11628 2017-10-06 Sudakshina Das <sudi.das@arm.com>
11631 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
11632 aarch64_simd_valid_immediate on CONST_VECTORs.
11633 (aarch64_reg_or_bic_imm): Likewise.
11635 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
11637 PR rtl-optimization/82396
11638 * haifa-sched.c (ready_sort_real): Disable qsort checking.
11640 2017-10-06 Sebastian Pop <sebpop@gmail.com>
11642 * graphite-dependences.c (scop_get_reads): Move code to...
11643 (scop_get_must_writes): Move code to...
11644 (scop_get_may_writes): Move code to...
11645 (scop_get_reads_and_writes): ... here.
11646 (scop_get_dependences): Call scop_get_reads_and_writes.
11648 2017-10-06 Jakub Jelinek <jakub@redhat.com>
11650 PR tree-optimization/82434
11651 * fold-const.h (can_native_encode_type_p,
11652 can_native_encode_string_p): Remove.
11653 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
11654 don't encode anything, just return what would be otherwise returned.
11655 (native_encode_fixed, native_encode_complex, native_encode_vector):
11657 (native_encode_string): Likewise. Inline by hand
11658 can_native_encode_string_p.
11659 (can_native_encode_type_p): Remove.
11660 (can_native_encode_string_p): Remove.
11661 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
11662 STRING_CSTs using can_native_encode_string_p, test all
11663 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
11664 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
11665 argument from native_encode_expr.
11666 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
11667 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
11670 2017-10-06 Richard Biener <rguenther@suse.de>
11672 PR tree-optimization/82397
11673 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
11674 operand_equal_p but rely on data_ref_compare_tree for detecting
11676 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
11677 to match up with dr_group_sort_cmp.
11679 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11682 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
11684 * config/s390/s390-builtin-types.def: Regenerate.
11686 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11689 * config/s390/s390-builtin-types.def: Regenerate.
11690 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
11691 Change flag from B_VXE to B_VX.
11692 (s390_vec_min_dbl): Remove B_VXE flag.
11694 2017-10-06 Richard Biener <rguenther@suse.de>
11696 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
11697 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
11698 translate_isl_ast_to_gimple::is_valid_rename,
11699 translate_isl_ast_to_gimple::get_rename,
11700 translate_isl_ast_to_gimple::get_def_bb_for_const,
11701 translate_isl_ast_to_gimple::get_new_name,
11702 translate_isl_ast_to_gimple::collect_all_ssa_names,
11703 translate_isl_ast_to_gimple::copy_loop_phi_args,
11704 translate_isl_ast_to_gimple::collect_all_ssa_names,
11705 translate_isl_ast_to_gimple::copy_loop_phi_args,
11706 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
11707 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
11708 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
11709 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
11710 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
11711 translate_isl_ast_to_gimple::copy_cond_phi_args,
11712 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
11713 translate_isl_ast_to_gimple::edge_for_new_close_phis,
11714 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
11715 translate_isl_ast_to_gimple::rename_uses,
11716 translate_isl_ast_to_gimple::rename_all_uses): Remove.
11717 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
11718 (set_rename_for_each_def): Likewise.
11719 (graphite_copy_stmts_from_block): Handle debug stmt resetting
11720 here. Handle rewriting SCEV analyzable uses here.
11721 (copy_bb_and_scalar_dependences): Generate code for PHI
11723 (graphite_regenerate_ast_isl): Adjust.
11724 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
11725 (add_write, add_read): New functions.
11726 (build_cross_bb_scalars_def): Use it and simplify.
11727 (build_cross_bb_scalars_use): Likewise.
11728 (graphite_find_cross_bb_scalar_vars): Inline into...
11729 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
11730 simulating out-of-SSA. Compute liveout and add dependencies.
11731 (build_scops): Force an empty entry block.
11732 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
11734 (sese_build_liveouts): Declare.
11735 (sese_trivially_empty_bb_p): Likewise.
11736 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
11737 compute liveout and debug_liveout.
11738 (sese_bad_liveouts_use): Remove.
11739 (sese_reset_debug_liveouts_bb): Likewise.
11740 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
11741 (sese_build_liveouts): Build liveout and debug_liveout and store
11743 (new_sese_info): Adjust.
11744 (free_sese_info): Likewise.
11745 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
11746 do not build liveout here.
11747 (move_sese_in_condition): Adjust region entry.
11748 (scev_analyzable_p): Match up with chrec_apply requirements.
11749 (sese_trivially_empty_bb_p): New.
11750 * tree-into-ssa.c (get_reaching_def): Properly support generating
11751 default-defs for incremental rewrite of anonymous names.
11753 2017-10-06 Richard Biener <rguenther@suse.de>
11755 * graphite-sese-to-poly.c (extract_affine): For casts increasing
11756 precision do not perform modulo reduction.
11758 2017-10-06 Richard Biener <rguenther@suse.de>
11760 PR tree-optimization/82436
11761 * tree-vect-slp.c (vect_supported_load_permutation_p): More
11762 conservatively choose the vectorization factor when checking
11763 whether we can perform the required load permutation.
11764 (vect_transform_slp_perm_load): Assert when we may not fail.
11766 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
11768 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
11769 message for incompatible -msdata=* and -mcall-* options.
11771 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
11773 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
11774 rate for post-reload scheduling.
11776 2017-10-05 Tamar Christina <tamar.christina@arm.com>
11778 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
11780 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
11782 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
11783 to improve monte carlo in scimark.
11785 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
11787 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
11788 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
11789 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
11790 pentium4_cost, nocona_cost): Set reassociation width to 1.
11791 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
11792 width to 2 for fp operations and 1 otherwise.
11793 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
11795 (atom_cost): Set reassociation width to 2.
11796 (slm_cost, generic_cost): Set fp reassociation width
11797 to 2 and 1 otherwise.
11798 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
11799 (core_cost): Set fp reassociation width to 4 and vector to 2.
11800 (ix86_reassociation_width): Rewrite using cost table; special case
11801 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
11802 and TARGET_AVX128_OPTIMAL.
11803 * config/i386/i386.h (processor_costs): Add
11804 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
11805 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
11806 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
11807 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
11808 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
11809 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
11811 2017-10-05 Nathan Sidwell <nathan@acm.org>
11813 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
11815 2017-10-05 Tamar Christina <tamar.christina@arm.com>
11817 * config/arm/arm.c (arm_test_fpu_data): New.
11818 (arm_run_selftests): Call arm_test_fpu_data.
11820 2017-10-04 Nathan Sidwell <nathan@acm.org>
11822 * toplev.c (toplev::main): Remove excess parens on pretty_printer
11824 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
11826 2017-10-04 Sudakshina Das <sudi.das@arm.com>
11828 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
11829 check type for aarch64_simd_valid_immediate.
11830 (aarch64_output_simd_mov_immediate): Update prototype.
11831 (aarch64_simd_valid_immediate): Update prototype.
11832 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
11833 support for ORR-immediate.
11834 (and<mode>3): modified pattern to add support for BIC-immediate.
11835 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
11836 now checks for valid immediate for BIC and ORR based on new enum
11838 (aarch64_output_simd_mov_immediate): Function now used to output
11839 BIC/ORR imm as well based on new enum argument.
11840 * config/aarch64/constraints.md (Do): New vector immediate constraint.
11842 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
11843 (aarch64_reg_or_bic_imm): Likewise.
11845 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11847 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
11848 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
11850 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
11854 PR rtl-optimization/82396
11855 * haifa-sched.c (autopref_multipass_init): Simplify
11857 (autopref_rank_data): Simplify sort order.
11858 * sched-int.h (autopref_multipass_data_): Remove
11859 multi_mem_insn_p, min_offset and max_offset.
11861 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11863 * doc/sourcebuild.texi: Document vect_peeling_profitable.
11865 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11867 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
11868 vect_doubleint_cvt.
11870 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11872 * doc/sourcebuild.texi: Document vect_long_mult.
11874 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
11876 PR tree-optimization/82413
11877 * fold-const.c (build_range_check): Use widest_int when comparing
11878 the maximum ETYPE value with HIGH.
11880 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
11882 PR rtl-optimization/82396
11883 * haifa-sched.c (autopref_multipass_init): Simplify
11885 (autopref_rank_data): Simplify sort order.
11886 * sched-int.h (autopref_multipass_data_): Remove
11887 multi_mem_insn_p, min_offset and max_offset.
11889 2017-10-04 Jakub Jelinek <jakub@redhat.com>
11891 PR tree-optimization/82381
11892 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
11893 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
11896 PR tree-optimization/82374
11897 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
11898 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
11899 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
11900 current_function_decl to the new decl.
11902 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
11904 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
11905 helper macro for IEEE float128 hardware built-in functions.
11906 (SQRTF128_ODD): Add built-in functions with the round-to-odd
11908 (TRUNCF128_ODD): Likewise.
11909 (ADDF128_ODD): Likewise.
11910 (SUBF128_ODD): Likewise.
11911 (MULF128_ODD): Likewise.
11912 (DIVF128_ODD): Likewise.
11913 (FMAF128_ODD): Likewise.
11914 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
11915 UNSPEC_TRUNC_ROUND_TO_ODD.
11916 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
11917 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
11918 floating point round to odd instructions.
11919 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
11920 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
11921 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
11922 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
11923 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
11924 (trunc<mode>sf2_hw): Change the truncate with round to odd
11925 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
11926 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
11927 to odd hardware instructions.
11928 (sub<mode>3_odd): Likewise.
11929 (mul<mode>3_odd): Likewise.
11930 (div<mode>3_odd): Likewise.
11931 (sqrt<mode>2_odd): Likewise.
11932 (fma<mode>4_odd): Likewise.
11933 (fms<mode>4_odd): Likewise.
11934 (nfma<mode>4_odd): Likewise.
11935 (nfms<mode>4_odd): Likewise.
11936 (trunc<mode>df2_odd): Change the truncate with round to odd
11937 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
11939 * doc/extend.texi (PowerPC built-in functions): Update documentation
11940 for existing IEEE float128-bit built-in functions. Add built-in
11941 functions that generate the IEEE 128-bit floating point round to
11944 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
11946 PR rtl-optimization/77729
11947 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
11948 to (X&(C1&~C2))|C2 transformations.
11950 2017-10-03 Martin Jambor <mjambor@suse.cz>
11952 PR tree-optimization/82363
11953 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
11954 mismatch, mark lacc written regardless of racc.
11956 2017-10-03 Jakub Jelinek <jakub@redhat.com>
11958 PR tree-optimization/82381
11959 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
11960 stmt_to_insert nor wheather SSA_NAMEs are default defs.
11961 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
11962 fallthrough into reassoc_stmt_dominates_stmt_p.
11965 * combine.c (combine_instructions): Don't combine in unreachable
11968 2017-10-02 Peter Bergner <bergner@vnet.ibm.com>
11971 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
11972 function to not use the have_cpu variable. Do not set cpu_index,
11973 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
11974 or the default cpu.
11975 (rs6000_valid_attribute_p): Remove duplicate initializations of
11976 old_optimize and func_optimize.
11977 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
11978 (rs6000_activate_target_options): Make global.
11979 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
11982 2017-10-02 Jakub Jelinek <jakub@redhat.com>
11984 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
11985 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
11986 Set *poffset to 0 and *psize and *pmax_size to -1 if
11987 *poffset + *psize overflows in HOST_WIDE_INT.
11989 PR tree-optimization/82387
11990 PR tree-optimization/82388
11991 PR tree-optimization/82389
11992 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
11993 instead of live_bytes non-NULL.
11995 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
11998 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
12001 2017-10-02 Richard Biener <rguenther@suse.de>
12003 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
12004 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
12005 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
12007 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
12009 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
12010 requested precision matches the type's.
12011 * calls.c (alloc_max_size): Calculate the new candidate size as
12012 a widest_int and use wi::to_widest when comparing it with the
12013 current candidate size.
12014 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
12015 zero rather than integer_zero_node.
12016 * match.pd: Check for a no-op conversion before using wi::add
12017 rather than after. Use tree_to_uhwi when summing small shift
12018 counts into an unsigned int.
12020 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
12021 Alan Hayward <alan.hayward@arm.com>
12022 David Sherwood <david.sherwood@arm.com>
12025 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
12026 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
12027 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
12028 POINTER_AND_FP_REGS.
12030 2017-10-02 Richard Biener <rguenther@suse.de>
12032 PR tree-optimization/82355
12033 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
12034 a mapping for the enclosing loop but avoid generating one for
12035 the loop tree root.
12036 (copy_bb_and_scalar_dependences): Remove premature codegen
12037 error on PHIs in blocks duplicated into multiple places.
12038 * graphite-scop-detection.c
12039 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
12040 in the region use it as loop and nest to analyze the DR in.
12041 (try_generate_gimple_bb): Likewise.
12042 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
12043 (add_loop_constraints): For blocks in a loop not in the region
12044 create a dimension with a single iteration.
12045 * sese.h (gbb_loop_at_index): Remove assert.
12047 2017-10-01 Kevin Buettner <kevinb@redhat.com>
12049 * omp-expand.c (adjust_context_scope): New function.
12050 (expand_parallel_call): Call adjust_context_scope.
12052 2017-10-01 Jeff Law <law@redhat.com>
12054 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
12055 dom_opt_dom_walker class with direct access to private members.
12056 Add comments. Call test_for_singularity.
12057 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
12058 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
12059 m_dummy_cond anymore.
12060 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
12062 (pass_dominator:execute): Build the dummy_cond here and pass it
12063 to the dom_opt_dom_walker ctor.
12064 (test_for_singularity): New function.
12066 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
12067 Maya Rashish <coypu@sdf.org>
12069 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
12070 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
12071 (alpha*-*-netbsd*) Use nbsd_tm_file.
12072 (arm*-*-netbsdelf*) Likewise.
12073 (i[34567]86-*-netbsdelf*) Likewise.
12074 (x86_64-*-netbsd*) Likewise.
12075 (mips*-*-netbsd*) Likewise.
12076 (powerpc-*-netbsd*) Likewise.
12077 (sh*-*-netbsd*) Likewise.
12078 (sparc-*-netbsdelf*) Likewise.
12079 (sparc64-*-netbsd*) Likewise.
12080 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
12082 (vax-*-netbsdelf*) Likewise.
12083 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
12084 (UINT_FAST8_TYPE) Likewise.
12085 (INT_FAST16_TYPE) Check CHAR_FAST16.
12086 (UINT_FAST16_TYPE) Likewise.
12088 2017-09-30 Jakub Jelinek <jakub@redhat.com>
12091 * config/i386/i386.md
12092 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
12093 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
12094 *divmodsi4_zext_2): New define_insn_and_split.
12095 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
12096 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
12097 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
12098 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
12099 New define_insn_and_split.
12100 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
12101 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
12102 operands[1] having DImode when mode is SImode.
12104 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
12105 always SImode for DIV and MOD in REG_EQUAL notes.
12107 2017-09-29 Yury Gribov <tetra2005@gmail.com>
12109 PR middle-end/82319
12110 * match.pd: Fix handling of NaNs in pattern.
12112 2017-09-29 Jeff Law <law@redhat.com>
12114 * sbitmap.c (bitmap_bit_in_range_p): New function.
12115 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
12116 * tree-ssa-dse.c (live_bytes_read): New function.
12117 (dse_classify_store): Ignore reads of dead bytes.
12119 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
12120 typos and whitespace errors.
12121 * config/i386/predicates.md (address_no_seg_operand): Likewise.
12122 * config/s390/s390.c (s390_emit_prologue): Likewise.
12124 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
12127 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
12128 with a symbol for LRA.
12130 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
12132 PR rtl-optimization/82338
12133 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
12135 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
12137 * genmodes.c (calc_wider_mode): Suppress qsort macro.
12138 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
12139 (qsort_chk): Declare.
12140 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
12141 (qsort_chk): New function.
12143 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12145 PR tree-optimization/82337
12146 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
12147 phi definition if the PHI result appears in an abnormal PHI.
12148 (find_basis_for_base_expr): Don't record a basis if the LHS of the
12149 basis appears in an abnormal PHI.
12151 2017-09-29 Richard Biener <rguenther@suse.de>
12153 * graphite-isl-ast-to-gimple.c
12154 (translate_isl_ast_to_gimple::set_codegen_error): New function.
12155 (binary_op_to_tree): Use it.
12156 (get_rename_from_scev): Likewise.
12157 (copy_loop_phi_nodes): Likewise.
12158 (copy_bb_and_scalar_dependences): Likewise.
12159 (translate_pending_phi_nodes): Likewise.
12161 2017-09-29 Jakub Jelinek <jakub@redhat.com>
12164 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
12165 for movabsq $(i32 << shift), r64.
12167 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
12169 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
12170 index when encoding %esp as %rsp to avoid 0x67 prefix.
12172 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
12174 * config/i386/i386.md (*movsf_internal, *movdf_internal):
12175 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
12177 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
12179 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
12180 Extensions with more than 16 double VFP registers.
12181 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
12182 to_clear_mask and all code related to it. Replace the remaining
12183 entry by a sbitmap and adapt code accordingly.
12185 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
12187 * brig-builtins.def: Change pure attributes to const.
12189 2017-09-28 Joseph Myers <joseph@codesourcery.com>
12191 * config.gcc (default_gnu_indirect_function): Default to yes for
12192 sparc*-*-linux* with glibc.
12194 2017-09-28 Joseph Myers <joseph@codesourcery.com>
12196 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
12197 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
12198 when creating .init_array and .fini_array sections with priority
12201 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
12204 * config/aarch64/aarch64.c
12205 (aarch64_builtin_support_vector_misalignment): Always return false
12206 when misalignment is unknown.
12208 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
12210 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
12211 this function to return false if the definition used by the swap
12212 instruction is artificial, or if the memory address from which the
12213 constant value is loaded is not represented by a base address held
12214 in a register or if the base address register is a frame or stack
12215 pointer. Additionally, return false if the base address of the
12216 loaded constant is a SYMBOL_REF but is not considered to be a
12218 (replace_swapped_load_constant): New function.
12219 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
12220 loaded constant vector with a load of a swapped constant vector.
12222 2017-09-27 Carl Love <cel@us.ibm.com>
12224 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
12225 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
12226 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
12229 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
12231 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
12232 first, always call autopref_rank_data otherwise.
12234 2017-09-27 Richard Biener <rguenther@suse.de>
12236 * graphite-scop-detection.c (find_scop_parameters): Move
12237 loop bound handling ...
12238 (gather_bbs::before_dom_children): ... here, avoiding the need
12239 to build scop_info->loop_nest.
12240 (record_loop_in_sese): Remove.
12241 * sese.h (sese_info_t::loop_nest): Remove.
12242 * sese.c (new_sese_info): Do not allocate loop_nest.
12243 (free_sese_info): Do not free loop_nest.
12245 2017-09-27 Jakub Jelinek <jakub@redhat.com>
12248 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
12249 lhs from calls if the lhs has addressable type.
12251 2017-09-27 Richard Biener <rguenther@suse.de>
12253 * graphite.h (scop::max_alias_set): New member.
12254 * graphite-scop-detection.c: Remove references to non-existing
12255 --param in comments.
12256 (build_alias_sets): Record the maximum alias set used for drs.
12257 (build_scops): Support zero as unlimited for
12258 --param graphite-max-arrays-per-scop.
12259 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
12260 and inline into ...
12261 (build_poly_sr_1): ... here. Compute alias set based on the
12262 maximum alias set used for drs rather than
12263 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
12265 2017-09-27 Richard Biener <rguenther@suse.de>
12267 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
12268 --param loop-block-tile-size=0 to disable tiling.
12270 2017-09-27 Richard Biener <rguenther@suse.de>
12272 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
12273 (graphite-max-nb-scop-params): Document special value zero.
12274 * domwalk.h (dom_walker::STOP): New symbolical constant.
12275 (dom_walker::dom_walker): Add optional parameter for bb to
12277 (dom_walker::~dom_walker): Declare.
12278 (dom_walker::before_dom_children): Document STOP return value.
12279 (dom_walker::m_user_bb_to_rpo): New member.
12280 (dom_walker::m_bb_to_rpo): Likewise.
12281 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
12282 mapping here if not provided by the user.
12283 (dom_walker::~dom_walker): Free bb to RPO mapping if not
12284 provided by the user.
12285 (dom_walker::STOP): Define.
12286 (dom_walker::walk): Do not compute bb to RPO mapping here.
12287 Support STOP return value from before_dom_children to stop
12289 * graphite-optimize-isl.c (optimize_isl): If the schedule
12290 is the same still generate code if -fgraphite-identity
12291 or -floop-parallelize-all are given.
12292 * graphite-scop-detection.c: Include cfganal.h.
12293 (gather_bbs::gather_bbs): Get and pass through bb to RPO
12295 (gather_bbs::before_dom_children): Return STOP for BBs
12297 (build_scops): Compute bb to RPO mapping and pass it to
12298 the domwalk. Treat --param graphite-max-nb-scop-params=0
12299 as not limiting the number of params.
12300 * graphite.c (graphite_initialize): Remove limit on the
12301 number of basic-blocks in a function.
12302 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
12303 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
12304 default value of 10.
12306 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
12308 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
12309 Adjust code to eliminate needing to do the shift right 32-bits
12310 operation after XSCVDPSPN.
12312 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
12314 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
12315 ((X / Y) != 0 -> X >= Y): Likewise.
12317 2017-09-26 Carl Love <cel@us.ibm.com>
12319 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
12320 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
12321 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
12322 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
12323 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
12324 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
12325 definitions and overloading.
12326 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
12327 statement for P9V_BUILTIN_XST_LEN_R.
12328 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
12329 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
12330 define_expand and define_insn for the instructions and builtins.
12331 * doc/extend.texi: Update the built-in documentation file for the new
12332 built-in functions.
12333 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
12334 define_insn for the instructions
12336 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
12339 * gcc/config/netbsd-protos.h: New file.
12340 * gcc/config/netbsd.c: New file.
12341 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
12342 * gcc/config/t-netbsd: New file.
12343 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
12344 (tmake_file) Add t-netbsd.
12345 (extra_objs) Add netbsd.o.
12347 2017-09-26 Janus Weil <janus@gcc.gnu.org>
12351 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
12353 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
12355 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
12356 sign extension from a vector register to a GPR by doing a 32-bit
12357 direct move and then an EXTSW.
12358 (extendsi<mode>2 splitter): Likewise.
12359 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
12360 right or vector extract after doing XSCVDPSPN. Use
12361 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
12363 (movdi_from_sf_zero_ext): Likewise.
12364 (reload_gpr_from_vsxsf): Likewise.
12365 (p8_mfvsrd_4_disf): Delete, no longer used.
12366 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
12367 then needing to move the SFmode to a GPR to use the XSCVDPSP
12368 instruction instead of FRSP and XSCVDPSPN.
12369 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
12370 it is adjacent to the other XSCVSPDP insns.
12371 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
12372 SFmode to be in traditional Altivec registers.
12373 (vsx_xscvdpspn): Eliminate useless alternative constraint.
12374 (vsx_xscvspdpn): Likewise.
12375 (vsx_xscvspdpn_scalar): Likewise.
12377 2017-09-26 Martin Jambor <mjambor@suse.cz>
12379 * tree-sra.c (compare_access_positions): Put integral types first,
12380 stabilize sorting of integral types, remove conditions putting
12381 non-full-precision integers last.
12382 (sort_and_splice_var_accesses): Disable scalarization if a
12383 non-integert would be represented by a non-full-precision integer.
12385 2017-09-26 Joseph Myers <joseph@codesourcery.com>
12387 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
12388 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
12389 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
12390 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
12391 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
12392 conditionals inside the function instead of around it. Call
12393 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
12394 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
12396 2017-09-26 Richard Biener <rguenther@suse.de>
12398 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
12400 (scop_detection::build_scop_breadth): ... this. Removed.
12401 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
12402 (scop_detection::harmful_stmt_in_bb): Likewise.
12403 (scop_detection::graphite_can_represent_stmt): Likewise.
12404 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
12405 (scop_detection::can_represent_loop): Remove recursion, fold in ...
12406 (scop_detection::can_represent_loop_1): ... this. Removed.
12407 (scop_detection::harmful_loop_in_region): Simplify after inlining
12408 the above and remove more quadraticness.
12409 (build_scops): Adjust.
12410 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
12413 2017-09-26 Jakub Jelinek <jakub@redhat.com>
12416 * config/i386/i386.c (ix86_print_operand_address_as): Only test
12417 REGNO (base) == SP_REG if base is a REG.
12419 PR middle-end/35691
12420 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
12421 if it is different SSA_NAME.
12422 (optimize_range_tests_cmp_bitwise): New function.
12423 (optimize_range_tests): Call it.
12425 2017-09-26 Richard Biener <rguenther@suse.de>
12427 PR tree-optimization/82321
12428 * graphite.c (canonicalize_loop_closed_ssa): Properly check
12429 for the def being inside the loop.
12431 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12433 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
12435 * config/s390/s390-builtins.def: Fix constraint on op4.
12437 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12439 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
12440 independent expanders.
12441 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
12442 ("vec_ordered", "vec_unordered"): New expanders.
12444 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12446 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
12449 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12451 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
12452 vec_unpacks_lo_v16qi.
12453 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
12455 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12457 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
12458 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
12459 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
12461 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12463 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
12465 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
12467 ("*vec_slb<mode>"): New insn pattern.
12468 ("vec_shr_<mode>"): New expander.
12469 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
12470 and force the shift count operand to V16QImode.
12471 ("vec_srb<mode>"): Set shift count mode to V16QI.
12473 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12475 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
12476 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
12477 ("vec_widen_smult_hi_<mode>"): New expander definitions.
12479 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
12482 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
12484 2017-09-26 Richard Biener <rguenther@suse.de>
12486 PR tree-optimization/82320
12487 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
12490 2017-09-25 Jeff Law <law@redhat.com>
12492 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
12493 prototype for new argument.
12494 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
12495 mostly extracted from rs6000_emit_allocate_stack.
12496 (rs6000_emit_probe_stack_range_stack_clash): New function.
12497 (rs6000_emit_allocate_stack): Call
12498 rs6000_emit_probe_stack_range_stack_clash as needed.
12499 (rs6000_emit_probe_stack_range): Add additional argument
12500 to call to gen_probe_stack_range{si,di}.
12501 (output_probe_stack_range): New.
12502 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
12503 (output_probe_stack_range_stack_clash): New.
12504 (rs6000_emit_prologue): Emit notes into dump file as requested.
12505 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
12506 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
12507 Add additional operand and pass it to output_probe_stack_range.
12509 2017-09-25 Bin Cheng <bin.cheng@arm.com>
12511 PR tree-optimization/82163
12512 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
12513 (checking_verify_loop_closed_ssa): New parameter.
12514 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
12515 (check_loop_closed_ssa_stmt): Delete.
12516 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
12517 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
12518 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
12521 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
12523 * brig-builtins.def: Treat HSAIL barrier builtins as
12524 setjmp/longjump style functions.
12526 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
12528 * target.def (constant_alignment): New hook.
12529 * defaults.h (CONSTANT_ALIGNMENT): Delete.
12530 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
12531 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
12532 * doc/tm.texi: Regenerate.
12533 * targhooks.h (default_constant_alignment): Declare.
12534 (constant_alignment_word_strings): Likewise.
12535 * targhooks.c (default_constant_alignment): New function.
12536 (constant_alignment_word_strings): Likewise.
12537 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
12538 instead of CONSTANT_ALIGNMENT.
12539 * varasm.c (align_variable, get_variable_align, build_constant_desc)
12540 (force_const_mem): Likewise.
12541 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
12542 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
12543 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
12544 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12545 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
12547 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
12548 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12549 constant_alignment_word_strings.
12550 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
12551 (CONSTANT_ALIGNMENT): Likewise.
12552 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12553 (arm_constant_alignment): New function.
12554 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
12555 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12556 constant_alignment_word_strings.
12557 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
12558 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12559 constant_alignment_word_strings.
12560 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
12561 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12562 (cris_constant_alignment): New function.
12563 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
12564 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12565 (epiphany_constant_alignment): New function.
12566 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
12567 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12568 constant_alignment_word_strings.
12569 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
12570 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12571 constant_alignment_word_strings.
12572 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
12573 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12574 constant_alignment_word_strings.
12575 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
12576 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
12577 * config/i386/i386.c (ix86_constant_alignment): Make static.
12578 Use the same interface as the target hook.
12579 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12580 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
12581 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12582 constant_alignment_word_strings.
12583 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
12584 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
12585 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12586 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
12587 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12588 constant_alignment_word_strings.
12589 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
12590 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12591 constant_alignment_word_strings.
12592 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
12593 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12594 constant_alignment_word_strings.
12595 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
12596 * config/microblaze/microblaze.c (microblaze_constant_alignment):
12598 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12599 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
12600 * config/mips/mips.c (mips_constant_alignment): New function.
12601 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12602 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
12603 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
12604 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12605 (mmix_constant_alignment): Make static. Use the same interface
12606 as the target hook.
12607 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
12608 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12609 constant_alignment_word_strings.
12610 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
12611 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12612 constant_alignment_word_strings.
12613 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
12614 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12615 constant_alignment_word_strings.
12616 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
12617 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12618 (rs6000_constant_alignment): New function.
12619 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
12620 * config/riscv/riscv.c (riscv_constant_alignment): New function.
12621 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12622 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
12623 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12624 (rs6000_constant_alignment): New function.
12625 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
12626 * config/s390/s390.c (s390_constant_alignment): New function.
12627 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12628 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
12629 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12630 constant_alignment_word_strings.
12631 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
12632 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12633 (sparc_constant_alignment): New function.
12634 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
12635 * config/spu/spu.c (spu_constant_alignment): New function.
12636 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12637 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
12638 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12639 constant_alignment_word_strings.
12640 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
12641 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12642 constant_alignment_word_strings.
12643 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
12644 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12645 constant_alignment_word_strings.
12646 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
12647 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12648 (visium_constant_alignment): New function.
12649 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
12650 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12651 (xtensa_constant_alignment): New function.
12652 * system.h (CONSTANT_ALIGNMENT): Poison.
12654 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
12656 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
12657 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
12658 (rs6000_builtin_valid_without_lhs): New helper function.
12659 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12660 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
12662 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
12664 * target.h (vec_perm_indices): Use unsigned short rather than
12666 (auto_vec_perm_indices): Likewise.
12667 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
12668 Use unsigned int rather than unsigned char.
12669 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
12671 2017-09-25 Richard Biener <rguenther@suse.de>
12673 * cfgloop.h (sort_sibling_loops): Declare.
12674 * cfgloop.c (sort_sibling_loops_cmp): New helper.
12675 (sort_sibling_loops): New function sorting the sibling loop list
12677 * graphite.c (graphite_transform_loops): Sort sibling loops.
12679 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
12681 * target.def (vec_perm_const_ok): Change sel parameter to
12683 * optabs-query.c (can_vec_perm_p): Update accordingly.
12684 * doc/tm.texi: Regenerate.
12685 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
12686 auto_vec_perm_indices and remove separate nelt field.
12687 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
12688 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
12689 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
12690 (aarch64_expand_vec_perm_const): Update accordingly.
12691 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
12692 to vec_perm_indices.
12693 * config/arm/arm.c (expand_vec_perm_d): Change perm to
12694 auto_vec_perm_indices and remove separate nelt field.
12695 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
12696 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
12697 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
12699 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
12700 to vec_perm_indices.
12701 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
12702 sel to vec_perm_indices.
12703 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
12704 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
12705 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
12707 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
12710 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
12713 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
12714 on the FUNCTION_DECL function context if it has a DIE that is a
12717 2017-09-25 Richard Biener <rguenther@suse.de>
12719 PR tree-optimization/82285
12720 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
12723 2017-09-25 Tom de Vries <tom@codesourcery.com>
12727 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
12730 2017-09-25 Richard Biener <rguenther@suse.de>
12732 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
12733 ISL errors other than isl_error_quota happen. Dump if the
12734 schedule is the same.
12735 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
12736 errors instead of aborting inside ISL.
12738 2017-09-25 Iain Sandoe <iain@codesourcery.com>
12741 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
12742 of libgcc_eh for m64.
12743 * config/i386/darwin64.h: Likewise.
12745 2017-09-25 Richard Biener <rguenther@suse.de>
12747 PR middle-end/82144
12748 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
12749 attribute for incomplete types nor twice for complete ones.
12751 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
12754 * config/i386/i386.c (ix86_print_operand_address_as): Encode
12755 %esp as %rsp to avoid 0x67 prefix if there is no index or base
12758 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
12761 * config/i386/i386.opt (mprefer-avx256): Use
12762 ix86_target_flags variable.
12763 * config/i386/i386.c (ix86_target_string): Move
12764 -mprefer-avx256 to flag2_opts.
12766 2017-09-22 Jakub Jelinek <jakub@redhat.com>
12768 PR middle-end/35691
12769 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
12770 and x != -1 | y != -1 into (x & y) != -1.
12772 2017-09-22 Steve Ellcey <sellcey@cavium.com>
12774 * config.gcc: Add new case statement to set
12775 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
12776 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
12777 s390x-*-linux* case statements. Added aarch64 to the list of
12778 supported architectures.
12780 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
12782 PR tree-optimization/82289
12783 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
12784 STMT_VINFO_RELEVANT_P.
12786 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
12787 Alan Hayward <alan.hayward@arm.com>
12788 David Sherwood <david.sherwood@arm.com>
12790 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
12791 for VR_RANGE only; don't allow VR_ANTI_RANGE.
12792 (extract_range_from_binary_expr_1): Don't call
12793 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
12795 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
12796 Alan Hayward <alan.hayward@arm.com>
12797 David Sherwood <david.sherwood@arm.com>
12799 * target.def (preferred_vector_alignment): New hook.
12800 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
12802 * doc/tm.texi: Regenerate.
12803 * targhooks.h (default_preferred_vector_alignment): Declare.
12804 * targhooks.c (default_preferred_vector_alignment): New function.
12805 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
12807 (DR_TARGET_ALIGNMENT): New macro.
12808 (aligned_access_p): Update commentary.
12809 (vect_known_alignment_in_bytes): New function.
12810 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
12812 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
12813 Calculate the misalignment based on the target alignment rather than
12815 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
12816 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
12817 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
12818 the target alignment, rather than masking the element misalignment
12819 with the number of elements in a vector. Also use the target
12820 alignment when calculating the maximum number of peels.
12821 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
12822 instead of TYPE_ALIGN_UNIT.
12823 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
12824 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
12825 (vect_create_addr_base_for_vector_ref): Update call accordingly.
12826 (vect_create_data_ref_ptr): Likewise.
12827 (vect_setup_realignment): Realign by ANDing with
12828 -DR_TARGET_MISALIGNMENT.
12829 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
12830 the number of peels based on DR_TARGET_ALIGNMENT.
12831 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
12832 with the guaranteed alignment boundary when deciding whether
12834 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
12835 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
12836 (ensure_base_align): Remove stmt_info parameter. Get the
12837 target base alignment from DR_TARGET_ALIGNMENT.
12838 (vectorizable_store): Update call accordingly. Interpret
12839 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
12841 (vectorizable_load): Likewise.
12843 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
12844 Alan Hayward <alan.hayward@arm.com>
12845 David Sherwood <david.sherwood@arm.com>
12847 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
12848 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
12849 (vect_enhance_data_refs_alignment): Likewise.
12851 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
12853 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
12854 error. Only quit immediately if parsing is complete.
12855 (BEGIN): Initialize fatal_err and parse_done.
12856 (begin fpu, end fpu): Check number of arguments.
12857 (begin arch, end arch): Likewise.
12858 (begin cpu, end cpu): Likewise.
12859 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
12860 (optalias): Likewise.
12862 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
12864 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
12865 * config/arm/arm-isa.h: Delete. Move definitions to ...
12866 * arm-cpus.in: ... here. Use new feature and fgroup values.
12867 * config/arm/arm.c (arm_option_override): Use lower case for feature
12869 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
12870 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
12871 * config/arm/parsecpu.awk (END): Add new command 'isa'.
12873 (print_isa_bits_for): New function.
12874 (gen_isa): New function.
12875 (gen_comm_data): Use print_isa_bits_for.
12876 (define feature): New keyword.
12877 (define fgroup): New keyword.
12878 * config/arm/t-arm (TM_H): Remove.
12879 (GTM_H): Add arm-isa.h.
12880 (arm-isa.h): Add rule to generate file.
12881 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
12882 case for feature bit names.
12884 2017-09-22 Richard Biener <rguenther@suse.de>
12886 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
12888 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
12889 print of no dependency loops ...
12890 * graphite.c (graphite_transform_loops): ... here.
12891 (canonicalize_loop_closed_ssa_form): Work from inner to outer
12893 (same_close_phi_node, remove_duplicate_close_phi,
12894 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
12895 (canonicalize_loop_closed_ssa): ... here and simplify.
12896 * graphite-optimize-isl.c: Include tree-vectorizer.h.
12897 (optimize_isl): Use dump_printf_loc to tell when we stopped
12898 optimizing because of an ISL timeout.
12900 2017-09-22 Richard Biener <rguenther@suse.de>
12902 PR tree-optimization/82291
12903 * tree-if-conv.c (predicate_mem_writes): Make sure to
12904 remove writes in blocks predicated with false.
12906 2017-09-22 Richard Biener <rguenther@suse.de>
12908 * sese.c: Include cfganal.h.
12909 (if_region_set_false_region): Remove.
12910 (create_if_region_on_edge): Likewise.
12911 (move_sese_in_condition): Re-implement without destroying
12914 2017-09-22 Richard Biener <rguenther@suse.de>
12916 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
12917 Verify both BBs contain loop PHI nodes before dispatching to
12918 copy_loop_phi_args.
12919 (graphite_regenerate_ast_isl): Do not recompute dominators,
12920 do not verify three times. Restructure for clarity.
12921 * graphite-scop-detection.c (same_close_phi_node,
12922 remove_duplicate_close_phi, make_close_phi_nodes_unique,
12923 defined_in_loop_p, canonicalize_loop_closed_ssa,
12924 canonicalize_loop_closed_ssa_form): Simplify, remove excess
12925 checking and SSA rewrite, move to ...
12926 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
12927 (graphite_initialize): Do not pass in ctx, do not reset the
12928 SCEV cache, compute only dominators.
12929 (graphite_transform_loops): Allocate ISL ctx after
12930 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
12931 Maintain post-dominators only around build_scops.
12932 * sese.c (if_region_set_false_region): Make static. Free
12933 and recompute dominators.
12934 (move_sese_in_condition): Assert we don't get called with
12935 post-dominators computed.
12936 * sese.h (if_region_set_false_region): Remove.
12938 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
12940 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
12941 mode attribute for TARGET_AVX512VL.
12943 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
12945 * config/i386/i386.opt (mprefer-avx256): New option.
12946 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
12948 (ix86_preferred_simd_mode): Return 256-bit AVX modes
12949 for TARGET_PREFER_AVX256.
12950 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
12952 2017-09-21 Jeff Law <law@redhat.com>
12954 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
12955 Fix dump output if the only stack space is for pushed registers.
12957 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
12959 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
12962 2017-09-21 Martin Sebor <msebor@redhat.com>
12965 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
12966 code (in C++) or code that triggers warnings.
12968 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
12970 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
12972 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
12974 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
12975 * sched-rgn.c: Ditto.
12976 * sel-sched-ir.c: Ditto.
12978 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
12980 * toplev.h (set_random_seed): Adjust return type.
12981 * toplev.c (init_local_tick): Move eager initialization of random_seed
12982 to get_random_seed. Adjust comment.
12983 (init_random_seed): Inline to get_random_seed, delete.
12984 (get_random_seed): Initialize random_seed lazily.
12985 (set_random_seed): Do not return previous value.
12986 (print_switch_value): Do not call get_random_seed.
12988 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
12990 * cgraph.c (delete_function_version): New, broken out from...
12991 (cgraph_node::delete_function_version): ...here. Rename to
12992 cgraph_node::delete_function_version_by_decl. Update all uses.
12993 (cgraph_node::remove): Call delete_function_version.
12995 2017-09-21 Jakub Jelinek <jakub@redhat.com>
12998 * tree-inline.c (expand_call_inline): Emit clobber stmts for
12999 VAR_DECLs to which addressable non-volatile parameters are mapped
13000 and for id->retvar after the return value assignment. Clear
13001 id->retval and id->retbnd after inlining.
13003 2017-09-21 Richard Biener <rguenther@suse.de>
13005 PR tree-optimization/82276
13006 PR tree-optimization/82244
13007 * tree-vrp.c (build_assert_expr_for): Set
13008 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
13010 (remove_range_assertions): Revert earlier change.
13012 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
13015 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
13017 2017-09-21 Richard Biener <rguenther@suse.de>
13019 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
13020 Restore valid IL after code generation errors.
13021 * graphite.c (graphite_transform_loops): Diagnose code
13022 generation issues as MSG_MISSED_OPTIMIZATION and continue
13023 with processing SCOPs.
13025 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13026 Alan Hayward <alan.hayward@arm.com>
13027 David Sherwood <david.sherwood@arm.com>
13029 * calls.c (compute_argument_addresses): Use simplify_gen_binary
13030 rather than choosing between plus_constant and gen_rtx_<CODE>.
13031 * expr.c (emit_push_insn): Likewise.
13032 (expand_expr_real_2): Likewise.
13034 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13035 Alan Hayward <alan.hayward@arm.com>
13036 David Sherwood <david.sherwood@arm.com>
13038 * loop-unroll.c (split_iv): Call copy_rtx on the step.
13040 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13041 Alan Hayward <alan.hayward@arm.com>
13042 David Sherwood <david.sherwood@arm.com>
13044 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
13045 calling tree_to_uhwi.
13047 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13048 Alan Hayward <alan.hayward@arm.com>
13049 David Sherwood <david.sherwood@arm.com>
13051 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
13052 INTEGER_CST rather than a negative test for ADDR_EXPR.
13054 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13055 Alan Hayward <alan.hayward@arm.com>
13056 David Sherwood <david.sherwood@arm.com>
13058 * tree-vrp.c (extract_range_from_binary_expr_1): Check
13059 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
13061 2017-09-21 Richard Biener <rguenther@suse.de>
13063 PR tree-optimization/71351
13064 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
13065 graphite_create_new_loop_guard): Remove, fold remaining parts
13067 (translate_isl_ast_node_for): ... here and simplify.
13069 2017-09-21 Jakub Jelinek <jakub@redhat.com>
13072 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
13073 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
13074 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
13075 alternative always use QI mode, for -Os imov (=R,R) alternative
13076 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
13079 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13080 Jeff Law <law@redhat.com>
13082 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
13083 (allocate_stack_space): New function, partially extracted from
13084 s390_emit_prologue.
13085 (s390_emit_prologue): Track offset to most recent stack probe.
13086 Code to allocate space moved into allocate_stack_space.
13087 Dump actions when no stack is allocated.
13088 (s390_prologue_plus_offset): New function.
13089 (s390_emit_stack_probe): Likewise.
13091 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
13093 * common.opt (Wa, Wl, Wp, g, gz=): Add
13095 (gno-column-info): Remove.
13096 (gcolumn-info): Drop RejectNegative.
13097 (gno-): New prefix.
13098 (gno-record-gcc-switches): Remove.
13099 (grecord-gcc-switches): Drop RejectNegative.
13100 (gno-split-dwarf): Remove.
13101 (gsplit-dwarf): Drop RejectNegative.
13102 (gno-strict-dwarf): Remove.
13103 (gstrict-dwarf): Drop RejectNegative.
13104 * config/darwin.opt (gfull, gused): Add RejectNegative.
13105 * dwarf2out.c (gen_producer_string): Drop
13106 gno-record-gcc-switches handler.
13107 * optc-gen.awk: Add g to prefixes with negative forms.
13108 * opts-common.c (remapping_prefix_p): New.
13109 (find_opt): Check it.
13110 (generate_canonical_option): Test g prefix.
13111 (option_map): Add -gno- mapping.
13112 (add_misspelling_candidates): Check remapping_prefix_p.
13114 2017-09-20 Jeff Law <law@redhat.com>
13116 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
13117 thinko in stack clash protection support.
13119 * explow.c (compute_stack_clash_protection_loop_data): Use
13120 CONST_INT_P instead of explicit test. Verify object is a
13121 CONST_INT_P before looking at INTVAL.
13122 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
13123 instead of explicit test.
13125 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
13128 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
13129 address instead of to r1 and r11.
13131 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
13133 * config.gcc: Support "knm".
13134 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
13135 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
13137 * config/i386/i386.c (m_KNM): Define.
13138 (processor_target_table): Add "knm".
13140 (ix86_option_override_internal): Add "knm".
13141 (ix86_issue_rate): Add PROCESSOR_KNM.
13142 (ix86_adjust_cost): Ditto.
13143 (ia32_multipass_dfa_lookahead): Ditto.
13144 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
13145 (fold_builtin_cpu): Add M_INTEL_KNM.
13146 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
13147 (processor_type): Add PROCESSOR_KNM.
13148 * config/i386/x86-tune.def: Add m_KNM.
13149 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
13151 2017-09-20 Richard Biener <rguenther@suse.de>
13153 PR tree-optimization/80213
13154 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
13155 are allowed in empty BBs as well.
13156 (canonicalize_loop_closed_ssa): Also look for other complex
13158 (scop_detection::get_sese): Include the loop-closed PHI block
13160 (scop_detection::merge_sese): Remove code adding extra blocks.
13161 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
13162 (build_scops): Assert the final returned scop is invalid.
13164 2017-09-20 Richard Biener <rguenther@suse.de>
13166 PR tree-optimization/82264
13167 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
13169 (vn_phi_lookup): Likewise.
13170 (vn_phi_insert): Likewise.
13172 2017-09-20 Jakub Jelinek <jakub@redhat.com>
13174 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
13175 that fits into uhwi or shwi, add DW_AT_const_value regardless
13176 of early_dwarf without going through RTL, using add_AT_unsigned
13179 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
13180 (DEBUG_LTO_ABBREV_SECTION): Likewise.
13181 (DEBUG_LTO_MACINFO_SECTION): Likewise.
13182 (DEBUG_MACRO_SECTION): Likewise.
13183 (DEBUG_LTO_MACRO_SECTION): Likewise.
13184 (DEBUG_STR_DWO_SECTION): Likewise.
13185 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
13186 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
13187 (DEBUG_LTO_DWO_LINE_SECTION): Define.
13188 (DEBUG_LTO_LINE_STR_SECTION): Define.
13189 (init_sections_and_labels): Initialize debug_line_str_section
13190 variable. Initialize debug_loc_section for -gdwarf-5 to
13191 DEBUG_LOCLISTS_SECTION. Formatting fixes.
13193 2017-09-20 Richard Biener <rguenther@suse.de>
13195 * graphite-sese-to-poly.c (extract_affine): Properly handle
13196 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
13198 2017-09-20 Richard Biener <rguenther@suse.de>
13200 PR tree-optimization/81373
13201 * graphite-scop-detection.c (build_cross_bb_scalars_def):
13202 Force SESE live-out defs to be handled even if they are
13205 2017-09-19 Jeff Law <law@redhat.com>
13207 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
13208 nothing for stack adjustments with REG_STACK_CHECK.
13209 * sched-deps.c (parse_add_or_inc): Reject insns with
13210 REG_STACK_CHECK from dependency breaking.
13211 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
13212 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
13213 * reg-notes.def (STACK_CHECK): New note.
13215 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
13216 (ix86_expand_prologue): Dump stack clash info as needed.
13217 Call ix86_adjust_stack_and_probe_stack_clash as needed.
13219 * function.c (dump_stack_clash_frame_info): New function.
13220 * function.h (dump_stack_clash_frame_info): Prototype.
13221 (enum stack_clash_probes): New enum.
13223 * config/alpha/alpha.c (alpha_expand_prologue): Also check
13224 flag_stack_clash_protection.
13225 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
13226 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
13227 (arm_frame_pointer_required): Likewise.
13228 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
13229 (ia64_expand_prologue): Likewise.
13230 * config/mips/mips.c (mips_expand_prologue): Likewise.
13231 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
13232 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
13233 (sparc_flat_expand_prologue): Likewise.
13234 * config/spu/spu.c (spu_expand_prologue): Likewise.
13236 * explow.c: Include "params.h".
13237 (anti_adjust_stack_and_probe_stack_clash): New function.
13238 (get_stack_check_protect): Likewise.
13239 (compute_stack_clash_protection_loop_data): Likewise.
13240 (emit_stack_clash_protection_loop_start): Likewise.
13241 (emit_stack_clash_protection_loop_end): Likewise.
13242 (allocate_dynamic_stack_space): Use get_stack_check_protect.
13243 Use anti_adjust_stack_and_probe_stack_clash.
13244 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
13245 (emit_stack_clash_protection_loop_start): Likewise.
13246 (emit_stack_clash_protection_loop_end): Likewise.
13247 * rtl.h (get_stack_check_protect): Prototype.
13248 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
13249 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
13250 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
13252 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
13254 * doc/tm.texi: Rebuilt.
13255 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
13256 get_stack_check_protect.
13257 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
13258 * config/arm/arm.c (arm_expand_prologue): Likewise.
13259 (arm_frame_pointer_required): Likewise.
13260 * config/i386/i386.c (ix86_expand_prologue): Likewise.
13261 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
13262 * config/mips/mips.c (mips_expand_prologue): Likewise.
13263 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
13264 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
13265 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
13266 (sparc_flat_expand_prologue): Likewise.
13268 * common.opt (-fstack-clash-protection): New option.
13269 * flag-types.h (enum stack_check_type): Note difference between
13270 -fstack-check= and -fstack-clash-protection.
13271 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
13272 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
13273 * toplev.c (process_options): Issue warnings/errors for cases
13274 not handled with -fstack-clash-protection.
13275 * doc/invoke.texi (-fstack-clash-protection): Document new option.
13276 (-fstack-check): Note additional problem with -fstack-check=generic.
13277 Note that -fstack-check is primarily for Ada and refer users
13278 to -fstack-clash-protection for stack-clash-protection.
13279 Document new params for stack clash protection.
13281 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
13283 * config/i386/i386.c (ix86_split_long_move): Do not handle
13284 address used for LEA in a special way.
13286 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
13288 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
13290 2017-09-19 Martin Sebor <msebor@redhat.com>
13293 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
13294 of incompatible types.
13296 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
13298 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13299 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
13300 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13301 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
13303 2017-09-19 Richard Biener <rguenther@suse.de>
13305 PR tree-optimization/82244
13306 * tree-vrp.c (remove_range_assertions): Do not propagate
13307 a constant to abnormals but replace the assert with a copy.
13309 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
13311 PR rtl-optimization/57878
13312 PR rtl-optimization/68988
13313 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
13314 avoidance test involving non_reload_pseudos. Move frequency test
13315 below the general fragmentation avoidance test.
13317 2017-09-19 Richard Biener <rguenther@suse.de>
13319 PR tree-optimization/69728
13320 * graphite-sese-to-poly.c (schedule_error): New global.
13321 (add_loop_schedule): Handle empty domain by failing the
13323 (build_original_schedule): Handle schedule_error.
13325 2017-09-19 Richard Biener <rguenther@suse.de>
13327 * graphite-scop-detection.c (scop_detection::can_represent_loop):
13328 Do not iterate to sibling loops but only to siblings of inner
13331 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
13334 * config/m68k/m68k.md (moveq feeding equality comparison): Check
13335 that the registers are different.
13337 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
13339 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
13340 to processor_model and "amdfam17h" to arch_names_table.
13341 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
13343 2017-09-18 Jakub Jelinek <jakub@redhat.com>
13346 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
13348 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13349 Alan Hayward <alan.hayward@arm.com>
13350 David Sherwood <david.sherwood@arm.com>
13352 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
13354 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
13356 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
13357 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
13358 vect_schedule_slp_instance.
13359 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
13360 Update call to vect_slp_analyze_node_operations. Simplify return
13362 (vect_slp_analyze_bb_1): Update call accordingly.
13363 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
13364 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
13365 (vect_schedule_slp): Update call accordingly.
13367 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13368 Alan Hayward <alan.hayward@arm.com>
13369 David Sherwood <david.sherwood@arm.com>
13371 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
13372 with types that aren't in fact scalar.
13374 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13376 * tree-vect-slp.c (vect_record_max_nunits): New function,
13378 (vect_build_slp_tree_1): ...here.
13379 (vect_build_slp_tree_2): Call it for phis too.
13381 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13383 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
13384 to vect_get_vec_def_for_operand when getting the mask operand.
13386 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13387 Alan Hayward <alan.hayward@arm.com>
13388 David Sherwood <david.sherwood@arm.com>
13390 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
13393 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13394 Alan Hayward <alan.hayward@arm.com>
13395 David Sherwood <david.sherwood@arm.com>
13397 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
13398 calculation for vector booleans.
13400 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13401 Alan Hayward <alan.hayward@arm.com>
13402 David Sherwood <david.sherwood@arm.com>
13404 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
13406 (vect_transform_stmt): ...here.
13407 (vect_analyze_stmt): Use it instead of calling
13408 vectorizable_live_operation directly.
13410 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
13412 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
13413 non-SIMT targets in acc vector loops.
13415 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
13417 * configure.ac: Add arc and check if assembler supports gdwarf2.
13418 * configure: Regenerate.
13420 2017-09-18 Richard Biener <rguenther@suse.de>
13422 PR tree-optimization/82220
13423 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
13424 epilogue niters from the min_profitable_iters compute.
13426 2017-09-18 Jakub Jelinek <jakub@redhat.com>
13429 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
13430 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
13431 (ix86_init_pic_reg): Revert 2017-09-01 changes.
13433 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
13436 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
13437 switching to a new text section.
13439 2017-09-18 Richard Biener <rguenther@suse.de>
13441 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
13443 (build_alias_set): Reject aliases with no access function.
13445 2017-09-18 Richard Biener <rguenther@suse.de>
13447 PR tree-optimization/79622
13448 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
13450 (build_cross_bb_scalars_use): Likewise.
13452 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
13454 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
13456 2017-09-18 Alan Modra <amodra@gmail.com>
13459 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
13460 stack_pointer_rtx for count 0. Update comments. Break up
13461 large rtl expression.
13463 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
13465 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
13466 Increase to 20 bytes.
13467 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
13468 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
13469 or avx version of the stub.
13471 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
13474 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
13475 compute the minimum stack alignment. Also update preferred stack
13476 boundary for leaf functions.
13478 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
13480 PR tree-optimization/82228
13481 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
13484 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
13486 * common/config/nds32/nds32-common.c
13487 (nds32_option_optimization_table): Refine formatting.
13488 (nds32_option_optimization_table): Use -fsched-pressure and
13489 -fomit-frame-pointer for specific optimization level.
13491 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
13493 * config/nds32/nds32.c: Refine formatting and comments.
13494 * config/nds32/nds32.h: Likewise.
13495 * config/nds32/nds32.md: Likewise.
13496 * config/nds32/nds32-cost.c: Likewise.
13497 * config/nds32/nds32-isr.c: Likewise.
13498 * config/nds32/nds32-md-auxiliary.c: Likewise.
13499 * config/nds32/nds32-multiple.md: Likewise.
13500 * config/nds32/nds32-predicates.c: Likewise.
13502 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
13503 Jakub Jelinek <jakub@redhat.com>
13505 Add support for -std=c++2a.
13506 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
13508 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
13510 2017-09-15 Steve Ellcey <sellcey@cavium.com>
13513 * doc/extend.texi (Common Function Attributes): Add
13514 references to ARM, AArch64, and S/390 specific attributes.
13515 (Function Specific Option Pragmas): Add AArch64 and S/390
13516 to list of back ends that support the target pragma.
13518 2017-09-15 Nathan Sidwell <nathan@acm.org>
13520 * doc/standards.texi: Fix C++17 description. Update URLs for
13523 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
13525 * common.opt (Wcast-align=strict): New warning option.
13526 * doc/invoke.texi: Document -Wcast-align=strict.
13528 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
13530 * cgraph.h (cgraph_thunk_info): Add comments.
13531 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
13532 assert for VIRTUAL_* arguments stricter.
13534 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
13536 PR tree-optimization/71026
13537 * match.pd: Move RDIV patterns from fold-const.c
13538 * fold-const.c (distribute_real_division): Removed.
13539 (fold_binary_loc): Remove calls to distribute_real_divison.
13541 2017-09-15 Jakub Jelinek <jakub@redhat.com>
13543 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
13544 c++1z and gnu++1z as deprecated. Change other references to
13545 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
13546 Change -Wc++1z-compat to -Wc++17-compat.
13547 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
13548 * dwarf2out.c (highest_c_language): Handle C++17.
13549 (gen_compile_unit_die): Likewise.
13551 2017-09-15 Jakub Jelinek <jakub@redhat.com>
13553 PR rtl-optimization/82192
13554 * combine.c (make_extraction): Don't look through non-paradoxical
13555 SUBREGs or TRUNCATE if pos + len is or might be bigger than
13558 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
13559 Alan Hayward <alan.hayward@arm.com>
13560 David Sherwood <david.sherwood@arm.com>
13562 * target.def (function_arg_offset): New hook.
13563 * targhooks.h (default_function_arg_offset): Declare.
13564 * targhooks.c (default_function_arg_offset): New function.
13565 * function.c (locate_and_pad_parm): Use
13566 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
13567 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
13568 (TARGET_FUNCTION_ARG_OFFSET): ...this.
13569 * doc/tm.texi: Regenerate.
13570 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
13571 * config/spu/spu.c (spu_function_arg_offset): New function.
13572 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
13573 * system.h (FUNCTION_ARG_OFFSET): Poison.
13575 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
13576 Alan Hayard <alan.hayward@arm.com>
13577 David Sherwood <david.sherwood@arm.com>
13579 * target.def (truly_noop_truncation): New hook.
13580 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
13581 than TRULY_NOOP_TRUNCATION.
13582 * hooks.h (hook_bool_uint_uint_true): Declare.
13583 * hooks.c (hook_bool_uint_uint_true): New function.
13584 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
13585 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
13586 * doc/tm.texi: Regenerate.
13587 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
13588 rather than TRULY_NOOP_TRUNCATION in comments.
13589 (simplify_comparison): Likewise.
13590 (record_truncated_value): Likewise.
13591 * expmed.c (extract_bit_field_1): Likewise.
13592 (extract_split_bit_field): Likewise.
13593 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
13594 instead of TRULY_NOOP_TRUNCATION.
13595 * function.c (assign_parm_setup_block): Likewise.
13596 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
13597 * rtlhooks.c: Include target.h.
13598 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
13599 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
13600 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
13601 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
13602 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
13603 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
13604 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
13605 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
13606 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
13607 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
13608 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
13609 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
13610 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
13611 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
13612 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
13613 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
13614 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
13615 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
13616 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
13617 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
13618 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
13619 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
13620 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
13621 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
13622 * config/mips/mips.c (mips_truly_noop_truncation): New function.
13623 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
13624 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
13625 rather than TRULY_NOOP_TRUNCATION in comments.
13626 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
13627 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
13628 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
13629 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
13630 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
13631 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
13632 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
13633 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
13634 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
13635 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
13636 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
13637 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
13638 rather than TRULY_NOOP_TRUNCATION in comments.
13639 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
13640 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
13641 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
13642 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
13643 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
13644 TRULY_NOOP_TRUNCATION condition.
13645 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
13646 (TRULY_NOOP_TRUNCATION): Delete.
13647 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
13648 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
13649 * config/spu/spu.c (spu_truly_noop_truncation): New function.
13650 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
13651 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
13652 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
13653 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
13654 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
13655 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
13656 rather than TRULY_NOOP_TRUNCATION in comments.
13657 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
13658 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
13659 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
13660 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
13661 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
13662 * system.h (TRULY_NOOP_TRUNCATION): Poison.
13664 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
13667 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
13668 (*cmp_ior): Likewise.
13669 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
13670 (*ior_scc_scc_cmp): Likewise.
13671 (*and_scc_scc): Likewise.
13672 (*and_scc_scc_cmp): Likewise.
13674 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
13675 Alan Hayard <alan.hayward@arm.com>
13676 David Sherwood <david.sherwood@arm.com>
13678 * target.def (can_change_mode_class): New hook.
13679 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
13680 (hard_regno_nregs): Likewise.
13681 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
13682 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
13683 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
13684 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
13685 (LOAD_EXTEND_OP): Update accordingly.
13686 * doc/tm.texi: Regenerate.
13687 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
13688 CANNOT_CHANGE_MODE_CLASS.
13689 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
13690 (REG_CAN_CHANGE_MODE_P): ...this new macro.
13691 * combine.c (simplify_set): Update accordingly.
13692 * emit-rtl.c (validate_subreg): Likewise.
13693 * recog.c (general_operand): Likewise.
13694 * regcprop.c (mode_change_ok): Likewise.
13695 * reload1.c (choose_reload_regs): Likewise.
13696 (inherit_piecemeal_p): Likewise.
13697 * rtlanal.c (simplify_subreg_regno): Likewise.
13698 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
13699 instead of CANNOT_CHANGE_MODE_CLASS.
13700 (reload_cse_simplify_operands): Likewise.
13701 * reload.c (push_reload): Use targetm.can_change_mode_class
13702 instead of CANNOT_CHANGE_MODE_CLASS.
13703 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
13704 REG_CANNOT_CHANGE_MODE_P.
13705 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13706 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
13707 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13708 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13709 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13710 (arm_can_change_mode_class): New function.
13711 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
13712 than CANNOT_CHANGE_MODE_CLASS in comments.
13713 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13714 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
13715 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
13716 (ix86_can_change_mode_class): ...this new function, inverting the
13717 sense of the return value.
13718 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13719 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13720 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13721 (ia64_can_change_mode_class): New function.
13722 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13723 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
13724 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
13725 (m32c_can_change_mode_class): ...this new function, inverting the
13726 sense of the return value.
13727 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13728 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13729 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
13730 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
13731 (mips_can_change_mode_class): ...this new function, inverting the
13732 sense of the return value.
13733 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13734 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13735 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13736 (msp430_can_change_mode_class): New function.
13737 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13738 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
13739 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13740 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13741 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13742 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
13743 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13744 (pa_cannot_change_mode_class): Replace with...
13745 (pa_can_change_mode_class): ...this new function, inverting the
13746 sense of the return value.
13747 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
13748 than CANNOT_CHANGE_MODE_CLASS in comments.
13749 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13750 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
13751 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13752 (pdp11_cannot_change_mode_class): Replace with...
13753 (pdp11_can_change_mode_class): ...this new function, inverting the
13754 sense of the return value.
13755 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13756 * config/powerpcspe/powerpcspe-protos.h
13757 (rs6000_cannot_change_mode_class_ptr): Delete.
13758 * config/powerpcspe/powerpcspe.c
13759 (rs6000_cannot_change_mode_class_ptr): Delete.
13760 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13761 (rs6000_option_override_internal): Assign to
13762 targetm.can_change_mode_class instead of
13763 rs6000_cannot_change_mode_class_ptr.
13764 (rs6000_cannot_change_mode_class): Replace with...
13765 (rs6000_can_change_mode_class): ...this new function, inverting the
13766 sense of the return value.
13767 (rs6000_debug_cannot_change_mode_class): Replace with...
13768 (rs6000_debug_can_change_mode_class): ...this new function.
13769 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13770 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
13771 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13772 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13773 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
13775 * config/rs6000/rs6000.c (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/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13786 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
13787 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
13788 (s390_can_change_mode_class): ...this new function, inverting the
13789 sense of the return value.
13790 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13791 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13792 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
13793 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13794 (sh_cannot_change_mode_class): Replace with...
13795 (sh_can_change_mode_class): ...this new function, inverting the
13796 sense of the return value.
13797 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13798 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13799 (sparc_can_change_mode_class): New function.
13800 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13801 * config/spu/spu.c (spu_can_change_mode_class): New function.
13802 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13803 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13804 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13805 (visium_can_change_mode_class): New function.
13806 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
13808 2017-09-15 Richard Biener <rguenther@suse.de>
13810 PR tree-optimization/82217
13811 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
13812 but not undefined case.
13814 2017-09-15 Jakub Jelinek <jakub@redhat.com>
13817 * postreload.c (reload_cse_simplify_operands): Skip
13818 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
13820 2017-09-15 Richard Biener <rguenther@suse.de>
13822 PR tree-optimization/68823
13823 * graphite-scop-detection.c (build_alias_set): If we have a
13824 possible dependence check whether we can handle them by just
13825 looking at the DRs DR_ACCESS_FNs.
13826 (build_scops): If build_alias_set fails, fail the SCOP.
13828 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
13830 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
13831 to support float128 built-in functions that require the ISA 3.0
13833 (BU_FLOAT128_3_HW): Likewise.
13834 (SQRTF128): Add support for the IEEE 128-bit square root and fma
13835 built-in functions.
13836 (FMAF128): Likewise.
13838 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
13839 support for built-in functions that need the ISA 3.0 IEEE 128-bit
13840 floating point instructions.
13841 (rs6000_invalid_builtin): Likewise.
13842 (rs6000_builtin_mask_names): Likewise.
13843 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
13844 (RS6000_BTM_FLOAT128_HW): Likewise.
13845 (RS6000_BTM_COMMON): Likewise.
13846 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
13848 * doc/extend.texi (RS/6000 built-in functions): Document the
13849 IEEE 128-bit floating point square root and fused multiply-add
13850 built-in functions.
13852 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
13854 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
13855 reg (r2) isn't in the set of registers defined in the prologue.
13857 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13858 Alan Hayward <alan.hayward@arm.com>
13859 David Sherwood <david.sherwood@arm.com>
13861 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
13862 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
13863 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
13864 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
13865 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
13867 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
13868 max_vectorization_factor.
13869 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
13871 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13872 Alan Hayward <alan.hayward@arm.com>
13873 David Sherwood <david.sherwood@arm.com>
13875 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
13876 (vect_worthwhile_without_simd_p): Declare.
13877 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
13878 (vectorizable_reduction): Use it.
13879 * tree-vect-stmts.c (vectorizable_shift): Likewise.
13880 (vectorizable_operation): Likewise.
13882 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13883 Alan Hayward <alan.hayward@arm.com>
13884 David Sherwood <david.sherwood@arm.com>
13886 * tree-vectorizer.h (vect_get_num_copies): New function.
13887 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
13888 * tree-vect-loop.c (vectorizable_reduction): Likewise.
13889 (vectorizable_induction): Likewise.
13890 (vectorizable_live_operation): Likewise.
13891 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
13892 (vectorizable_bswap): Likewise.
13893 (vectorizable_call): Likewise.
13894 (vectorizable_conversion): Likewise.
13895 (vectorizable_assignment): Likewise.
13896 (vectorizable_shift): Likewise.
13897 (vectorizable_operation): Likewise.
13898 (vectorizable_store): Likewise.
13899 (vectorizable_load): Likewise.
13900 (vectorizable_condition): Likewise.
13901 (vectorizable_comparison): Likewise.
13902 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
13904 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13905 Alan Hayward <alan.hayward@arm.com>
13906 David Sherwood <david.sherwood@arm.com>
13908 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
13909 of vect_init_vector.
13911 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13912 Alan Hayward <alan.hayward@arm.com>
13913 David Sherwood <david.sherwood@arm.com>
13915 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
13916 an inline wrapper that provides a location.
13917 (gimple_build_vector): Likewise.
13918 * gimple-fold.c (gimple_build_vector_from_val): New function.
13919 (gimple_build_vector): Likewise.
13920 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
13921 functions to build the initial value. Always return a gimple value.
13922 (get_initial_defs_for_reduction): Likewise. Only compute
13924 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
13925 vect_init_vector on the results from get_initial_def(s)_for_reduction.
13926 (vectorizable_induction): Use gimple_build_vector rather than
13929 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13930 Alan Hayward <alan.hayward@arm.com>
13931 David Sherwood <david.sherwood@arm.com>
13933 * target.h (vec_perm_indices): New typedef.
13934 (auto_vec_perm_indices): Likewise.
13935 * optabs-query.h: Include target.h
13936 (can_vec_perm_p): Take a vec_perm_indices *.
13937 * optabs-query.c (can_vec_perm_p): Likewise.
13938 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
13939 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
13940 * tree-vect-generic.c (lower_vec_perm): Likewise.
13941 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
13942 (vect_grouped_load_supported): Likewise.
13943 (vect_shift_permute_load_chain): Likewise.
13944 (vect_permute_store_chain): Use auto_vec_perm_indices.
13945 (vect_permute_load_chain): Likewise.
13946 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
13947 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
13948 Update uses of can_vec_perm_p.
13949 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
13950 mode with a number of elements. Take a vec_perm_indices *.
13951 (vect_create_epilog_for_reduction): Update accordingly.
13952 Use auto_vec_perm_indices.
13953 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
13954 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
13955 (vect_transform_slp_perm_load): Likewise.
13956 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
13957 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
13958 (vect_gen_perm_mask_checked): Likewise.
13959 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
13960 (vect_gen_perm_mask_checked): Likewise.
13961 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
13962 (vectorizable_store): Likewise.
13963 (vectorizable_load): Likewise.
13964 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
13965 (vectorizable_bswap): Likewise.
13967 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13968 Alan Hayward <alan.hayward@arm.com>
13969 David Sherwood <david.sherwood@arm.com>
13971 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
13972 * tree.c (build_vector): Likewise.
13973 (build_vector_from_ctor): Update accordingly.
13974 (build_vector_from_val): Likewise.
13975 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
13976 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
13977 * tree-vect-generic.c (add_rshift): Likewise.
13978 (expand_vector_divmod): Likewise.
13979 (optimize_vector_constructor): Likewise.
13980 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
13981 (vect_transform_slp_perm_load): Likewise.
13982 (vect_schedule_slp_instance): Likewise.
13983 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
13984 (vectorizable_call): Likewise.
13985 (vect_gen_perm_mask_any): Likewise. Add elements in order.
13986 * expmed.c (make_tree): Likewise.
13987 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
13988 a vector passed to build_vector.
13989 (fold_convert_const): Likewise.
13990 (exact_inverse): Likewise.
13991 (fold_ternary_loc): Likewise.
13992 (fold_relational_const): Likewise.
13993 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
13994 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
13995 (const_unop): Likewise. Store the reduction accumulator in a
13996 variable rather than an array.
13997 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
13998 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
13999 the new vector, rather than constructing it after the input arrays.
14000 (native_interpret_vector): Use auto_vec<tree> when building
14001 a vector passed to build_vector. Add elements in order.
14002 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
14003 auto_vec<tree> when building a vector passed to build_vector.
14004 (vect_create_epilog_for_reduction): Likewise.
14005 (vectorizable_induction): Likewise.
14006 (get_initial_def_for_reduction): Likewise. Fix indentation of
14008 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
14010 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
14011 passed to build_vector.
14013 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14014 Alan Hayward <alan.hayward@arm.com>
14015 David Sherwood <david.sherwood@arm.com>
14017 * tree-core.h (tree_base::u): Add an "nelts" field.
14018 (tree_vector): Use VECTOR_CST_NELTS as the length.
14019 * tree.c (tree_size): Likewise.
14020 (make_vector): Initialize VECTOR_CST_NELTS.
14021 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
14022 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
14023 TYPE_VECTOR_SUBPARTS.
14024 * expr.c (const_vector_mask_from_tree): Consistently use "units"
14025 as the number of units, setting it from VECTOR_CST_NELTS.
14026 (const_vector_from_tree): Likewise.
14027 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
14028 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
14029 (fold_negate_expr_1): Likewise.
14030 (fold_convert_const): Likewise.
14031 (const_binop): Likewise. Differentiate the number of output and
14033 (const_unop): Likewise.
14034 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
14035 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
14036 in cases that did the opposite.
14038 2017-09-14 Richard Biener <rguenther@suse.de>
14040 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
14043 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
14045 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
14047 2017-09-14 Jakub Jelinek <jakub@redhat.com>
14050 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
14051 if and where to split a bb, except for splitting before debug insn
14052 sequences followed by non-label real insn. Delete debug insns
14053 in between basic blocks.
14055 * combine.c (make_compound_operation_int): Formatting fixes.
14057 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
14058 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14059 * config/netbsd.h (LINK_EH_SPEC): Likewise.
14060 * config/sol2.h (LINK_EH_SPEC): Likewise.
14061 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14062 * config/s390/linux.h (LINK_SPEC): Likewise.
14063 * config/freebsd.h (LINK_EH_SPEC): Likewise.
14064 * config/openbsd.h (LINK_EH_SPEC): Likewise.
14065 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14066 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
14067 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
14068 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14069 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
14070 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
14072 2017-09-13 Jakub Jelinek <jakub@redhat.com>
14074 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
14076 (ENDFILE_LINUX_SPEC): Likewise.
14077 (LINK_EH_SPEC): Likewise.
14078 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
14079 (LINK_OS_LINUX_SPEC32): Likewise.
14080 (LINK_OS_LINUX_SPEC64): Likewise.
14081 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
14082 (LINK_OS_LINUX_SPEC): Likewise.
14084 2017-09-13 Martin Liska <mliska@suse.cz>
14086 PR middle-end/82154
14087 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
14088 CASE_HIGH is NULL_TREE.
14090 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
14091 Alan Hayward <alan.hayward@arm.com>
14092 David Sherwood <david.sherwood@arm.com>
14094 * target.def (secondary_memory_needed): New hook.
14095 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
14096 instead of SECONDARY_MEMORY_NEEDED.
14097 (secondary_memory_needed_mode): Likewise.
14098 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
14099 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
14100 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
14101 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
14102 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
14103 * doc/tm.texi: Regenerate.
14104 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
14105 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
14106 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14107 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
14108 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
14109 * config/i386/i386.c (inline_secondary_memory_needed): Put the
14110 mode argument first and change the reg_class arguments to reg_class_t.
14111 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
14112 Make static. Update the call to inline_secondary_memory_needed.
14113 (ix86_register_move_cost): Update the call to
14114 inline_secondary_memory_needed.
14115 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14116 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
14118 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
14119 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
14121 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
14122 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
14123 * config/mips/mips.c (mips_secondary_memory_needed): Make static
14124 and match hook interface. Add comment from mips.h.
14125 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14126 * config/mmix/mmix.md (truncdfsf2): Refer to
14127 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
14129 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
14130 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
14131 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14132 (pa_secondary_memory_needed): New function.
14133 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
14134 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
14135 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14136 (pdp11_secondary_memory_needed): Make static and match hook interface.
14137 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
14138 * config/powerpcspe/powerpcspe-protos.h
14139 (rs6000_secondary_memory_needed_ptr): Delete.
14140 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
14142 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14143 (rs6000_option_override_internal): Assign to
14144 targetm.secondary_memory_needed rather than
14145 rs6000_secondary_memory_needed_ptr.
14146 (rs6000_secondary_memory_needed): Match hook interface.
14147 (rs6000_debug_secondary_memory_needed): Likewise.
14148 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
14149 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
14150 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
14151 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14152 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
14153 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
14155 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
14156 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14157 (rs6000_option_override_internal): Assign to
14158 targetm.secondary_memory_needed rather than
14159 rs6000_secondary_memory_needed_ptr.
14160 (rs6000_secondary_memory_needed): Match hook interface.
14161 (rs6000_debug_secondary_memory_needed): Likewise.
14162 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
14163 * config/s390/s390.c (s390_secondary_memory_needed): New function.
14164 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14165 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
14166 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14167 (sparc_secondary_memory_needed): New function.
14168 * lra-constraints.c (check_and_process_move): Refer to
14169 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
14171 (curr_insn_transform): Likewise.
14172 (process_alt_operands): Use targetm.secondary_memory_needed
14173 instead of TARGET_SECONDARY_MEMORY_NEEDED.
14174 (check_secondary_memory_needed_p): Likewise.
14175 (choose_split_class): Likewise.
14176 * reload.c: Unconditionally include code that was previously
14177 conditional on SECONDARY_MEMORY_NEEDED.
14178 (push_secondary_reload): Use targetm.secondary_memory_needed
14179 instead of TARGET_SECONDARY_MEMORY_NEEDED.
14180 (push_reload): Likewise.
14181 * reload1.c: Unconditionally include code that was previously
14182 conditional on SECONDARY_MEMORY_NEEDED.
14183 (choose_reload_regs): Use targetm.secondary_memory_needed
14184 instead of TARGET_SECONDARY_MEMORY_NEEDED.
14185 (gen_reload): Likewise.
14186 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
14188 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
14189 Alan Hayward <alan.hayward@arm.com>
14190 David Sherwood <david.sherwood@arm.com>
14192 * target.def (secondary_memory_needed_mode): New hook:
14193 * targhooks.c (default_secondary_memory_needed_mode): Declare.
14194 * targhooks.h (default_secondary_memory_needed_mode): New function.
14195 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
14196 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
14197 * doc/tm.texi: Regenerate.
14198 * lra-constraints.c (check_and_process_move): Use
14199 targetm.secondary_memory_needed_mode instead of
14200 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
14201 (curr_insn_transform): Likewise.
14202 * reload.c (get_secondary_mem): Likewise.
14203 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14204 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
14206 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14207 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14208 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
14209 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14210 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
14212 * config/powerpcspe/powerpcspe-protos.h
14213 (rs6000_secondary_memory_needed_mode): Delete.
14214 * config/powerpcspe/powerpcspe.c
14215 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14216 (rs6000_secondary_memory_needed_mode): Make static.
14217 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14218 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
14220 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
14222 (rs6000_secondary_memory_needed_mode): Make static.
14223 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14224 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
14225 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14226 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14227 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
14229 (sparc_secondary_memory_needed_mode): New function.
14230 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
14232 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
14234 * config/aarch64/constraints.md (Umq): New constraint.
14235 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
14237 (mov<mode>): Update condition.
14239 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14241 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
14242 when bitposition is the same.
14244 2017-09-13 Richard Biener <rguenther@suse.de>
14246 * dwarf2out.c (output_die_symbol): Remove.
14247 (output_die): Do not output a DIEs symbol.
14249 2017-09-13 Richard Biener <rguenther@suse.de>
14251 PR middle-end/82128
14252 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
14253 default-def to avoid breaking iterator update with the weird
14254 interaction with cgraph_update_edges_for_call_stmt_node.
14256 2017-09-13 Richard Biener <rguenther@suse.de>
14258 * tree-cfg.c (verify_gimple_assign_binary): Add verification
14259 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
14260 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
14261 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
14263 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
14265 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
14266 Disable pc relative literal load irrespective of
14267 TARGET_FIX_ERR_A53_84341 for default.
14269 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
14271 * config/sparc/sparc.c (output_return): Output the source location of
14272 the insn in the delay slot, if any.
14273 (output_sibcall): Likewise.
14275 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
14278 * common.opt (-static-pie): New alias.
14279 (shared): Negate static-pie.
14280 (-no-pie): Update help text.
14282 (static-pie): New option.
14283 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
14284 -static-pie support.
14285 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
14286 (LINK_EH_SPEC): Likewise.
14287 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14288 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
14289 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
14290 * gcc.c (LINK_COMMAND_SPEC): Likewise.
14291 (init_gcc_specs): Likewise.
14292 (init_spec): Likewise.
14293 (display_help): Update help message for -pie.
14294 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
14297 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
14299 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
14300 (movdi_aarch64): Likewise.
14301 (movti_aarch64): Likewise.
14303 2017-09-12 Simon Wright <simon@pushface.org>
14306 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
14307 calculation of the minor version, always output as 0.
14309 2017-09-12 Jakub Jelinek <jakub@redhat.com>
14312 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
14313 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
14314 on it early, rather than manual conversion late. For
14315 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
14316 instead of performing manual conversion.
14318 2017-09-12 Carl Love <cel@us.ibm.com>
14320 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
14321 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
14323 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
14324 VMULOSW): Add definitions.
14325 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14326 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
14327 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
14328 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
14329 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
14331 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
14333 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
14335 (movti_aarch64): Likewise.
14336 (movdf_aarch64): Likewise.
14337 (movtf_aarch64): Likewise.
14338 (load_pairdi): Likewise.
14339 (store_pairdi): Likewise.
14340 (load_pairdf): Likewise.
14341 (store_pairdf): Likewise.
14342 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
14343 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
14344 (ldr_got_small_<mode>): Likewise.
14345 (ldr_got_small_28k_<mode>): Likewise.
14346 (ldr_got_tiny): Likewise.
14347 * config/aarch64/iterators.md (ldst_sz): New.
14348 (ldpstp_sz): Likewise.
14349 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
14351 (thunderx_load): Split load_8 to load_16.
14352 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
14354 (thunderx2t99_storepair_basic): Split store_8 to store_16.
14355 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
14356 (xgene1_store_pair): Split store_8 to store_16.
14357 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
14358 (falkor_st_0_st_sd): Split store_8 to store_16.
14360 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
14362 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
14363 and store1/2/3/4 to store_4/8/12/16.
14364 * config/aarch64/aarch64.md: Update for rename.
14365 * config/arm/arm.md: Likewise.: Likewise.
14366 * config/arm/arm.c: Likewise.
14367 * config/arm/thumb1.md: Likewise.
14368 * config/arm/thumb2.md: Likewise.
14369 * config/arm/vfp.md: Likewise.
14370 * config/arm/arm-generic.md: Likewise.
14371 * config/arm/arm1020e.md: Likewise.
14372 * config/arm/arm1026ejs.md: Likewise.
14373 * config/arm/arm1136jfs.md: Likewise.
14374 * config/arm/arm926ejs.md: Likewise.
14375 * config/arm/cortex-a15.md: Likewise.
14376 * config/arm/cortex-a17.md: Likewise.
14377 * config/arm/cortex-a5.md: Likewise.
14378 * config/arm/cortex-a53.md: Likewise.
14379 * config/arm/cortex-a57.md: Likewise.
14380 * config/arm/cortex-a7.md: Likewise.
14381 * config/arm/cortex-a8.md: Likewise.
14382 * config/arm/cortex-a9.md: Likewise.
14383 * config/arm/cortex-m4.md: Likewise.
14384 * config/arm/cortex-m7.md: Likewise.
14385 * config/arm/cortex-r4.md: Likewise.
14386 * config/arm/exynos-m1.md: Likewise.
14387 * config/arm/fa526.md: Likewise.
14388 * config/arm/fa606te.md: Likewise.
14389 * config/arm/fa626te.md: Likewise.
14390 * config/arm/fa726te.md: Likewise.
14391 * config/arm/fmp626.md: Likewise.
14392 * config/arm/iwmmxt.md: Likewise.
14393 * config/arm/ldmstm.md: Likewise.
14394 * config/arm/marvell-pj4.md: Likewise.
14395 * config/arm/xgene1.md: Likewise.
14396 * config/aarch64/thunderx.md: Likewise.
14397 * config/aarch64/thunderx2t99.md: Likewise.
14398 * config/aarch64/falkor.md: Likewise.
14400 2017-09-12 Martin Liska <mliska@suse.cz>
14402 * attribs.c (private_lookup_attribute): New function.
14403 * attribs.h (private_lookup_attribute): Declared here.
14404 (lookup_attribute): Called from this place.
14406 2017-09-12 Richard Biener <rguenther@suse.de>
14408 PR tree-optimization/82157
14409 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
14410 stmts with side-effects.
14412 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14413 Alan Hayward <alan.hayward@arm.com>
14414 David Sherwood <david.sherwood@arm.com>
14416 * target.def (hard_regno_nregs): New hook.
14417 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
14418 * targhooks.h (default_hard_regno_nregs): Declare.
14419 * targhooks.c (default_hard_regno_nregs): New function.
14420 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
14421 (TARGET_HARD_REGNO_NREGS): ...this hook.
14422 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
14423 (CLASS_MAX_NREGS): Likewise.
14424 * doc/tm.texi: Regenerate.
14425 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
14426 instead of HARD_REGNO_NREGS.
14427 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
14428 HARD_REGNO_NREGS in the comment.
14429 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
14430 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
14431 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
14432 Return an unsigned int.
14433 (TARGET_HARD_REGNO_NREGS): Redefine.
14434 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
14435 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
14436 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
14437 (arc_hard_regno_nregs): New function.
14438 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
14439 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
14440 (arm_hard_regno_nregs): New function.
14441 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
14442 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
14443 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
14444 (TARGET_HARD_REGNO_NREGS): Redefine.
14445 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
14446 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
14447 (HARD_REGNO_NREGS): Delete.
14448 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
14449 (cr16_hard_regno_nregs): New function.
14450 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
14451 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
14452 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
14453 (cris_hard_regno_nregs): New function.
14454 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
14455 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
14456 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
14457 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
14458 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
14459 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
14460 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
14461 (frv_hard_regno_nregs): Make static. Take and return an
14463 (frv_class_max_nregs): Remove outdated copy of documentation.
14464 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
14465 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
14466 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
14467 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
14468 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
14469 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
14470 (TARGET_HARD_REGNO_NREGS): Redefine.
14471 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
14472 (CLASS_MAX_NREGS): Update comment.
14473 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
14474 (ia64_hard_regno_nregs): New function.
14475 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
14476 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
14477 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
14478 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
14479 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
14481 (m32c_hard_regno_nregs): Likewise. Make static.
14482 (TARGET_HARD_REGNO_NREGS): Redefine.
14483 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
14484 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
14485 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
14486 (m68k_hard_regno_nregs): New function.
14487 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
14488 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
14489 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
14490 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
14491 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
14492 Take and return an unsigned int.
14493 (TARGET_HARD_REGNO_NREGS): Redefine.
14494 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
14495 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
14496 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
14497 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
14498 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
14499 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
14500 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
14501 (msp430_hard_regno_nregs): Make static. Take and return an
14503 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
14504 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
14505 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
14506 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
14507 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
14508 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
14509 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
14510 (TARGET_HARD_REGNO_NREGS): Redefine.
14511 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
14512 (PA_HARD_REGNO_NREGS): ...this.
14513 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
14514 (PA_HARD_REGNO_NREGS): ...this.
14515 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
14516 (pa_hard_regno_nregs): New function.
14517 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
14518 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
14519 (pdp11_hard_regno_nregs): New function.
14520 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
14521 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
14522 (rs6000_hard_regno_nregs_hook): New function.
14523 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
14524 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
14525 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
14526 Take and return an unsigned int. Move earlier in file.
14527 (TARGET_HARD_REGNO_NREGS): Redefine.
14528 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
14529 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
14530 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
14531 (rl78_hard_regno_nregs): Make static. Take and return an
14533 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
14534 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
14535 (rs6000_hard_regno_nregs_hook): New function.
14536 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
14537 * config/rx/rx.c (rx_hard_regno_nregs): New function.
14538 (TARGET_HARD_REGNO_NREGS): Redefine.
14539 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
14540 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
14541 instead of HARD_REGNO_NREGS.
14542 (s390_hard_regno_nregs): New function.
14543 (s390_hard_regno_mode_ok): Add comment from s390.h.
14544 (TARGET_HARD_REGNO_NREGS): Redefine.
14545 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
14546 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
14547 (sh_hard_regno_nregs): New function.
14548 (sh_pass_in_reg_p): Use it.
14549 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
14550 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
14551 (sparc_hard_regno_nregs): New function.
14552 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
14553 * config/spu/spu.c (spu_hard_regno_nregs): New function.
14554 (spu_function_arg_advance): Use it, supplying a valid register number.
14555 (TARGET_HARD_REGNO_NREGS): Redefine.
14556 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
14557 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
14558 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
14559 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
14560 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
14561 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
14562 (CLASS_MAX_NREGS): Remove copy of old documentation.
14563 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
14564 (visium_hard_regno_nregs): New function.
14565 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
14566 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
14567 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
14568 xtensa_hard_regno_nregs): New function.
14569 * system.h (HARD_REGNO_NREGS): Poison.
14571 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14573 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
14574 hard_regno_nregs instead of HARD_REGNO_NREGS.
14575 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
14576 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
14577 (c6x_expand_epilogue): Likewise.
14578 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
14579 (frv_read_iacc_argument): Likewise.
14580 * config/sh/sh.c: Include regs.h.
14581 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
14582 (regs_used): Likewise.
14583 (output_stack_adjust): Likewise.
14584 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
14585 * expmed.c: Include regs.h.
14586 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
14587 * ree.c: Include regs.h.
14588 (combine_reaching_defs): Use hard_regno_nregs instead of
14590 (add_removable_extension): Likewise.
14592 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14594 * regs.h (hard_regno_nregs): Turn into a function.
14595 (end_hard_regno): Update accordingly.
14596 * caller-save.c (setup_save_areas): Likewise.
14597 (save_call_clobbered_regs): Likewise.
14598 (replace_reg_with_saved_mem): Likewise.
14599 (insert_restore): Likewise.
14600 (insert_save): Likewise.
14601 * combine.c (can_change_dest_mode): Likewise.
14602 (move_deaths): Likewise.
14603 (distribute_notes): Likewise.
14604 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
14605 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
14606 (rs6000_split_multireg_move): Likewise.
14607 (rs6000_register_move_cost): Likewise.
14608 (rs6000_memory_move_cost): Likewise.
14609 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
14610 (rs6000_split_multireg_move): Likewise.
14611 (rs6000_register_move_cost): Likewise.
14612 (rs6000_memory_move_cost): Likewise.
14613 * cselib.c (cselib_reset_table): Likewise.
14614 (cselib_lookup_1): Likewise.
14615 * emit-rtl.c (set_mode_and_regno): Likewise.
14616 * function.c (aggregate_value_p): Likewise.
14617 * ira-color.c (setup_profitable_hard_regs): Likewise.
14618 (check_hard_reg_p): Likewise.
14619 (calculate_saved_nregs): Likewise.
14620 (assign_hard_reg): Likewise.
14621 (improve_allocation): Likewise.
14622 (calculate_spill_cost): Likewise.
14623 * ira-emit.c (modify_move_list): Likewise.
14624 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
14625 (ira_hard_reg_in_set_p): Likewise.
14626 * ira.c (setup_reg_mode_hard_regset): Likewise.
14627 (clarify_prohibited_class_mode_regs): Likewise.
14628 (check_allocation): Likewise.
14629 * lra-assigns.c (find_hard_regno_for_1): Likewise.
14630 (lra_setup_reg_renumber): Likewise.
14631 (setup_try_hard_regno_pseudos): Likewise.
14632 (spill_for): Likewise.
14633 (assign_hard_regno): Likewise.
14634 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
14635 * lra-constraints.c (in_class_p): Likewise.
14636 (lra_constraint_offset): Likewise.
14637 (simplify_operand_subreg): Likewise.
14638 (lra_constraints): Likewise.
14639 (split_reg): Likewise.
14640 (split_if_necessary): Likewise.
14641 (invariant_p): Likewise.
14642 (inherit_in_ebb): Likewise.
14643 * lra-lives.c (process_bb_lives): Likewise.
14644 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
14645 (get_hard_regs): Likewise.
14646 (do_remat): Likewise.
14647 * lra-spills.c (assign_spill_hard_regs): Likewise.
14648 * mode-switching.c (create_pre_exit): Likewise.
14649 * postreload.c (reload_combine_recognize_pattern): Likewise.
14650 * recog.c (peep2_find_free_register): Likewise.
14651 * regcprop.c (kill_value_regno): Likewise.
14652 (set_value_regno): Likewise.
14653 (copy_value): Likewise.
14654 (maybe_mode_change): Likewise.
14655 (find_oldest_value_reg): Likewise.
14656 (copyprop_hardreg_forward_1): Likewise.
14657 * regrename.c (check_new_reg_p): Likewise.
14658 (regrename_do_replace): Likewise.
14659 * reload.c (push_reload): Likewise.
14660 (combine_reloads): Likewise.
14661 (find_dummy_reload): Likewise.
14662 (operands_match_p): Likewise.
14663 (find_reloads): Likewise.
14664 (find_equiv_reg): Likewise.
14665 (reload_adjust_reg_for_mode): Likewise.
14666 * reload1.c (count_pseudo): Likewise.
14667 (count_spilled_pseudo): Likewise.
14668 (find_reg): Likewise.
14669 (clear_reload_reg_in_use): Likewise.
14670 (free_for_value_p): Likewise.
14671 (allocate_reload_reg): Likewise.
14672 (choose_reload_regs): Likewise.
14673 (reload_adjust_reg_for_temp): Likewise.
14674 (emit_reload_insns): Likewise.
14675 (delete_output_reload): Likewise.
14676 * rtlanal.c (subreg_get_info): Likewise.
14677 * sched-deps.c (sched_analyze_reg): Likewise.
14678 * sel-sched.c (init_regs_for_mode): Likewise.
14679 (mark_unavailable_hard_regs): Likewise.
14680 (choose_best_reg_1): Likewise.
14681 (verify_target_availability): Likewise.
14682 * valtrack.c (dead_debug_insert_temp): Likewise.
14683 * var-tracking.c (track_loc_p): Likewise.
14684 (emit_note_insn_var_location): Likewise.
14685 * varasm.c (make_decl_rtl): Likewise.
14686 * reginfo.c (choose_hard_reg_mode): Likewise.
14687 (init_reg_modes_target): Refer directly to
14688 this_target_regs->x_hard_regno_nregs.
14690 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14692 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
14693 instead of hard_regno_nregs.
14695 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14697 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
14698 end_hard_regno instead of hard_regno_nregs.
14699 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
14700 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
14701 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
14702 * ira-color.c (improve_allocation): Likewise.
14703 * lra-assigns.c (find_hard_regno_for_1): Likewise.
14704 * lra-lives.c (mark_regno_live): Likewise.
14705 (mark_regno_dead): Likewise.
14706 * lra-remat.c (operand_to_remat): Likewise.
14707 * lra.c (collect_non_operand_hard_regs): Likewise.
14708 * postreload.c (reload_combine_note_store): Likewise.
14709 (move2add_valid_value_p): Likewise.
14710 * reload.c (regno_clobbered_p): Likewise.
14712 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14714 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
14716 * config/v850/v850.c (v850_reorg): Likewise.
14717 * reload.c (refers_to_regno_for_reload_p): Likewise.
14718 (find_equiv_reg): Likewise.
14719 * reload1.c (reload_reg_reaches_end_p): Likewise.
14721 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14723 * caller-save.c (add_used_regs): Use REG_NREGS instead of
14725 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
14726 * config/arm/arm.c (output_move_neon): Likewise.
14727 (arm_attr_length_move_neon): Likewise.
14728 (neon_split_vcombine): Likewise.
14729 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
14730 (c6x_mark_reg_written): Likewise.
14731 (c6x_dwarf_register_span): Likewise.
14732 * config/i386/i386.c (ix86_save_reg): Likewise.
14733 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
14734 (rws_access_reg): Likewise.
14735 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
14736 * mode-switching.c (create_pre_exit): Likewise.
14737 * ree.c (combine_reaching_defs): Likewise.
14738 (add_removable_extension): Likewise.
14739 * regcprop.c (find_oldest_value_reg): Likewise.
14740 (copyprop_hardreg_forward_1): Likewise.
14741 * reload.c (reload_inner_reg_of_subreg): Likewise.
14742 (push_reload): Likewise.
14743 (combine_reloads): Likewise.
14744 (find_dummy_reload): Likewise.
14745 (reload_adjust_reg_for_mode): Likewise.
14746 * reload1.c (find_reload_regs): Likewise.
14747 (forget_old_reloads_1): Likewise.
14748 (reload_reg_free_for_value_p): Likewise.
14749 (reload_adjust_reg_for_temp): Likewise.
14750 (emit_reload_insns): Likewise.
14751 (delete_output_reload): Likewise.
14752 * sel-sched.c (choose_best_reg_1): Likewise.
14753 (choose_best_pseudo_reg): Likewise.
14755 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14756 Alan Hayward <alan.hayward@arm.com>
14757 David Sherwood <david.sherwood@arm.com>
14759 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
14760 * target.def (slow_unaligned_access): New hook.
14761 * targhooks.h (default_slow_unaligned_access): Declare.
14762 * targhooks.c (default_slow_unaligned_access): New function.
14763 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
14764 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
14765 * doc/tm.texi: Regenerate.
14766 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
14767 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
14768 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
14770 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
14771 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
14773 (rs6000_slow_unaligned_access): New function.
14774 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
14775 (expand_block_compare): Likewise.
14776 (expand_strn_compare): Likewise.
14777 (rs6000_rtx_costs): Likewise.
14778 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
14779 (riscv_slow_unaligned_access): Likewise.
14780 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
14781 (riscv_slow_unaligned_access_p): ...this and make static.
14782 (riscv_option_override): Update accordingly.
14783 (riscv_slow_unaligned_access): New function.
14784 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
14785 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
14786 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
14787 (rs6000_slow_unaligned_access): New function.
14788 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
14789 (rs6000_rtx_costs): Likewise.
14790 * config/rs6000/rs6000-string.c (expand_block_compare)
14791 (expand_strn_compare): Use targetm.slow_unaligned_access instead
14792 of SLOW_UNALIGNED_ACCESS.
14793 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
14794 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
14795 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
14796 of SLOW_UNALIGNED_ACCESS.
14797 * expmed.c (simple_mem_bitfield_p): Likewise.
14798 * expr.c (alignment_for_piecewise_move): Likewise.
14799 (emit_group_load_1): Likewise.
14800 (emit_group_store): Likewise.
14801 (copy_blkmode_from_reg): Likewise.
14802 (emit_push_insn): Likewise.
14803 (expand_assignment): Likewise.
14804 (store_field): Likewise.
14805 (expand_expr_real_1): Likewise.
14806 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
14807 * lra-constraints.c (simplify_operand_subreg): Likewise.
14808 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
14809 * gimple-ssa-store-merging.c: Likewise in block comment at start
14811 * tree-ssa-strlen.c: Include target.h.
14812 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
14813 of SLOW_UNALIGNED_ACCESS.
14814 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
14816 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14818 PR rtl-optimization/82185
14819 * expmed.c (emit_store_flag_int): Only test tem if it has been
14822 2017-09-12 Richard Biener <rguenther@suse.de>
14824 PR middle-end/82149
14825 * match.pd ((FTYPE) N CMP CST): Fix typo.
14827 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
14829 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
14831 (mips_near_type_p): Add 'short_call' attribute as a synonym
14833 * doc/extend.texi (short_call): Document new function attribute.
14835 2017-09-12 Jakub Jelinek <jakub@redhat.com>
14838 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
14839 assertion check that in the condition.
14840 (get_atomic_generic_size): Likewise. Before testing if parameter
14841 has pointer type, if it has array type, call for C++
14842 default_conversion to perform array-to-pointer conversion.
14844 2017-09-12 Richard Biener <rguenther@suse.de>
14846 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
14847 for operations we cannot scalarize.
14849 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
14851 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
14852 vectors heap vectors. Clean up comments.
14853 Make visited_bbs a reference.
14854 (profitable_jump_thread_path): Make GC
14855 vectors heap vectors. Clean up comments.
14857 (convert_and_register_jump_thread_path): Make GC vectors heap
14859 (check_subpath_and_update_thread_path): Same. Clean up comments.
14860 Make visited_bbs a reference.
14861 (handle_phi): Abstract common code to to
14862 register_jump_thread_path_if_profitable.
14863 Rename VAR_BB to DEF_BB.
14865 Make GC vectors heap vectors.
14866 Make visited_bbs a reference.
14867 (handle_assignment): Same.
14868 (register_jump_thread_path_if_profitable): New.
14869 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
14871 Make GC vectors heap vectors. Clean up comments.
14872 Make visited_bbs a reference.
14873 (find_jump_threads_backwards): Make visited_bbs live in the stack.
14874 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
14877 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
14880 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
14881 words of E_DImode object are reachable by xtensa_uimm8x4 access.
14883 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
14885 Revert r251800 and r251799.
14887 2017-09-11 Martin Jambor <mjambor@suse.cz>
14890 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
14891 arguments in advance.
14892 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
14893 use it to find predecessor edges.
14894 (naive_outof_ssa): Collect vector of predecessors.
14896 2017-09-08 Jason Merrill <jason@redhat.com>
14898 PR c++/70029 - ICE with ref-qualifier and -flto
14899 * langhooks.h (struct lang_hooks_for_types): Add
14900 copy_lang_qualifiers.
14901 * attribs.c (build_type_attribute_qual_variant): Use it.
14902 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
14904 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
14905 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
14907 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
14910 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
14911 (*mulsi3_sp64): New instruction.
14912 (mulsi3): New expander.
14914 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
14916 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
14918 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
14920 * sancov.c: Include memmodel.h.
14922 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
14925 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
14928 2017-09-07 Carl Love <cel@us.ibm.com>
14930 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
14931 the sldi instruction.
14933 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
14935 * sancov.c: Include tm_p.h.
14937 2017-09-07 Jakub Jelinek <jakub@redhat.com>
14940 * output.h (switch_to_other_text_partition): New declaration.
14941 * varasm.c (switch_to_other_text_partition): New function.
14942 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
14943 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
14944 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
14945 to the other text partition before emitting LCL label and switch back
14946 after emitting the word after it.
14948 2017-09-07 Richard Biener <rguenther@suse.de>
14950 * passes.def (pass_split_crit_edges): Remove instance before PRE.
14951 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
14952 critical edges here, after loop init.
14953 (pass_data_pre): Remove PROP_no_crit_edges flags.
14954 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
14955 for valueization of call args to avoid leaking VN_TOP.
14956 (visit_use): Assert we do not visit default defs.
14957 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
14958 Use error_mark_node to more easily detect leaking VN_TOP.
14959 All default-defs are varying, not VN_TOP. Mark them visited.
14960 (run_scc_vn): Make code match comment.
14962 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
14964 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
14965 OPTION_MASK_FLOAT128_KEYWORD.
14966 (POWERPC_MASKS): Likewise.
14967 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
14968 support for the -mfloat128-type option, and make -mfloat128
14969 default on PowerPC Linux systems. Define or undefine
14970 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
14971 Define __float128 to be __ieee128 if IEEE 128-bit support is
14972 enabled, or undefine it.
14973 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
14974 Delete defining __FLOAT128_TYPE__.
14975 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
14976 -mfloat128-type option and make -mfloat128 default on PowerPC
14978 (TARGET_FLOAT128_TYPE): Likewise.
14979 (-mfloat128-type): Likewise.
14980 * config/rs6000/rs6000.c (rs6000_option_override_internal):
14981 Delete the -mfloat128-type option and make -mfloat128 default on
14982 PowerPC Linux systems. Always use __ieee128 to be the keyword for
14983 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
14984 128-bit floating point is enabled. Change tests from using
14985 -mfloat128-type to -mfloat128.
14986 (rs6000_mangle_type): Use the correct mangling for the __float128
14987 type even if normal long double is restricted to 64-bits.
14988 (floatn_mode): Enable the _Float128 type by default on VSX Linux
14990 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
14991 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
14992 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
14994 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
14995 documentation for -mfloat128.
14997 2017-09-06 Olivier Hainque <hainque@adacore.com>
14999 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
15001 2017-09-06 Wish Wu <wishwu007@gmail.com>
15002 Jakub Jelinek <jakub@redhat.com>
15004 * asan.c (initialize_sanitizer_builtins): Add
15005 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
15006 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
15007 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
15008 BT_FN_VOID_UINT64_PTR variables.
15009 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
15010 (BT_FN_VOID_UINT16_UINT16): Likewise.
15011 (BT_FN_VOID_UINT32_UINT32): Likewise.
15012 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
15013 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
15014 (BT_FN_VOID_UINT64_PTR): Likewise.
15015 * common.opt (flag_sanitize_coverage): New variable.
15016 (fsanitize-coverage=trace-pc): Remove.
15017 (fsanitize-coverage=): Add.
15018 * flag-types.h (enum sanitize_coverage_code): New enum.
15019 * fold-const.c (fold_range_test): Disable non-short-circuit
15020 optimization if flag_sanitize_coverage.
15021 (fold_truth_andor): Likewise.
15022 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
15023 * opts.c (COVERAGE_SANITIZER_OPT): Define.
15024 (coverage_sanitizer_opts): New array.
15025 (get_closest_sanitizer_option): Add OPTS argument, handle also
15026 OPT_fsanitize_coverage_.
15027 (parse_sanitizer_options): Adjusted to also handle
15028 OPT_fsanitize_coverage_.
15029 (common_handle_option): Add OPT_fsanitize_coverage_.
15030 * sancov.c (instrument_comparison, instrument_switch): New function.
15031 (sancov_pass): Add trace-cmp support.
15032 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
15033 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
15034 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
15035 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
15036 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
15037 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
15038 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
15039 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
15040 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
15041 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
15043 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
15045 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
15046 error. Only quit immediately if parsing is complete.
15047 (BEGIN): Initialize fatal_err and parse_done.
15048 (begin fpu, end fpu): Check number of arguments.
15049 (begin arch, end arch): Likewise.
15050 (begin cpu, end cpu): Likewise.
15051 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
15052 (optalias): Likewise.
15054 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
15056 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
15057 * config/arm/arm-isa.h: Delete. Move definitions to ...
15058 * arm-cpus.in: ... here. Use new feature and fgroup values.
15059 * config/arm/arm.c (arm_option_override): Use lower case for feature
15061 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
15062 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
15063 * config/arm/parsecpu.awk (END): Add new command 'isa'.
15065 (print_isa_bits_for): New function.
15066 (gen_isa): New function.
15067 (gen_comm_data): Use print_isa_bits_for.
15068 (define feature): New keyword.
15069 (define fgroup): New keyword.
15070 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
15071 (arm-isa.h): Add rule to generate file.
15072 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
15073 case for feature bit names.
15075 2017-09-06 Richard Biener <rguenther@suse.de>
15077 * tree-ssa-pre.c (NECESSARY): Remove.
15078 (create_expression_by_pieces): Do not touch pass-local flags.
15079 (insert_into_preds_of_block): Likewise.
15080 (do_pre_regular_insertion): Likewise.
15081 (eliminate_insert): Likewise.
15082 (eliminate_dom_walker::before_dom_children): Likewise.
15083 (fini_eliminate): Do not look at inserted_exprs.
15084 (mark_operand_necessary): Remove.
15085 (remove_dead_inserted_code): Replace with simple work-list
15086 algorithm based on inserted_exprs and SSA uses.
15087 (pass_pre::execute): Re-order fini_eliminate and
15088 remove_dead_inserted_code.
15090 2017-09-06 Olivier Hainque <hainque@adacore.com>
15092 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
15093 for VxWorks 7. Adjust surrounding comments.
15095 2017-09-06 Richard Biener <rguenther@suse.de>
15097 * gimple-ssa-strength-reduction.c
15098 (find_candidates_dom_walker::before_dom_children): Also allow
15101 2017-09-06 Richard Biener <rguenther@suse.de>
15103 PR tree-optimization/82108
15104 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
15105 for gap in the non-permutation SLP case.
15107 2017-09-06 Martin Jambor <mjambor@suse.cz>
15109 PR tree-optimization/82078
15110 * tree-sra.c (sort_and_splice_var_accesses): Move call to
15111 add_access_to_work_queue...
15112 (build_accesses_from_assign): ...here.
15113 (propagate_all_subaccesses): Make sure racc is the group
15114 representative, if there is one.
15116 2017-09-06 Jakub Jelinek <jakub@redhat.com>
15118 PR middle-end/82095
15119 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
15122 2017-09-06 Richard Biener <rguenther@suse.de>
15124 * gimple-ssa-strength-reduction.c
15125 (find_candidates_dom_walker::before_doom_children): Use a
15126 type and not a mode check.
15128 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
15131 * config/arm/predicates.md (arm_general_adddi_operand): Create new
15133 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
15135 2017-09-05 Jeff Law <law@redhat.com>
15137 PR tree-optimization/64910
15138 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
15139 cases where we have 3 or more operands.
15141 2017-09-05 Jakub Jelinek <jakub@redhat.com>
15143 PR middle-end/81768
15144 * omp-low.c (lower_omp_for): Recompute tree invariant if
15145 gimple_omp_for_initial/final is ADDR_EXPR.
15147 PR middle-end/81768
15148 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
15149 into gimple val before gimplification fo the COND_EXPR.
15151 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
15153 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
15154 REGION_COPY argument.
15155 (thread_through_all_blocks): Remove unused argument to
15156 duplicate_thread_path.
15158 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15159 Alan Hayward <alan.hayward@arm.com>
15160 David Sherwood <david.sherwood@arm.com>
15162 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
15163 Take a scalar_mode rather than a machine_mode.
15164 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
15165 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
15166 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
15167 (aarch64_gen_adjusted_ldpstp): Likewise.
15168 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
15170 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15171 Alan Hayward <alan.hayward@arm.com>
15172 David Sherwood <david.sherwood@arm.com>
15174 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
15175 Take a scalar_int_mode instead of a machine_mode.
15176 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
15177 (aarch64_output_scalar_simd_mov_immediate): Likewise.
15178 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
15179 (aarch64_simd_attr_length_rglist): Delete.
15180 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
15181 a scalar_int_mode instead of a machine_mode.
15182 (aarch64_add_offset): Likewise.
15183 (aarch64_internal_mov_immediate): Likewise
15184 (aarch64_add_constant_internal): Likewise.
15185 (aarch64_add_constant): Likewise.
15186 (aarch64_movw_imm): Likewise.
15187 (aarch64_rtx_arith_op_extract_p): Likewise.
15188 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
15189 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
15190 Remove assert that the mode isn't a vector.
15191 (aarch64_output_scalar_simd_mov_immediate): Likewise.
15192 (aarch64_expand_mov_immediate): Update calls after above changes.
15193 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
15194 (aarch64_and_bitmask_imm): Check for scalar integer modes.
15195 (aarch64_move_imm): Likewise.
15196 (aarch64_can_const_movi_rtx_p): Likewise.
15197 (aarch64_strip_extend): Likewise.
15198 (aarch64_extr_rtx_p): Likewise.
15199 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
15200 a CONST_INT when the mode parameter is VOIDmode.
15201 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
15203 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15205 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
15206 * stor-layout.c (bitwise_mode_for_mode): Likewise.
15207 (bitwise_type_for_mode): Update accordingly.
15209 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15211 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
15212 * stor-layout.c (mode_for_size_tree): Likewise.
15213 (mode_for_array): Update accordingly.
15214 (layout_decl): Likewise.
15215 (compute_record_mode): Likewise. Only set the mode once.
15217 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15219 * target.def (get_mask_mode): Change return type to opt_mode.
15221 * doc/tm.texi: Regenerate.
15222 * targhooks.h (default_get_mask_mode): Return an opt_mode.
15223 * targhooks.c (default_get_mask_mode): Likewise.
15224 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
15225 * optabs-query.c (can_vec_mask_load_store_p): Update use of
15226 targetm.get_mask_mode.
15227 * tree.c (build_truth_vector_type): Likewise.
15229 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15231 * machmode.h (mode_for_vector): Return an opt_mode.
15232 * stor-layout.c (mode_for_vector): Likewise.
15233 (mode_for_int_vector): Update accordingly.
15234 (layout_type): Likewise.
15235 * config/i386/i386.c (emit_memmov): Likewise.
15236 (ix86_expand_set_or_movmem): Likewise.
15237 (ix86_expand_vector_init): Likewise.
15238 (ix86_get_mask_mode): Likewise.
15239 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
15241 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
15242 * expmed.c (extract_bit_field_1): Likewise.
15243 * expr.c (expand_expr_real_2): Likewise.
15244 * optabs-query.c (can_vec_perm_p): Likewise.
15245 (can_vec_mask_load_store_p): Likewise.
15246 * optabs.c (expand_vec_perm): Likewise.
15247 * targhooks.c (default_get_mask_mode): Likewise.
15248 * tree-vect-stmts.c (vectorizable_store): Likewise.
15249 (vectorizable_load): Likewise.
15250 (get_vectype_for_scalar_type_and_size): Likewise.
15252 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15254 * machmode.h (mode_for_int_vector): New function.
15255 * stor-layout.c (mode_for_int_vector): Likewise.
15256 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
15257 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
15258 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
15259 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
15260 (s390_expand_vcond): Likewise.
15262 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15264 * machmode.h (opt_machine_mode): New type.
15265 (opt_mode<T>): Allow construction from anything that can be
15267 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
15268 (mode_for_size): Return an opt_machine_mode.
15269 * stor-layout.c (mode_for_size): Likewise.
15270 (mode_for_size_tree): Update call accordingly.
15271 (bitwise_mode_for_mode): Likewise.
15272 (make_fract_type): Likewise.
15273 (make_accum_type): Likewise.
15274 * caller-save.c (replace_reg_with_saved_mem): Update call
15276 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15277 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15278 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15279 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15280 * expmed.c (extract_bit_field_1): Likewise.
15281 * reload.c (get_secondary_mem): Likewise.
15282 * varasm.c (assemble_integer): Likewise.
15283 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
15286 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15288 * machmode.h (decimal_float_mode_for_size): New function.
15289 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
15290 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
15291 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
15292 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
15293 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
15294 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
15296 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15298 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
15299 (get_builtin_sync_mode): Likewise.
15300 (expand_ifn_atomic_compare_exchange): Likewise.
15301 (expand_builtin_atomic_clear): Likewise.
15302 (expand_builtin_atomic_test_and_set): Likewise.
15303 (fold_builtin_atomic_always_lock_free): Likewise.
15304 * calls.c (compute_argument_addresses): Likewise.
15305 (emit_library_call_value_1): Likewise.
15306 (store_one_arg): Likewise.
15307 * combine.c (combine_instructions): Likewise.
15308 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
15309 * config/arm/arm.c (arm_function_value): Likewise.
15310 (aapcs_allocate_return_reg): Likewise.
15311 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
15312 * config/i386/i386.c (construct_container): Likewise.
15313 (ix86_gimplify_va_arg): Likewise.
15314 (ix86_expand_sse_cmp): Likewise.
15315 (emit_memmov): Likewise.
15316 (emit_memset): Likewise.
15317 (expand_small_movmem_or_setmem): Likewise.
15318 (ix86_expand_pextr): Likewise.
15319 (ix86_expand_pinsr): Likewise.
15320 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
15321 * config/microblaze/microblaze.c (microblaze_block_move_straight):
15323 * config/mips/mips.c (mips_function_value_1) Likewise.
15324 (mips_block_move_straight): Likewise.
15325 (mips_expand_ins_as_unaligned_store): Likewise.
15326 * config/powerpcspe/powerpcspe.c
15327 (rs6000_darwin64_record_arg_advance_flush): Likewise.
15328 (rs6000_darwin64_record_arg_flush): Likewise.
15329 * config/rs6000/rs6000.c
15330 (rs6000_darwin64_record_arg_advance_flush): Likewise.
15331 (rs6000_darwin64_record_arg_flush): Likewise.
15332 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
15333 (sparc_function_value_1): Likewise.
15334 * config/spu/spu.c (adjust_operand): Likewise.
15335 (spu_emit_branch_or_set): Likewise.
15336 (arith_immediate_p): Likewise.
15337 * emit-rtl.c (gen_lowpart_common): Likewise.
15338 * expr.c (expand_expr_real_1): Likewise.
15339 * function.c (assign_parm_setup_block): Likewise.
15340 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
15341 * reload1.c (alter_reg): Likewise.
15342 * stor-layout.c (mode_for_vector): Likewise.
15343 (layout_type): Likewise.
15345 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15347 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
15348 (spu_convert_move): Likewise.
15349 * lower-subreg.c (resolve_simple_move): Likewise.
15351 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15354 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
15355 define_insn to a define_expand.
15356 (altivec_vsum2sws_direct): New define_insn.
15357 (altivec_vsumsws): Convert from a define_insn to a define_expand.
15359 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
15361 * config/arm/arm.c (arm_option_params_internal): Improve setting of
15364 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
15369 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
15370 realign stack if stack alignment needed is less than incoming
15373 2017-09-05 Marek Polacek <polacek@redhat.com>
15376 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
15379 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
15381 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
15383 2017-09-05 Richard Biener <rguenther@suse.de>
15385 PR tree-optimization/82084
15386 * fold-const.c (can_native_encode_string_p): Handle wide characters.
15388 2017-09-05 Richard Biener <rguenther@suse.de>
15390 PR tree-optimization/82102
15391 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
15393 2017-09-05 Martin Liska <mliska@suse.cz>
15395 PR tree-optimization/82032
15396 * tree-cfg.c (generate_range_test): New function.
15397 * tree-cfg.h (generate_range_test): Declared here.
15398 * tree-cfgcleanup.c (convert_single_case_switch): New function.
15399 (cleanup_control_expr_graph): Use it.
15400 * tree-switch-conversion.c (try_switch_expansion): Remove
15402 (emit_case_nodes): Use generate_range_test.
15404 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
15407 * config/i386/i386.md (*<btsc><mode>_mask): Add
15408 TARGET_USE_BT to insn constraint.
15409 (*btr<mode>_mask): Ditto.
15411 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
15413 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
15414 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
15416 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
15419 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
15420 TARGET_NEON and TARGET_IWMMXT.
15421 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
15422 TARGET_NEON and TARGET_IWMMXT.
15423 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
15425 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
15427 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
15428 (ix86_rewrite_tls_address): Ditto.
15429 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
15430 (ix86_rewrite_tls_address_1): Ditto.
15431 (ix86_rewrite_tls_address): Ditto.
15432 * config/i386/predicates.md (tls_address_pattern): New predicate.
15433 * config/i386/i386.md (TLS address splitter): New splitter.
15435 2017-09-04 Richard Biener <rguenther@suse.de>
15437 PR tree-optimization/82084
15438 * fold-const.h (can_native_encode_string_p): Declare.
15439 * fold-const.c (can_native_encode_string_p): Factor out from ...
15440 (native_encode_string): ... here.
15441 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
15442 vectorizing stores from constants we later cannot handle.
15444 2017-09-04 Marek Polacek <polacek@redhat.com>
15447 * doc/invoke.texi: Update -Wtautological-compare documentation.
15449 2017-09-04 Jeff Law <law@redhat.com>
15451 PR tree-optimization/64910
15452 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
15453 swap the first and last operand if the last is a constant.
15455 2017-09-04 Marek Polacek <polacek@redhat.com>
15458 * convert.c (do_narrow): When sanitizing signed integer overflows,
15459 bail out for signed types.
15460 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
15462 2017-09-04 Richard Biener <rguenther@suse.de>
15464 PR tree-optimization/82060
15465 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
15466 Move devirtualization after stmt folding and before EH/AB/noreturn
15467 cleanup to get the stmt refs canonicalized. Use a bool instead
15468 of gimple_modified_p since that doesn't work for NOPs. Schedule
15469 NOPs generated by folding for removal.
15471 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
15472 Alan Hayward <alan.hayward@arm.com>
15473 David Sherwood <david.sherwood@arm.com>
15475 * coretypes.h (pad_direction): New enum.
15476 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
15477 (FUNCTION_ARG_PADDING): Likewise.
15478 * target.def (function_arg_padding): New hook.
15479 * targhooks.h (default_function_arg_padding): Declare.
15480 * targhooks.c (default_function_arg_padding): New function.
15481 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
15482 (TARGET_FUNCTION_ARG_PADDING): ...this.
15483 * doc/tm.texi: Regenerate.
15484 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
15485 instead of direction.
15486 (compute_argument_addresses): Likewise.
15487 (load_register_parameters): Likewise.
15488 (emit_library_call_value_1): Likewise.
15489 (store_one_arg): Use targetm.calls.function_arg_padding instead
15490 of FUNCTION_ARG_PADDING.
15491 (must_pass_in_stack_var_size_or_pad): Likewise.
15492 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
15493 (emit_group_store): Likewise.
15494 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
15495 instead of FUNCTION_ARG_PADDING.
15496 (emit_push_insn): Likewise, and propagate enum change throughout
15498 * function.h (direction): Delete.
15499 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
15501 * function.c (assign_parm_find_stack_rtl): Likewise.
15502 (assign_parm_setup_block_p): Likewise.
15503 (assign_parm_setup_block): Likewise.
15504 (gimplify_parameters): Likewise.
15505 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
15506 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
15508 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
15509 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
15510 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
15511 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
15512 (aarch64_function_arg_padding): ...this new function.
15513 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
15514 (TARGET_FUNCTION_ARG_PADDING): Redefine.
15515 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
15516 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
15517 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
15518 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
15519 (arm_pad_arg_upward): Replace with...
15520 (arm_function_arg_padding): ...this new function.
15521 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
15523 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
15524 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
15525 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
15526 (ia64_hpux_function_arg_padding): Replace with...
15527 (ia64_function_arg_padding): ...this new function. Use pad_direction
15528 instead of direction. Check for TARGET_HPUX.
15529 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
15530 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
15531 (iq2000_function_arg_padding): New function.
15532 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
15533 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
15534 (mips_function_arg_padding): ...this new function.
15535 (mips_pad_reg_upward): Update accordingly.
15536 (TARGET_FUNCTION_ARG_PADDING): Redefine.
15537 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
15538 targetm.calls.function_arg_padding.
15539 (FUNCTION_ARG_PADDING): Delete.
15540 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
15541 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
15542 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
15543 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
15544 (nios2_block_reg_padding): Return pad_direction instead of direction.
15545 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
15546 instead of direction.
15547 (nios2_function_arg_padding): Likewise. Make static.
15548 (TARGET_FUNCTION_ARG_PADDING): Redefine.
15549 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
15550 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
15551 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
15552 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
15553 (pa_function_arg_padding): Make static. Return pad_direction instead
15555 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
15556 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
15557 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
15558 instead of direction. Use targetm.calls.function_arg_padding.
15559 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
15560 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
15561 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
15562 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
15563 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
15565 (function_arg_padding): Rename to...
15566 (rs6000_function_arg_padding): ...this. Make static. Return
15567 pad_direction instead of direction.
15568 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
15569 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
15570 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
15571 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
15572 instead of direction. Use targetm.calls.function_arg_padding.
15573 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
15574 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
15575 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
15576 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
15577 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
15578 (function_arg_padding): Rename to...
15579 (rs6000_function_arg_padding): ...this. Make static. Return
15580 pad_direction instead of direction.
15581 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
15582 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
15583 * config/s390/s390.c (s390_function_arg_padding): New function.
15584 (TARGET_FUNCTION_ARG_PADDING): Redefine.
15585 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
15586 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
15587 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
15588 (function_arg_padding): Rename to...
15589 (sparc_function_arg_padding): ...this. Make static. Return
15590 pad_direction instead of direction.
15591 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
15592 * config/spu/spu.c (spu_function_arg_padding): New function.
15593 (TARGET_FUNCTION_ARG_PADDING): Redefine.
15594 * system.h (FUNCTION_ARG_PADDING): Poison.
15596 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
15597 Alan Hayward <alan.hayward@arm.com>
15598 David Sherwood <david.sherwood@arm.com>
15600 * target.def (modes_tieable_p): New hook.
15601 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
15602 (TARGET_MODES_TIEABLE_P): ...this.
15603 * doc/tm.texi.in: Regenerate.
15604 * hooks.h (hook_bool_mode_mode_true): Declare.
15605 * hooks.c (hook_bool_mode_mode_true): New function.
15606 * combine.c (subst): Use targetm.modes_tieable_p instead of
15608 * dse.c (find_shift_sequence): Likewise.
15609 * expmed.c (extract_low_bits): Likewise.
15610 * lower-subreg.c: Include target.h.
15611 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
15613 * rtlanal.c (rtx_cost): Likewise.
15614 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
15615 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
15616 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
15617 (TARGET_MODES_TIEABLE_P): Redefine.
15618 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
15619 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
15620 (TARGET_MODES_TIEABLE_P): Redefine.
15621 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
15622 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
15623 (arc_modes_tieable_p): New function.
15624 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
15625 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
15626 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
15627 (arm_modes_tieable_p): Make static.
15628 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
15629 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
15630 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
15631 (TARGET_MODES_TIEABLE_P): Redefine.
15632 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
15633 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
15634 (TARGET_MODES_TIEABLE_P): Redefine.
15635 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
15636 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
15637 (cr16_modes_tieable_p): New function.
15638 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
15639 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
15640 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
15641 (TRULY_NOOP_TRUNCATION): Update comment.
15642 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
15643 (TRULY_NOOP_TRUNCATION): Update comment.
15644 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
15645 (frv_modes_tieable_p): New function.
15646 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
15647 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
15648 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
15649 (TARGET_MODES_TIEABLE_P): Redefine.
15650 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
15651 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
15652 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
15653 (TARGET_MODES_TIEABLE_P): Redefine.
15654 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
15655 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
15656 (ia64_modes_tieable_p): New function.
15657 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
15658 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
15659 (iq2000_modes_tieable_p): New function.
15660 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
15661 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
15662 (lm32_modes_tieable_p): New function.
15663 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
15664 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
15665 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
15666 (TARGET_MODES_TIEABLE_P): Redefine.
15667 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
15668 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
15669 (m32r_modes_tieable_p): New function.
15670 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
15671 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
15672 (m68k_modes_tieable_p): New function.
15673 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
15674 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
15675 (mcore_modes_tieable_p): New function.
15676 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
15677 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
15679 (TARGET_MODES_TIEABLE_P): Redefine.
15680 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
15681 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
15682 * config/mips/mips.c (mips_modes_tieable_p): Make static.
15683 (TARGET_MODES_TIEABLE_P): Redefine.
15684 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
15685 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
15686 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
15687 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
15688 (mn10300_modes_tieable_p): ...this and make static.
15689 (TARGET_MODES_TIEABLE_P): Redefine.
15690 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
15691 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
15692 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
15693 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
15694 (msp430_modes_tieable_p): Make static.
15695 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
15696 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
15697 (TARGET_MODES_TIEABLE_P): Redefine.
15698 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
15699 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
15700 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
15701 (TARGET_MODES_TIEABLE_P): Redefine.
15702 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
15703 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
15704 * config/pa/pa.c (pa_modes_tieable_p): Make static.
15705 (TARGET_MODES_TIEABLE_P): Redefine.
15706 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
15707 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
15708 (pdp11_modes_tieable_p): New function.
15709 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
15710 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
15711 (rs6000_modes_tieable_p): New function.
15712 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
15713 * config/powerpcspe/powerpcspe.md: Update comment.
15714 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
15715 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
15716 (TARGET_MODES_TIEABLE_P): Redefine.
15717 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
15718 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
15719 (rl78_modes_tieable_p): New function.
15720 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
15721 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
15722 (rs6000_modes_tieable_p): New function.
15723 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
15724 * config/rs6000/rs6000.md: Update comment.
15725 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
15726 * config/rx/rx.c (rx_modes_tieable_p): New function.
15727 (TARGET_MODES_TIEABLE_P): Redefine.
15728 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
15729 * config/s390/s390.c (s390_modes_tieable_p): New function.
15730 (TARGET_MODES_TIEABLE_P): Redefine.
15731 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
15732 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
15733 (sh_modes_tieable_p): New function.
15734 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
15735 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
15736 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
15737 (sparc_modes_tieable_p): Make static.
15738 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
15739 * config/spu/spu.c (spu_modes_tieable_p): New function.
15740 (TARGET_MODES_TIEABLE_P): Redefine.
15741 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
15742 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
15743 (TARGET_MODES_TIEABLE_P): Redefine.
15744 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
15745 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
15746 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
15747 * config/v850/v850.c (v850_modes_tieable_p): New function.
15748 (TARGET_MODES_TIEABLE_P): Redefine.
15749 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
15750 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
15751 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
15752 (visium_modes_tieable_p): New function.
15753 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
15754 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
15755 (xtensa_modes_tieable_p): New function.
15756 * system.h (MODES_TIEABLE_P): Poison.
15758 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
15759 Alan Hayward <alan.hayward@arm.com>
15760 David Sherwood <david.sherwood@arm.com>
15762 * target.def (hard_regno_mode_ok): New hook.
15763 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
15764 (TARGET_HARD_REGNO_MODE_OK): ...this.
15765 * doc/tm.texi.in: Regenerate.
15766 * hooks.h (hook_bool_uint_mode_true): Declare.
15767 * hooks.c (hook_bool_uint_mode_true): New function.
15768 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
15769 HARD_REGNO_MODE_OK.
15770 * genpreds.c (write_insn_preds_c): Add an include of target.h.
15771 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
15772 instead of HARD_REGNO_MODE_OK.
15773 * caller-save.c: Include target.h.
15774 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
15775 HARD_REGNO_MODE_OK.
15776 * combine.c (can_combine_p): Likewise.
15777 (combinable_i3pat): Likewise.
15778 (can_change_dest_mode): Likewise.
15779 * expr.c (init_expr_target): Likewise.
15780 (convert_move): Likewise.
15781 (convert_modes): Likewise.
15782 * ira.c (setup_prohibited_class_mode_regs): Likewise.
15783 (setup_prohibited_mode_move_regs): Likewise.
15784 * ira.h (target_ira): Likewise.
15785 * lra-assigns.c (find_hard_regno_for_1): Likewise.
15786 * lra-constraints.c (process_alt_operands): Likewise.
15787 (split_reg): Likewise.
15788 * recog.c (peep2_find_free_register): Likewise.
15789 * ree.c (combine_reaching_defs): Likewise.
15790 * regcprop.c (maybe_mode_change): Likewise.
15791 * reginfo.c (init_reg_sets_1): Likewise.
15792 (choose_hard_reg_mode): Likewise.
15793 (simplifiable_subregs): Likewise.
15794 * regrename.c (check_new_reg_p): Likewise.
15795 * reload.c (find_valid_class): Likewise.
15796 (find_valid_class_1): Likewise.
15797 (reload_inner_reg_of_subreg): Likewise.
15798 (push_reload): Likewise.
15799 (combine_reloads): Likewise.
15800 (find_dummy_reload): Likewise.
15801 (find_reloads): Likewise.
15802 * reload1.c (find_reg): Likewise.
15803 (set_reload_reg): Likewise.
15804 (allocate_reload_reg): Likewise.
15805 (choose_reload_regs): Likewise.
15806 (reload_adjust_reg_for_temp): Likewise.
15807 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
15808 (simplify_subreg_regno): Likewise.
15809 * sel-sched.c (init_regs_for_mode): Likewise.
15810 * varasm.c (make_decl_rtl): Likewise.
15811 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
15812 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
15813 HARD_REGNO_MODE_OK.
15814 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
15815 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
15816 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15817 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
15818 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
15819 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15820 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
15821 (arc_mode_class): Delete.
15822 (HARD_REGNO_MODE_OK): Delete.
15823 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15824 (arc_hard_regno_mode_ok): Rename old array to...
15825 (arc_hard_regno_mode_ok_modes): ...this.
15826 (arc_conditional_register_usage): Update accordingly.
15827 (arc_mode_class): Make static.
15828 (arc_hard_regno_mode_ok): New function.
15829 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
15830 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
15831 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15832 (arm_hard_regno_mode_ok): Make static.
15833 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
15834 HARD_REGNO_MODE_OK.
15835 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
15836 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
15837 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
15839 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15840 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
15841 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
15842 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
15843 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
15844 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15845 * config/bfin/predicates.md (valid_reg_operand): Use
15846 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
15847 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
15848 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
15849 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15850 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
15851 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
15852 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15853 (cr16_hard_regno_mode_ok): Make static and return a bool.
15854 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
15855 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15856 (cris_hard_regno_mode_ok): New function.
15857 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
15858 (epiphany_mode_class): Delete.
15859 (HARD_REGNO_MODE_OK): Delete.
15860 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
15861 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15862 (hard_regno_mode_ok): Rename to...
15863 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
15864 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
15865 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
15866 HARD_REGNO_MODE_OK.
15867 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
15868 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
15869 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15870 (frv_hard_regno_mode_ok): Make static and return a bool.
15871 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
15872 HARD_REGNO_MODE_OK.
15873 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
15874 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
15875 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
15876 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
15878 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15879 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
15880 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
15881 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
15883 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15884 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
15885 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15886 (ia64_hard_regno_mode_ok): New function.
15887 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
15888 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15889 (iq2000_hard_regno_mode_ok): New function.
15890 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
15891 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15892 (lm32_hard_regno_mode_ok): New function.
15893 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
15894 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
15895 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
15896 instead of HARD_REGNO_MODE_OK.
15897 (m32c_hard_regno_ok): Rename to...
15898 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
15899 (m32c_cannot_change_mode_class): Update accordingly.
15900 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15901 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
15902 (m32r_mode_class): Delete.
15903 (HARD_REGNO_MODE_OK): Delete.
15904 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15905 (m32r_hard_regno_mode_ok): Rename to...
15906 (m32r_hard_regno_modes): ...this.
15907 (m32r_mode_class): Make static.
15908 (m32r_hard_regno_mode_ok): New function.
15909 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
15910 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
15911 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15912 (m68k_hard_regno_mode_ok): Make static.
15913 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
15914 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15915 (mcore_hard_regno_mode_ok): New function.
15916 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
15917 (HARD_REGNO_MODE_OK): Delete.
15918 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
15920 (microblaze_hard_regno_mode_ok_p): ...this and make static.
15921 (microblaze_hard_regno_mode_ok): New function.
15922 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15923 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
15924 (mips_hard_regno_mode_ok): Delete.
15925 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
15926 (mips_hard_regno_mode_ok_p): ...this and make static.
15927 (mips_hard_regno_mode_ok_p): Rename to...
15928 (mips_hard_regno_mode_ok_uncached): ...this.
15929 (mips_hard_regno_mode_ok): New function.
15930 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
15931 of HARD_REGNO_MODE_OK.
15932 (mips_option_override): Update after above name changes.
15933 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15934 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
15935 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
15936 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
15937 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
15938 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15939 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
15940 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
15941 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
15942 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15943 (msp430_hard_regno_mode_ok): Make static and return a bool.
15944 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
15945 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
15946 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
15948 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15949 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
15950 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
15951 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
15952 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
15953 (PA_HARD_REGNO_MODE_OK): ...this
15954 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
15955 (PA_HARD_REGNO_MODE_OK): ...this.
15956 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15957 (pa_hard_regno_mode_ok): New function.
15958 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
15959 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15960 (pdp11_hard_regno_mode_ok): New function.
15961 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
15962 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
15964 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
15966 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15967 (rs6000_hard_regno_mode_ok): Rename to...
15968 (rs6000_hard_regno_mode_ok_uncached): ...this.
15969 (rs6000_init_hard_regno_mode_ok): Update accordingly.
15970 (rs6000_hard_regno_mode_ok): New function.
15971 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
15972 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
15973 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
15974 (riscv_hard_regno_mode_ok): ...this and make static.
15975 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15976 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
15977 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
15978 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15979 (rl78_hard_regno_mode_ok): Make static and return bool.
15980 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
15981 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
15983 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
15984 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15985 (rs6000_hard_regno_mode_ok): Rename to...
15986 (rs6000_hard_regno_mode_ok_uncached): ...this.
15987 (rs6000_init_hard_regno_mode_ok): Update accordingly.
15988 (rs6000_hard_regno_mode_ok): New function.
15989 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
15990 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
15991 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15992 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
15993 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
15994 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
15995 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15996 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
15997 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
15998 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15999 (sh_hard_regno_mode_ok): Make static.
16000 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
16001 instead of HARD_REGNO_MODE_OK.
16002 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
16003 (sparc_mode_class): Delete.
16004 (HARD_REGNO_MODE_OK): Delete.
16005 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16006 (hard_regno_mode_classes): Make static.
16007 (sparc_mode_class): Likewise.
16008 (sparc_hard_regno_mode_ok): New function.
16009 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
16010 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
16011 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
16013 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16014 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
16015 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
16016 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
16017 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
16018 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16019 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
16020 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
16021 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16022 (visium_hard_regno_mode_ok): New function.
16023 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
16024 instead of HARD_REGNO_MODE_OK.
16025 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
16026 (HARD_REGNO_MODE_OK): Delete.
16027 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
16028 (xtensa_hard_regno_mode_ok_p): ...this and make static.
16029 (xtensa_option_override): Update accordingly.
16030 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16031 (xtensa_hard_regno_mode_ok): New function.
16032 * system.h (HARD_REGNO_MODE_OK): Poison.
16034 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16035 Alan Hayward <alan.hayward@arm.com>
16036 David Sherwood <david.sherwood@arm.com>
16038 * target.def (hard_regno_call_part_clobbered): New hook.
16039 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
16040 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
16041 * doc/tm.texi: Regenerate.
16042 * hooks.h (hook_bool_uint_mode_false): Declare.
16043 * hooks.c (hook_bool_uint_mode_false): New function.
16044 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16045 * cselib.c (cselib_process_insn): Use
16046 targetm.hard_regno_call_part_clobbered instead of
16047 HARD_REGNO_CALL_PART_CLOBBERED.
16048 * ira-conflicts.c (ira_build_conflicts): Likewise.
16049 * ira-costs.c (ira_tune_allocno_costs): Likewise.
16050 * lra-constraints.c (need_for_call_save_p): Likewise.
16051 * lra-lives.c: Include target.h.
16052 (check_pseudos_live_through_calls): Use
16053 targetm.hard_regno_call_part_clobbered instead of
16054 HARD_REGNO_CALL_PART_CLOBBERED.
16055 * regcprop.c: Include target.h.
16056 (copyprop_hardreg_forward_1): Use
16057 targetm.hard_regno_call_part_clobbered instead of
16058 HARD_REGNO_CALL_PART_CLOBBERED.
16059 * reginfo.c (choose_hard_reg_mode): Likewise.
16060 * regrename.c (check_new_reg_p): Likewise.
16061 * reload.c (find_equiv_reg): Likewise.
16062 * reload1.c (emit_reload_insns): Likewise.
16063 * sched-deps.c (deps_analyze_insn): Likewise.
16064 * sel-sched.c (init_regs_for_mode): Likewise.
16065 (mark_unavailable_hard_regs): Likewise.
16066 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
16067 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16068 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
16070 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16071 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16072 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
16074 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
16076 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16077 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16078 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
16080 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16081 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16082 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
16084 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16085 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
16087 * config/powerpcspe/powerpcspe.c
16088 (rs6000_hard_regno_call_part_clobbered): New function.
16089 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16090 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16091 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
16093 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16094 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16095 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
16097 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16098 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16099 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
16101 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16102 Alan Hayward <alan.hayward@arm.com>
16103 David Sherwood <david.sherwood@arm.com>
16105 * rtl.h (subreg_memory_offset): Declare.
16106 * emit-rtl.c (subreg_memory_offset): New function.
16107 * expmed.c (store_bit_field_1): Use it.
16108 * expr.c (undefined_operand_subword_p): Likewise.
16109 * simplify-rtx.c (simplify_subreg): Likewise.
16111 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
16113 PR rtl-optimization/57448
16116 * optabs.c (expand_atomic_load): Place compiler memory barriers if
16117 using atomic_load pattern.
16118 (expand_atomic_store): Likewise.
16120 2017-09-04 Jakub Jelinek <jakub@redhat.com>
16123 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
16124 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
16125 handling. Use replace_call_with_value with NULL instead of
16126 gsi_replace, unlink_stmt_vdef and release_defs.
16128 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
16131 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
16133 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16136 * rtl.h (emit_library_call_value_1): Declare.
16137 (emit_library_call): Replace declaration with a series of overloads.
16138 Remove the parameter count argument.
16139 (emit_library_call_value): Likewise.
16140 * calls.c (emit_library_call_value_1): Make global. Replace varargs
16141 with an "rtx_mode_t *".
16142 (emit_library_call_value): Delete.
16143 (emit_library_call): Likewise.
16144 * asan.c (asan_emit_stack_protection): Update calls accordingly.
16145 (asan_emit_allocas_unpoison): Likewise.
16146 * builtins.c (expand_builtin_powi): Likewise.
16147 (expand_asan_emit_allocas_unpoison): Likewise.
16148 * cfgexpand.c (expand_main_function): Likewise.
16149 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
16150 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
16151 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
16152 * config/arm/arm.c (arm_trampoline_init): Likewise.
16153 (arm_call_tls_get_addr): Likewise.
16154 (arm_expand_divmod_libfunc): Likewise.
16155 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
16156 (smulsi3_highpart): Likewise.
16157 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
16158 (c6x_expand_compare): Likewise.
16159 (c6x_expand_movmem): Likewise.
16160 * config/frv/frv.c (frv_trampoline_init): Likewise.
16161 * config/i386/i386.c (ix86_trampoline_init): Likewise.
16162 (ix86_expand_divmod_libfunc): Likewise.
16163 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
16164 (ia64_expand_compare): Likewise.
16165 (ia64_profile_hook): Likewise.
16166 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
16167 (nonlocal_goto): Likewise.
16168 (restore_stack_nonlocal): Likewise.
16169 * config/m32r/m32r.c (block_move_call): Likewise.
16170 (m32r_trampoline_init): Likewise.
16171 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
16172 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
16173 (m68k_call_m68k_read_tp): Likewise.
16174 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
16175 (microblaze_expand_divide): Likewise.
16176 * config/mips/mips.h (mips_args): Likewise.
16177 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
16178 (MIPS_ICACHE_SYNC): Likewise.
16179 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
16180 (nios2_trampoline_init): Likewise.
16181 * config/pa/pa.c (hppa_tls_call): Likewise.
16182 (pa_trampoline_init): Likewise.
16183 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
16184 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
16185 (expand_strn_compare): Likewise.
16186 (rs6000_generate_compare): Likewise.
16187 (rs6000_expand_float128_convert): Likewise.
16188 (output_profile_hook): Likewise.
16189 (rs6000_trampoline_init): Likewise.
16190 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
16191 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
16192 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
16193 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
16194 (rs6000_generate_compare): Likewise.
16195 (rs6000_expand_float128_convert): Likewise.
16196 (output_profile_hook): Likewise.
16197 (rs6000_trampoline_init): Likewise.
16198 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
16199 * config/sh/sh.c (sh_trampoline_init): Likewise.
16200 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
16201 (sparc_emit_float_lib_cmp): Likewise.
16202 (sparc32_initialize_trampoline): Likewise.
16203 (sparc64_initialize_trampoline): Likewise.
16204 (sparc_profile_hook): Likewise.
16205 * config/spu/spu.c (ea_load_store): Likewise.
16206 * config/spu/spu.md (floatunssidf2): Likewise.
16207 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
16208 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
16209 * config/visium/visium.c (expand_block_move_4): Likewise.
16210 (expand_block_move_2): Likewise.
16211 (expand_block_move_1): Likewise.
16212 (expand_block_set_4): Likewise.
16213 (expand_block_set_2): Likewise.
16214 (expand_block_set_1): Likewise.
16215 (visium_trampoline_init): Likewise.
16216 (visium_profile_hook): Likewise.
16217 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
16218 (xtensa_setup_frame_addresses): Likewise.
16219 (xtensa_trampoline_init): Likewise.
16220 * except.c (sjlj_emit_function_enter): Likewise.
16221 (sjlj_emit_function_exit): Likewise.
16222 * explow.c (allocate_dynamic_stack_space): Likewise.
16223 (probe_stack_range): Likewise.
16224 * expr.c (convert_mode_scalar): Likewise.
16225 * optabs.c (expand_binop): Likewise.
16226 (expand_twoval_binop_libfunc): Likewise.
16227 (expand_unop): Likewise.
16228 (prepare_cmp_insn): Likewise.
16229 (prepare_float_lib_cmp): Likewise.
16230 (expand_float): Likewise.
16231 (expand_fix): Likewise.
16232 (expand_fixed_convert): Likewise.
16233 (maybe_emit_sync_lock_test_and_set): Likewise.
16234 (expand_atomic_compare_and_swap): Likewise.
16235 (expand_mem_thread_fence): Likewise.
16236 (expand_atomic_fetch_op): Likewise.
16238 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
16240 * doc/generic.texi (OpenACC): Adjust URL.
16241 * doc/invoke.texi (C Dialect Options): Ditto.
16243 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
16245 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
16246 predicate for operand 1. Add (m,<S>) constraint.
16247 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
16248 Prevent memory operand 1 with register operand 2.
16250 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
16252 PR rtl-optimization/82024
16253 * combine.c (try_combine): If the combination result is a PARALLEL,
16254 and we only need to retain the SET in there that would be placed
16255 at I2, check that we can place that at I3 instead, before doing so.
16257 2017-09-01 Jakub Jelinek <jakub@redhat.com>
16260 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
16262 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
16263 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
16266 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
16267 Jeff Law <law@redhat.com>
16269 * varasm.c (bss_initializer_p): Do not put constants into .bss
16270 (categorize_decl_for_section): Handle bss_initializer_p returning
16271 false when DECL_INITIAL is NULL.
16273 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16276 * config/s390/s390.c (s390_can_inline_p): New function.
16278 2017-09-01 Jeff Law <law@redhat.com>
16280 PR tree-optimization/82052
16281 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
16282 Always initialize the returned slot after a hash table miss
16283 when INSERT is true.
16285 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
16287 * config/s390/s390.md (mem_signal_fence): Remove.
16288 * doc/md.texi (mem_signal_fence): Remove.
16289 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
16291 * target-insns.def (mem_signal_fence): Remove.
16293 2017-09-01 Jakub Jelinek <jakub@redhat.com>
16296 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
16299 * asan.c (create_odr_indicator): Strip name encoding from assembler
16300 name before appending it after __odr_asan_.
16302 2017-09-01 Martin Liska <mliska@suse.cz>
16304 PR tree-optimization/82059
16305 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
16306 frequency only when an edge is redirected.
16308 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
16310 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
16311 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
16312 (arc_conditional_register_usage): Remove ARC600 lp_count
16314 (arc_file_start): Emit Tag_ARC_CPU_variation.
16315 (arc_can_use_doloop_p): New conditions to use ZOLs.
16316 (hwloop_fail): New function.
16317 (hwloop_optimize): Likewise.
16318 (hwloop_pattern_reg): Likewise.
16319 (arc_doloop_hooks): New struct, to be used with reorg_loops.
16320 (arc_reorg_loops): New function, calls reorg_loops.
16321 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
16322 (arc600_corereg_hazard): Remove ZOL checking, case handled by
16324 (arc_loop_hazard): Remove function, functionality moved into
16326 (arc_hazard): Remove arc_loop_hazard call.
16327 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
16328 into hwloop_optimize.
16329 (arc_label_align): Remove ZOL handling.
16330 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
16331 * config/arc/arc.md (doloop_begin): Remove pattern.
16332 (doloop_begin_i): Likewise.
16333 (doloop_end_i): Likewise.
16334 (doloop_fallback): Likewise.
16335 (doloop_fallback_m): Likewise.
16336 (doloop_end): Reimplement expand.
16337 (arc_lp): New pattern for LP instruction.
16338 (loop_end): New pattern.
16339 (loop_fail): Likewise.
16340 (decrement_and_branch_until_zero): Likewise.
16341 * config/arc/arc.opt (mlpc-width): New option.
16342 * doc/invoke.texi (mlpc-width): Document option.
16344 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
16346 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
16347 (arc_ccfsm_advance): Fix checking for delay slots.
16348 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
16350 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
16352 * config/arc/arc.md (movqi_insn): Add stores to save constant long
16354 (movhi_insn): Update store instruction constraint which are saving
16355 6-bit short immediates.
16356 (movsi_insn): Consider also short scaled load operations.
16357 (zero_extendhisi2_i): Use Usd constraint instead of T.
16358 (extendhisi2_i): Add q constraint.
16359 (arc_clzsi2): Add type and length attributes.
16360 (arc_ctzsi2): Likewise.
16361 * config/arc/constraints.md (Usc): Update constraint, the
16362 assembler can parse two relocations for a single instruction.
16364 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
16366 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
16367 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
16369 2017-08-31 Olivier Hainque <hainque@adacore.com>
16371 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
16372 match as powerpc-wrs-vxworks*.
16374 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
16376 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
16377 register constraint for by-element operand.
16378 (aarch64_mls_elt_merge<mode>): Likewise.
16380 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
16382 * config/arc/arc.c (arc_can_follow_jump): Check for short
16385 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
16387 * config.gcc: Use g.opt for arc.
16388 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
16389 functionality moved to ...
16390 (legitimate_scaled_address_p): New function, ...here.
16391 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
16392 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
16393 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
16395 (arc_override_options): Handle G option.
16396 (arc_output_pic_addr_const): Correct function definition.
16397 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
16398 (arc_decl_anon_ns_mem_p): Delete.
16399 (arc_in_small_data_p): Overhaul this function to take into
16400 consideration the value given via G option.
16401 (arc_rewrite_small_data_1): Renamed and corrected old
16402 arc_rewrite_small_data function.
16403 (arc_rewrite_small_data): New function.
16404 (small_data_pattern): Don't use pic_offset_table_rtx.
16405 * config/arc/arc.h (CC1_SPEC): Recognize G option.
16406 * config/arc/simdext.md (movmisalignv2hi): Use
16407 prepare_move_operands function.
16409 (movmisalign*): Likewise.
16410 * doc/invoke.texi (ARC options): Document -G option.
16412 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
16414 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
16416 * config/arc/arc.c (arc_print_operand): Output scalled address for
16417 sdata whenever is possible.
16418 (arc_in_small_data_p): Allow sdata for 64bit datum when double
16419 load/stores are available.
16420 (compact_sda_memory_operand): Check for the alignment required by
16421 code density instructions.
16422 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
16424 * config/arc/constraints.md (Usd): Update constraint.
16425 (Us0): New constraint.
16426 (Usc): Update constraint.
16428 2017-08-31 Richard Biener <rguenther@suse.de>
16430 PR middle-end/82054
16431 * dwarf2out.c (dwarf2out_early_global_decl): Process each
16432 function only once.
16434 2017-08-31 Tamar Christina <tamar.christina@arm.com>
16436 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
16437 Resize type_signature.
16439 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
16440 Alan Hayward <alan.hayward@arm.com>
16441 David Sherwood <david.sherwood@arm.com>
16443 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
16444 subregs whose inner modes can be stored in GPRs.
16445 (aarch64_classify_index): Likewise.
16447 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
16448 Alan Hayward <alan.hayward@arm.com>
16449 David Sherwood <david.sherwood@arm.com>
16451 * config/aarch64/iterators.md (V_cmp_result): Rename to...
16452 (V_INT_EQUIV): ...this.
16453 (v_cmp_result): Rename to...
16454 (v_int_equiv): ...this.
16455 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
16456 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
16457 (copysign<mode>3): Likewise.
16458 (aarch64_simd_bsl<mode>_internal): Likewise.
16459 (aarch64_simd_bsl<mode>): Likewise.
16460 (vec_cmp<mode><mode>): Likewise.
16461 (vcond<mode><mode>): Likewise.
16462 (vcond<v_cmp_mixed><mode>): Likewise.
16463 (vcondu<mode><v_cmp_mixed>): Likewise.
16464 (aarch64_cm<optab><mode>): Likewise.
16465 (aarch64_cmtst<mode>): Likewise.
16466 (aarch64_fac<optab><mode>): Likewise.
16467 (vec_perm_const<mode>): Likewise.
16468 (vcond_mask_<mode><v_cmp_result>): Rename to...
16469 (vcond_mask_<mode><v_int_equiv>): ...this.
16470 (vec_cmp<mode><v_cmp_result>): Rename to...
16471 (vec_cmp<mode><v_int_equiv>): ...this.
16473 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
16474 Alan Hayward <alan.hayward@arm.com>
16475 David Sherwood <david.sherwood@arm.com>
16477 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
16479 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
16480 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
16481 (UNSPEC_LD4_DREG): New unspecs.
16482 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
16483 (aarch64_ld2<mode>_dreg_be): Replace with...
16484 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
16486 (aarch64_ld3<mode>_dreg_le)
16487 (aarch64_ld3<mode>_dreg_be): Replace with...
16488 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
16490 (aarch64_ld4<mode>_dreg_le)
16491 (aarch64_ld4<mode>_dreg_be): Replace with...
16492 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
16495 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16497 PR tree-optimization/81987
16498 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
16499 insert an initializer in a location not dominated by the stride
16502 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
16504 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
16505 on the entire header of the finally block in the fallthru case.
16507 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
16509 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
16511 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
16513 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
16514 rs6000_emit_move_from_cr and call renamed function.
16515 (rs6000_emit_prologue): Call renamed functions.
16516 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
16517 movesi_from_cr, remove volatile CRs.
16519 2017-08-30 Jon Beniston <jon@beniston.com>
16520 Richard Biener <rguenther@suse.de>
16522 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
16523 of VECTOR_MODE_P check.
16524 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
16525 element vector types.
16527 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16529 * df.h (df_read_modify_subreg_p): Remove in favor of...
16530 * rtl.h (read_modify_subreg_p): ...this new function. Take a
16531 const_rtx instead of an rtx.
16532 * cprop.c (local_cprop_find_used_regs): Update accordingly.
16533 * df-problems.c (df_word_lr_mark_ref): Likewise.
16534 * ira-lives.c (mark_pseudo_reg_live): Likewise.
16535 (mark_pseudo_reg_dead): Likewise.
16536 (mark_ref_dead): Likewise.
16537 * reginfo.c (init_subregs_of_mode): Likewise.
16538 * sched-deps.c (sched_analyze_1): Likewise.
16539 * df-scan.c (df_def_record_1): Likewise.
16540 (df_uses_record): Likewise.
16541 (df_read_modify_subreg_p): Remove in favor of...
16542 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
16543 const_rtx instead of an rtx.
16545 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16546 Alan Hayward <alan.hayward@arm.com>
16547 David Sherwood <david.sherwood@arm.com>
16549 * rtl.h (partial_subreg_p): New function.
16550 * caller-save.c (save_call_clobbered_regs): Use it.
16551 * calls.c (expand_call): Likewise.
16552 * combine.c (combinable_i3pat): Likewise.
16553 (simplify_set): Likewise.
16554 (make_extraction): Likewise.
16555 (make_compound_operation_int): Likewise.
16556 (gen_lowpart_or_truncate): Likewise.
16557 (force_to_mode): Likewise.
16558 (make_field_assignment): Likewise.
16559 (reg_truncated_to_mode): Likewise.
16560 (record_truncated_value): Likewise.
16561 (move_deaths): Likewise.
16562 * cse.c (record_jump_cond): Likewise.
16563 (cse_insn): Likewise.
16564 * cselib.c (cselib_lookup_1): Likewise.
16565 * expmed.c (extract_bit_field_using_extv): Likewise.
16566 * function.c (assign_parm_setup_reg): Likewise.
16567 * ifcvt.c (noce_convert_multiple_sets): Likewise.
16568 * ira-build.c (create_insn_allocnos): Likewise.
16569 * lra-coalesce.c (merge_pseudos): Likewise.
16570 * lra-constraints.c (match_reload): Likewise.
16571 (simplify_operand_subreg): Likewise.
16572 (curr_insn_transform): Likewise.
16573 * lra-lives.c (process_bb_lives): Likewise.
16574 * lra.c (new_insn_reg): Likewise.
16575 (lra_substitute_pseudo): Likewise.
16576 * regcprop.c (mode_change_ok): Likewise.
16577 (maybe_mode_change): Likewise.
16578 (copyprop_hardreg_forward_1): Likewise.
16579 * reload.c (push_reload): Likewise.
16580 (find_reloads): Likewise.
16581 (find_reloads_subreg_address): Likewise.
16582 * reload1.c (alter_reg): Likewise.
16583 (eliminate_regs_1): Likewise.
16584 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
16586 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
16588 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
16589 back to if statements, including unpack.
16591 2017-08-30 Martin Liska <mliska@suse.cz>
16593 PR inline-asm/82001
16594 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
16596 (func_checker::compare_asm_inputs_outputs): ... this function.
16597 (func_checker::compare_gimple_asm): Use the function to compare
16598 also ASM constrains.
16599 * ipa-icf-gimple.h: Rename the function.
16601 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16602 Alan Hayward <alan.hayward@arm.com>
16603 David Sherwood <david.sherwood@arm.com>
16605 * coretypes.h (complex_mode): New type.
16606 * gdbhooks.py (build_pretty_printer): Handle it.
16607 * machmode.h (complex_mode): New class.
16608 (complex_mode::includes_p): New function.
16609 (is_complex_int_mode): Likewise.
16610 (is_complex_float_mode): Likewise.
16611 * genmodes.c (get_mode_class): Handle complex mode classes.
16612 * function.c (expand_function_end): Use is_complex_int_mode.
16614 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16615 Alan Hayward <alan.hayward@arm.com>
16616 David Sherwood <david.sherwood@arm.com>
16618 * coretypes.h (scalar_mode_pod): New typedef.
16619 * gdbhooks.py (build_pretty_printer): Handle it.
16620 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
16621 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
16622 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
16623 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
16624 as_a <scalar_mode>.
16626 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16627 Alan Hayward <alan.hayward@arm.com>
16628 David Sherwood <david.sherwood@arm.com>
16630 * machmode.h (mode_for_vector): Take a scalar_mode instead
16632 * stor-layout.c (mode_for_vector): Likewise.
16633 * explow.c (promote_mode): Use as_a <scalar_mode>.
16634 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
16636 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16637 Alan Hayward <alan.hayward@arm.com>
16638 David Sherwood <david.sherwood@arm.com>
16640 * target.def (preferred_simd_mode): Take a scalar_mode
16641 instead of a machine_mode.
16642 * targhooks.h (default_preferred_simd_mode): Likewise.
16643 * targhooks.c (default_preferred_simd_mode): Likewise.
16644 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
16645 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
16646 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
16647 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
16648 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
16649 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
16650 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
16651 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
16653 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
16654 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
16655 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
16656 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
16657 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
16658 * doc/tm.texi: Regenerate.
16659 * optabs-query.c (can_vec_mask_load_store_p): Return false for
16662 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16663 Alan Hayward <alan.hayward@arm.com>
16664 David Sherwood <david.sherwood@arm.com>
16666 * target.def (scalar_mode_supported_p): Take a scalar_mode
16667 instead of a machine_mode.
16668 * targhooks.h (default_scalar_mode_supported_p): Likewise.
16669 * targhooks.c (default_scalar_mode_supported_p): Likewise.
16670 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
16671 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
16672 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
16673 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
16674 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
16675 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
16676 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
16677 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
16678 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
16679 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
16680 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
16681 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
16683 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
16684 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
16685 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
16686 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
16687 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
16689 * doc/tm.texi: Regenerate.
16691 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16692 Alan Hayward <alan.hayward@arm.com>
16693 David Sherwood <david.sherwood@arm.com>
16695 * coretypes.h (opt_scalar_mode): New typedef.
16696 * gdbhooks.py (build_pretty_printers): Handle it.
16697 * machmode.h (mode_iterator::get_2xwider): Add overload for
16699 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
16701 * expr.c (convert_mode_scalar): Likewise.
16702 * omp-low.c (omp_clause_aligned_alignment): Likewise.
16703 * optabs.c (expand_float): Likewise.
16704 (expand_fix): Likewise.
16705 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
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 * optabs.c (expand_float): Explicitly check for scalars before
16712 using a branching expansion.
16713 (expand_fix): Likewise.
16715 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16716 Alan Hayward <alan.hayward@arm.com>
16717 David Sherwood <david.sherwood@arm.com>
16719 * expr.c (convert_mode): Split scalar handling out into...
16720 (convert_mode_scalar): ...this new function. Treat the modes
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 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
16729 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): 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 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
16736 rather than a machine_mode.
16737 (fixed_from_string): Likewise.
16738 (fixed_convert): Likewise.
16739 (fixed_convert_from_int): Likewise.
16740 (fixed_convert_from_real): Likewise.
16741 (real_convert_from_fixed): Likewise.
16742 * fixed-value.c (fixed_from_double_int): Likewise.
16743 (fixed_from_string): Likewise.
16744 (fixed_convert): Likewise.
16745 (fixed_convert_from_int): Likewise.
16746 (fixed_convert_from_real): Likewise.
16747 (real_convert_from_fixed): Likewise.
16748 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
16750 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16751 Alan Hayward <alan.hayward@arm.com>
16752 David Sherwood <david.sherwood@arm.com>
16754 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
16755 of separate mode class checks. Do not allow vector modes here.
16756 (immed_wide_int_const): Use as_a <scalar_mode>.
16757 * explow.c (trunc_int_for_mode): Likewise.
16758 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
16759 (wi::shwi): Likewise.
16760 (wi::min_value): Likewise.
16761 (wi::max_value): Likewise.
16762 * dwarf2out.c (loc_descriptor): Likewise.
16763 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
16764 for CONST_WIDE_INT.
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 * tree.h (SCALAR_TYPE_MODE): New macro.
16771 * expr.c (expand_expr_addr_expr_1): Use it.
16772 (expand_expr_real_2): Likewise.
16773 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
16774 (fold_convert_const_fixed_from_int): Likewise.
16775 (fold_convert_const_fixed_from_real): Likewise.
16776 (native_encode_fixed): Likewise
16777 (native_encode_complex): Likewise
16778 (native_encode_vector): Likewise.
16779 (native_interpret_fixed): Likewise.
16780 (native_interpret_real): Likewise.
16781 (native_interpret_complex): Likewise.
16782 (native_interpret_vector): Likewise.
16783 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
16784 (simd_clone_adjust_argument_types): Likewise.
16785 (simd_clone_init_simd_arrays): Likewise.
16786 (simd_clone_adjust): Likewise.
16787 * stor-layout.c (layout_type): Likewise.
16788 * tree.c (build_minus_one_cst): Likewise.
16789 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
16790 * tree-inline.c (estimate_move_cost): Likewise.
16791 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
16792 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
16793 (vectorizable_reduction): Likewise.
16794 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
16795 (vect_recog_mixed_size_cond_pattern): Likewise.
16796 (check_bool_pattern): Likewise.
16797 (adjust_bool_pattern): Likewise.
16798 (search_type_for_mask_1): Likewise.
16799 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
16800 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
16801 (vectorizable_load): Likewise.
16802 (vectorizable_store): Likewise.
16803 * ubsan.c (ubsan_encode_value): Likewise.
16804 * varasm.c (output_constant): Likewise.
16806 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16807 Alan Hayward <alan.hayward@arm.com>
16808 David Sherwood <david.sherwood@arm.com>
16810 * coretypes.h (scalar_mode): New class.
16811 * machmode.h (scalar_mode): Likewise.
16812 (scalar_mode::includes_p): New function.
16813 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
16814 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
16815 * genmodes.c (get_mode_class): Handle remaining scalar modes.
16816 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
16817 * expmed.c (store_bit_field_1): Likewise.
16818 (extract_bit_field_1): Likewise.
16819 * expr.c (write_complex_part): Likewise.
16820 (read_complex_part): Likewise.
16821 (emit_move_complex_push): Likewise.
16822 (expand_expr_real_2): Likewise.
16823 * function.c (assign_parm_setup_reg): Likewise.
16824 (assign_parms_unsplit_complex): Likewise.
16825 * optabs.c (expand_binop): Likewise.
16826 * rtlanal.c (subreg_get_info): Likewise.
16827 * simplify-rtx.c (simplify_immed_subreg): Likewise.
16828 * varasm.c (output_constant_pool_2): Likewise.
16830 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16831 Alan Hayward <alan.hayward@arm.com>
16832 David Sherwood <david.sherwood@arm.com>
16834 * expmed.c (extract_high_half): Use scalar_int_mode and remove
16836 (expmed_mult_highpart_optab): Likewise.
16837 (expmed_mult_highpart): Likewise.
16839 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16840 Alan Hayward <alan.hayward@arm.com>
16841 David Sherwood <david.sherwood@arm.com>
16843 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
16844 instead of a machine_mode.
16845 (builtin_memset_read_str): Likewise.
16846 * builtins.c (c_readstr): Likewise.
16847 (builtin_memcpy_read_str): Likewise.
16848 (builtin_strncpy_read_str): Likewise.
16849 (builtin_memset_read_str): Likewise.
16850 (builtin_memset_gen_str): Likewise.
16851 (expand_builtin_signbit): Use scalar_int_mode for local variables.
16852 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
16853 instead of a machine_mode.
16854 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
16856 (make_extraction): Likewise.
16857 (try_widen_shift_mode): Take and return scalar_int_modes instead
16859 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
16860 a scalar_int_mode instead of a machine_mode.
16861 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
16862 (avr_addr_space_pointer_mode): Likewise.
16863 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
16864 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
16865 (msp430_unwind_word_mode): Likewise.
16866 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
16867 (spu_addr_space_pointer_mode): Likewise.
16868 (spu_addr_space_address_mode): Likewise.
16869 (spu_libgcc_cmp_return_mode): Likewise.
16870 (spu_libgcc_shift_count_mode): Likewise.
16871 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
16872 (rl78_addr_space_pointer_mode): Likewise.
16873 (fl78_unwind_word_mode): Likewise.
16874 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
16876 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
16877 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
16878 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
16879 (mips_valid_pointer_mode): Likewise.
16880 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
16881 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
16882 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
16884 (ft32_addr_space_address_mode): Likewise.
16885 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
16886 scalar_int_mode instead of a machine_mode.
16887 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
16889 (m32c_addr_space_address_mode): Likewise.
16890 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
16891 (rs6000_eh_return_filter_mode): Likewise.
16892 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
16893 (rs6000_eh_return_filter_mode): Likewise.
16894 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
16895 (s390_libgcc_shift_count_mode): Likewise.
16896 (s390_unwind_word_mode): Likewise.
16897 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
16899 * target.def (mode_rep_extended): Likewise.
16900 (valid_pointer_mode): Likewise.
16901 (addr_space.valid_pointer_mode): Likewise.
16902 (eh_return_filter_mode): Return a scalar_int_mode rather than
16904 (libgcc_cmp_return_mode): Likewise.
16905 (libgcc_shift_count_mode): Likewise.
16906 (unwind_word_mode): Likewise.
16907 (addr_space.pointer_mode): Likewise.
16908 (addr_space.address_mode): Likewise.
16909 * doc/tm.texi: Regenerate.
16910 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
16912 (do_jump): Use scalar_int_mode for local variables.
16913 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
16914 rather than a machine_mode.
16915 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
16916 (scompare_loc_descriptor_wide): Likewise.
16917 (scompare_loc_descriptor_narrow): Likewise.
16918 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
16920 * except.c (sjlj_emit_dispatch_table): Likewise.
16921 (expand_builtin_eh_copy_values): Likewise.
16922 * explow.c (convert_memory_address_addr_space_1): Likewise.
16923 Take a scalar_int_mode rather than a machine_mode.
16924 (convert_memory_address_addr_space): Take a scalar_int_mode rather
16925 than a machine_mode.
16926 (memory_address_addr_space): Use scalar_int_mode for local variables.
16927 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
16928 rather than a machine_mode.
16929 * expmed.c (mask_rtx): Likewise.
16930 (init_expmed_one_conv): Likewise.
16931 (expand_mult_highpart_adjust): Likewise.
16932 (extract_high_half): Likewise.
16933 (expmed_mult_highpart_optab): Likewise.
16934 (expmed_mult_highpart): Likewise.
16935 (expand_smod_pow2): Likewise.
16936 (expand_sdiv_pow2): Likewise.
16937 (emit_store_flag_int): Likewise.
16938 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
16940 (extract_low_bits): Likewise.
16941 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
16943 * expr.c (pieces_addr::adjust): Likewise.
16944 (can_store_by_pieces): Likewise.
16945 (store_by_pieces): Likewise.
16946 (clear_by_pieces_1): Likewise.
16947 (expand_expr_addr_expr_1): Likewise.
16948 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
16949 (expand_expr_real_1): Likewise.
16950 (try_casesi): Likewise.
16951 * final.c (shorten_branches): Likewise.
16952 * fold-const.c (fold_convert_const_int_from_fixed): Change the
16953 type of "mode" to machine_mode.
16954 * internal-fn.c (expand_arith_overflow_result_store): Take a
16955 scalar_int_mode rather than a machine_mode.
16956 (expand_mul_overflow): Use scalar_int_mode for local variables.
16957 * loop-doloop.c (doloop_modify): Likewise.
16958 (doloop_optimize): Likewise.
16959 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
16960 than a machine_mode.
16961 (expand_doubleword_shift_condmove): Likewise.
16962 (expand_doubleword_shift): Likewise.
16963 (expand_doubleword_clz): Likewise.
16964 (expand_doubleword_popcount): Likewise.
16965 (expand_doubleword_parity): Likewise.
16966 (expand_absneg_bit): Use scalar_int_mode for local variables.
16967 (prepare_float_lib_cmp): Likewise.
16968 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
16969 rather than a machine_mode.
16970 (convert_memory_address_addr_space): Likewise.
16971 (get_mode_bounds): Likewise.
16972 (get_address_mode): Return a scalar_int_mode rather than a
16974 * rtlanal.c (get_address_mode): Likewise.
16975 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
16976 than a machine_mode.
16977 * targhooks.c (default_mode_rep_extended): Likewise.
16978 (default_valid_pointer_mode): Likewise.
16979 (default_addr_space_valid_pointer_mode): Likewise.
16980 (default_eh_return_filter_mode): Return a scalar_int_mode rather
16981 than a machine_mode.
16982 (default_libgcc_cmp_return_mode): Likewise.
16983 (default_libgcc_shift_count_mode): Likewise.
16984 (default_unwind_word_mode): Likewise.
16985 (default_addr_space_pointer_mode): Likewise.
16986 (default_addr_space_address_mode): Likewise.
16987 * targhooks.h (default_eh_return_filter_mode): Likewise.
16988 (default_libgcc_cmp_return_mode): Likewise.
16989 (default_libgcc_shift_count_mode): Likewise.
16990 (default_unwind_word_mode): Likewise.
16991 (default_addr_space_pointer_mode): Likewise.
16992 (default_addr_space_address_mode): Likewise.
16993 (default_mode_rep_extended): Take a scalar_int_mode rather than
16995 (default_valid_pointer_mode): Likewise.
16996 (default_addr_space_valid_pointer_mode): Likewise.
16997 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
16999 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
17000 rather than a machine_mode.
17001 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
17002 for local variables.
17003 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
17004 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
17005 than a machine_mode.
17007 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17008 Alan Hayward <alan.hayward@arm.com>
17009 David Sherwood <david.sherwood@arm.com>
17011 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
17012 the mode argument to scalar_int_mode.
17013 (do_jump_by_parts_zero_rtx): Likewise.
17014 (do_jump_by_parts_equality_rtx): Likewise.
17015 (do_jump_by_parts_greater): Take a mode argument.
17016 (do_jump_by_parts_equality): Likewise.
17017 (do_jump_1): Update calls accordingly.
17019 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17020 Alan Hayward <alan.hayward@arm.com>
17021 David Sherwood <david.sherwood@arm.com>
17023 * is-a.h (safe_dyn_cast): New function.
17024 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
17025 (jump_table_for_label): Likewise.
17026 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
17027 instead of an rtx_insn *.
17028 (shorten_branches): Use dyn_cast instead of LABEL_P and
17029 JUMP_TABLE_DATA_P. Use jump_table_for_label and
17030 rtx_jump_table_data::get_data_mode.
17031 (final_scan_insn): Likewise.
17033 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17034 Alan Hayward <alan.hayward@arm.com>
17035 David Sherwood <david.sherwood@arm.com>
17037 * combine.c (try_combine): Use is_a <scalar_int_mode> when
17038 trying to combine a full-register integer set with a subreg
17041 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17042 Alan Hayward <alan.hayward@arm.com>
17043 David Sherwood <david.sherwood@arm.com>
17045 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
17046 that is always either address_mode or pointer_mode.
17048 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17049 Alan Hayward <alan.hayward@arm.com>
17050 David Sherwood <david.sherwood@arm.com>
17052 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
17053 when the two are known to be equal.
17055 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17056 Alan Hayward <alan.hayward@arm.com>
17057 David Sherwood <david.sherwood@arm.com>
17059 * simplify-rtx.c (simplify_const_unary_operation): Use
17060 is_a <scalar_int_mode> instead of checking for a nonzero
17061 precision. Forcibly convert op_mode to a scalar_int_mode
17062 in that case. More clearly differentiate the operand and
17063 result modes and use the former when deciding what the value
17064 of a count-bits operation should be. Use is_int_mode instead
17065 of checking for a MODE_INT. Remove redundant check for whether
17066 this mode has a zero precision.
17068 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17069 Alan Hayward <alan.hayward@arm.com>
17070 David Sherwood <david.sherwood@arm.com>
17072 * optabs.c (widen_leading): Change the type of the mode argument
17073 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
17074 (widen_bswap): Likewise.
17075 (expand_parity): Likewise.
17076 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
17077 (expand_ffs): Likewise.
17078 (epand_unop): Check for scalar integer modes before calling the
17081 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17082 Alan Hayward <alan.hayward@arm.com>
17083 David Sherwood <david.sherwood@arm.com>
17085 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
17087 (expand_expr_real_1): Update call accordingly.
17089 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17090 Alan Hayward <alan.hayward@arm.com>
17091 David Sherwood <david.sherwood@arm.com>
17093 * expmed.c (store_bit_field_using_insv): Add op0_mode and
17094 value_mode arguments. Use scalar_int_mode internally.
17095 (store_bit_field_1): Rename the new integer mode from imode
17096 to op0_mode and use it instead of GET_MODE (op0). Update calls
17097 to store_split_bit_field, store_bit_field_using_insv and
17098 store_fixed_bit_field.
17099 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
17100 Use scalar_int_mode internally. Use a bit count rather than a mode
17101 when calculating the largest bit size for get_best_mode.
17102 Update calls to store_split_bit_field and store_fixed_bit_field_1.
17103 (store_fixed_bit_field_1): Add mode and value_mode arguments.
17104 Remove assertion that OP0 has a scalar integer mode.
17105 (store_split_bit_field): Add op0_mode and value_mode arguments.
17106 Update calls to extract_fixed_bit_field.
17107 (extract_bit_field_using_extv): Add an op0_mode argument.
17108 Use scalar_int_mode internally.
17109 (extract_bit_field_1): Rename the new integer mode from imode to
17110 op0_mode and use it instead of GET_MODE (op0). Update calls to
17111 extract_split_bit_field, extract_bit_field_using_extv and
17112 extract_fixed_bit_field.
17113 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
17114 to extract_split_bit_field and extract_fixed_bit_field_1.
17115 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
17116 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
17117 on the target mode.
17118 (extract_split_bit_field): Add an op0_mode argument. Update call
17119 to extract_fixed_bit_field.
17121 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17122 Alan Hayward <alan.hayward@arm.com>
17123 David Sherwood <david.sherwood@arm.com>
17125 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
17126 * explow.c (hard_function_value): Likewise.
17127 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
17128 convert_to_mode call outside the loop.
17129 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
17130 for the mode iterator. Require the mode specified by max_pieces
17132 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
17134 (copy_blkmode_to_reg): Likewise.
17135 (set_storage_via_setmem): Likewise.
17136 * optabs.c (prepare_cmp_insn): Likewise.
17137 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
17138 * stor-layout.c (finish_bitfield_representative): Likewise.
17140 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17141 Alan Hayward <alan.hayward@arm.com>
17142 David Sherwood <david.sherwood@arm.com>
17144 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
17145 * expr.c (convert_move): Use them.
17146 (convert_modes): Likewise.
17147 (store_expr_with_bounds): Likewise.
17149 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17150 Alan Hayward <alan.hayward@arm.com>
17151 David Sherwood <david.sherwood@arm.com>
17153 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
17154 parameter for the mode of "x". Remove the "known_x", "known_mode"
17155 and "known_ret" arguments. Change the type of the mode argument
17156 to scalar_int_mode.
17157 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
17158 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
17159 (reg_num_sign_bit_copies_for_combine): Likewise.
17160 * rtlanal.c (nonzero_bits1): Likewise.
17161 (num_sign_bit_copies1): Likewise.
17162 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
17163 (reg_num_sign_bit_copies_general): Likewise.
17164 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
17165 (reg_nonzero_bits_general): Likewise.
17167 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17168 Alan Hayward <alan.hayward@arm.com>
17169 David Sherwood <david.sherwood@arm.com>
17171 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
17172 than in subroutines. Return 1 for non-integer modes.
17173 (cached_num_sign_bit_copies): Change the type of the mode parameter
17174 to scalar_int_mode.
17175 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
17176 classes. Handle CONST_INT_P first and then check whether X also
17177 has a scalar integer mode. Check the same thing for inner registers
17178 of a SUBREG and for values that are being extended or truncated.
17180 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17181 Alan Hayward <alan.hayward@arm.com>
17182 David Sherwood <david.sherwood@arm.com>
17184 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
17185 in subroutines. Return the mode mask for non-integer modes.
17186 (cached_nonzero_bits): Change the type of the mode parameter
17187 to scalar_int_mode.
17188 (nonzero_bits1): Likewise. Remove early exit for other mode
17189 classes. Handle CONST_INT_P first and then check whether X
17190 also has a scalar integer mode.
17192 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17193 Alan Hayward <alan.hayward@arm.com>
17194 David Sherwood <david.sherwood@arm.com>
17196 * expr.c (widest_int_mode_for_size): Make the comment match the code.
17197 Return a scalar_int_mode and assert that the size is greater than
17199 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
17200 (op_by_pieces_d::op_by_pieces_d): Likewise.
17201 (op_by_pieces_d::run): Likewise.
17202 (can_store_by_pieces): Likewise.
17204 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17205 Alan Hayward <alan.hayward@arm.com>
17206 David Sherwood <david.sherwood@arm.com>
17208 * combine.c (extract_left_shift): Add a mode argument and update
17210 (make_compound_operation_int): Change the type of the mode parameter
17211 to scalar_int_mode and update the call to extract_left_shift.
17213 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17214 Alan Hayward <alan.hayward@arm.com>
17215 David Sherwood <david.sherwood@arm.com>
17217 * combine.c (simplify_and_const_int): Change the type of the mode
17218 parameter to scalar_int_mode.
17219 (simplify_and_const_int_1): Likewise. Update recursive call.
17221 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17222 Alan Hayward <alan.hayward@arm.com>
17223 David Sherwood <david.sherwood@arm.com>
17225 * combine.c (simplify_compare_const): Check that the mode is a
17226 scalar_int_mode (rather than VOIDmode) before testing its
17228 (simplify_comparison): Move COMPARISON_P handling out of the
17229 loop and restrict the latter part of the loop to scalar_int_modes.
17230 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
17231 and when considering SUBREG_REGs. Use is_int_mode instead of
17232 checking GET_MODE_CLASS against MODE_INT.
17234 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17235 Alan Hayward <alan.hayward@arm.com>
17236 David Sherwood <david.sherwood@arm.com>
17238 * combine.c (try_widen_shift_mode): Move check for equal modes to...
17239 (simplify_shift_const_1): ...here. Use scalar_int_mode for
17240 shift_unit_mode and for modes involved in scalar shifts.
17242 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17243 Alan Hayward <alan.hayward@arm.com>
17244 David Sherwood <david.sherwood@arm.com>
17246 * combine.c (force_int_to_mode): New function, split out from...
17247 (force_to_mode): ...here. Keep xmode up-to-date and use it
17248 instead of GET_MODE (x).
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 * optabs-query.h (extraction_insn::struct_mode): Change type to
17255 opt_scalar_int_mode and update comment.
17256 (extraction_insn::field_mode): Change type to scalar_int_mode.
17257 (extraction_insn::pos_mode): Likewise.
17258 * combine.c (make_extraction): Update accordingly.
17259 * optabs-query.c (get_traditional_extraction_insn): Likewise.
17260 (get_optab_extraction_insn): Likewise.
17261 * recog.c (simplify_while_replacing): Likewise.
17262 * expmed.c (narrow_bit_field_mem): Change the type of the mode
17263 parameter to opt_scalar_int_mode.
17265 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17266 Alan Hayward <alan.hayward@arm.com>
17267 David Sherwood <david.sherwood@arm.com>
17269 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
17270 to a scalar_int_mode instead of a machine_mode.
17271 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
17272 (get_best_mode): Return a boolean and use a pointer argument to store
17273 the selected mode. Replace the limit mode parameter with a bit limit.
17274 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
17275 for the values returned by bit_field_mode_iterator::next_mode.
17276 (store_bit_field): Update call to get_best_mode.
17277 (store_fixed_bit_field): Likewise.
17278 (extract_fixed_bit_field): Likewise.
17279 * expr.c (optimize_bitfield_assignment_op): Likewise.
17280 * fold-const.c (optimize_bit_field_compare): Likewise.
17281 (fold_truth_andor_1): Likewise.
17282 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
17283 Update for new type of m_mode.
17284 (get_best_mode): As above.
17286 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17287 Alan Hayward <alan.hayward@arm.com>
17288 David Sherwood <david.sherwood@arm.com>
17290 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
17291 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
17292 (store_bit_field): Check is_a <scalar_int_mode> before calling
17293 strict_volatile_bitfield_p.
17294 (extract_bit_field): Likewise.
17296 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17297 Alan Hayward <alan.hayward@arm.com>
17298 David Sherwood <david.sherwood@arm.com>
17300 * target.def (cstore_mode): Return a scalar_int_mode.
17301 * doc/tm.texi: Regenerate.
17302 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
17303 * targhooks.h (default_cstore_mode): Likewise.
17304 * targhooks.c (default_cstore_mode): Likewise, using a forced
17306 * expmed.c (emit_cstore): Expect the target of the cstore to be
17309 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17310 Alan Hayward <alan.hayward@arm.com>
17311 David Sherwood <david.sherwood@arm.com>
17313 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
17315 (niter_desc): Likewise mode.
17316 (iv_analyze): Add a mode parameter.
17318 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
17319 and change its type to scalar_int_mode.
17320 * loop-iv.c: Update commentary at head of file.
17321 (iv_constant): Pass the mode paraeter before the rtx it describes
17322 and change its type to scalar_int_mode. Remove VOIDmode handling.
17323 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
17324 (iv_extend): Likewise.
17325 (shorten_into_mode): Likewise.
17326 (iv_add): Use scalar_int_mode.
17327 (iv_mult): Likewise.
17328 (iv_shift): Likewise.
17329 (canonicalize_iv_subregs): Likewise.
17330 (get_biv_step_1): Pass the outer_mode parameter before the rtx
17331 it describes and change its mode to scalar_int_mode. Also change
17332 the type of the returned inner_mode to scalar_int_mode.
17333 (get_biv_step): Likewise, turning outer_mode from a pointer
17334 into a direct parameter. Update call to get_biv_step_1.
17335 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
17336 iv_constant and get_biv_step.
17337 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
17338 and change its type to scalar_int_mode. Don't initialise iv->mode
17339 to VOIDmode and remove later checks for its still being VOIDmode.
17340 Update calls to iv_analyze_op and iv_analyze_expr. Check
17341 is_a <scalar_int_mode> when changing the mode under consideration.
17342 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
17343 Update call to iv_analyze_expr.
17344 (iv_analyze_op): Add a mode parameter. Reject subregs whose
17345 inner register is not also a scalar_int_mode. Update call to
17347 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
17348 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
17349 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
17350 separate mode class checks. Update calls to iv_analyze. Remove
17351 fix-up of VOIDmodes after iv_analyze_biv.
17352 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
17353 don't have a scalar_int_mode. Update call to biv_p.
17355 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17356 Alan Hayward <alan.hayward@arm.com>
17357 David Sherwood <david.sherwood@arm.com>
17359 * cfgexpand.c (convert_debug_memory_address): Use
17360 as_a <scalar_int_mode>.
17361 * combine.c (expand_compound_operation): Likewise.
17362 (make_extraction): Likewise.
17363 (change_zero_ext): Likewise.
17364 (simplify_comparison): Likewise.
17365 * cse.c (cse_insn): Likewise.
17366 * dwarf2out.c (minmax_loc_descriptor): Likewise.
17367 (mem_loc_descriptor): Likewise.
17368 (loc_descriptor): Likewise.
17369 * expmed.c (init_expmed_one_mode): Likewise.
17370 (synth_mult): Likewise.
17371 (emit_store_flag_1): Likewise.
17372 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
17373 of a comparison with size.
17374 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
17375 (reduce_to_bit_field_precision): Likewise.
17376 * function.c (expand_function_end): Likewise.
17377 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
17378 * loop-doloop.c (doloop_modify): Likewise.
17379 * optabs.c (expand_binop): Likewise.
17380 (expand_unop): Likewise.
17381 (expand_copysign_absneg): Likewise.
17382 (prepare_cmp_insn): Likewise.
17383 (maybe_legitimize_operand): Likewise.
17384 * recog.c (const_scalar_int_operand): Likewise.
17385 * rtlanal.c (get_address_mode): Likewise.
17386 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17387 (simplify_cond_clz_ctz): Likewise.
17388 * tree-nested.c (get_nl_goto_field): Likewise.
17389 * tree.c (build_vector_type_for_mode): Likewise.
17390 * var-tracking.c (use_narrower_mode): Likewise.
17392 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17393 Alan Hayward <alan.hayward@arm.com>
17394 David Sherwood <david.sherwood@arm.com>
17396 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
17397 * builtins.c (expand_builtin_signbit): Use it.
17398 * cfgexpand.c (expand_debug_expr): Likewise.
17399 * dojump.c (do_jump): Likewise.
17400 (do_compare_and_jump): Likewise.
17401 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
17402 * expmed.c (make_tree): Likewise.
17403 * expr.c (expand_expr_real_2): Likewise.
17404 (expand_expr_real_1): Likewise.
17405 (try_casesi): Likewise.
17406 * fold-const-call.c (fold_const_call_ss): Likewise.
17407 * fold-const.c (unextend): Likewise.
17408 (extract_muldiv_1): Likewise.
17409 (fold_single_bit_test): Likewise.
17410 (native_encode_int): Likewise.
17411 (native_encode_string): Likewise.
17412 (native_interpret_int): Likewise.
17413 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
17414 * internal-fn.c (expand_addsub_overflow): Likewise.
17415 (expand_neg_overflow): Likewise.
17416 (expand_mul_overflow): Likewise.
17417 (expand_arith_overflow): Likewise.
17418 * match.pd: Likewise.
17419 * stor-layout.c (layout_type): Likewise.
17420 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
17421 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
17422 * tree-ssanames.c (get_range_info): Likewise.
17423 * tree-switch-conversion.c (array_value_type) Likewise.
17424 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
17425 (vect_recog_divmod_pattern): Likewise.
17426 (vect_recog_mixed_size_cond_pattern): Likewise.
17427 * tree-vrp.c (extract_range_basic): Likewise.
17428 (simplify_float_conversion_using_ranges): Likewise.
17429 * tree.c (int_fits_type_p): Likewise.
17430 * ubsan.c (instrument_bool_enum_load): Likewise.
17431 * varasm.c (mergeable_string_section): Likewise.
17432 (narrowing_initializer_constant_valid_p): Likewise.
17433 (output_constant): Likewise.
17435 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17436 Alan Hayward <alan.hayward@arm.com>
17437 David Sherwood <david.sherwood@arm.com>
17439 * machmode.h (NARROWEST_INT_MODE): New macro.
17440 * expr.c (alignment_for_piecewise_move): Use it instead of
17441 GET_CLASS_NARROWEST_MODE (MODE_INT).
17442 (push_block): Likewise.
17443 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
17445 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
17447 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17448 Alan Hayward <alan.hayward@arm.com>
17449 David Sherwood <david.sherwood@arm.com>
17451 * postreload.c (move2add_valid_value_p): Change the type of the
17452 mode parameter to scalar_int_mode.
17453 (move2add_use_add2_insn): Add a mode parameter and use it instead
17455 (move2add_use_add3_insn): Likewise.
17456 (reload_cse_move2add): Update accordingly.
17458 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17459 Alan Hayward <alan.hayward@arm.com>
17460 David Sherwood <david.sherwood@arm.com>
17462 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
17464 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
17465 * optabs.c (expand_unop): Likewise.
17467 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17468 Alan Hayward <alan.hayward@arm.com>
17469 David Sherwood <david.sherwood@arm.com>
17471 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
17472 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
17473 (popcount_loc_descriptor): Likewise.
17474 (bswap_loc_descriptor): Likewise.
17475 (rotate_loc_descriptor): Likewise.
17476 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
17477 calling the functions above.
17479 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17480 Alan Hayward <alan.hayward@arm.com>
17481 David Sherwood <david.sherwood@arm.com>
17483 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
17485 (try_combine): Likewise.
17486 (simplify_if_then_else): Likewise.
17487 * cse.c (cse_insn): Likewise.
17488 * dwarf2out.c (mem_loc_descriptor): Likewise.
17489 * emit-rtl.c (gen_lowpart_common): Likewise.
17490 * simplify-rtx.c (simplify_truncation): Likewise.
17491 (simplify_binary_operation_1): Likewise.
17492 (simplify_const_relational_operation): Likewise.
17493 (simplify_ternary_operation): Likewise.
17494 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
17496 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17497 Alan Hayward <alan.hayward@arm.com>
17498 David Sherwood <david.sherwood@arm.com>
17500 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
17501 * reload.c (push_reload): Likewise.
17502 (find_reloads): Likewise.
17504 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17505 Alan Hayward <alan.hayward@arm.com>
17506 David Sherwood <david.sherwood@arm.com>
17508 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
17509 (make_compound_operation_int): Likewise.
17510 (change_zero_ext): Likewise.
17511 * expr.c (convert_move): Likewise.
17512 (convert_modes): Likewise.
17513 * fwprop.c (forward_propagate_subreg): Likewise.
17514 * loop-iv.c (get_biv_step_1): Likewise.
17515 * optabs.c (widen_operand): Likewise.
17516 * postreload.c (move2add_valid_value_p): Likewise.
17517 * recog.c (simplify_while_replacing): Likewise.
17518 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17519 (simplify_binary_operation_1): Likewise. Remove redundant
17520 mode equality check.
17522 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17523 Alan Hayward <alan.hayward@arm.com>
17524 David Sherwood <david.sherwood@arm.com>
17526 * combine.c (combine_simplify_rtx): Add checks for
17527 is_a <scalar_int_mode>.
17528 (simplify_if_then_else): Likewise.
17529 (make_field_assignment): Likewise.
17530 (simplify_comparison): Likewise.
17531 * ifcvt.c (noce_try_bitop): Likewise.
17532 * loop-invariant.c (canonicalize_address_mult): Likewise.
17533 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17535 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17536 Alan Hayward <alan.hayward@arm.com>
17537 David Sherwood <david.sherwood@arm.com>
17539 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
17540 is_a <scalar_int_mode> instead of != BLKmode.
17542 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17543 Alan Hayward <alan.hayward@arm.com>
17544 David Sherwood <david.sherwood@arm.com>
17546 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
17547 instead of != VOIDmode.
17548 * combine.c (if_then_else_cond): Likewise.
17549 (change_zero_ext): Likewise.
17550 * dwarf2out.c (mem_loc_descriptor): Likewise.
17551 (loc_descriptor): Likewise.
17552 * rtlanal.c (canonicalize_condition): Likewise.
17553 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
17555 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17556 Alan Hayward <alan.hayward@arm.com>
17557 David Sherwood <david.sherwood@arm.com>
17559 * simplify-rtx.c (simplify_binary_operation_1): Use
17560 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
17562 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17563 Alan Hayward <alan.hayward@arm.com>
17564 David Sherwood <david.sherwood@arm.com>
17566 * wide-int.h (int_traits<unsigned char>) New class.
17567 (int_traits<unsigned short>) Likewise.
17568 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
17569 Use GET_MODE_UNIT_PRECISION and remove redundant test for
17571 * combine.c (set_nonzero_bits_and_sign_copies): Use
17572 is_a <scalar_int_mode>.
17573 (find_split_point): Likewise.
17574 (combine_simplify_rtx): Likewise.
17575 (simplify_logical): Likewise.
17576 (expand_compound_operation): Likewise.
17577 (expand_field_assignment): Likewise.
17578 (make_compound_operation): Likewise.
17579 (extended_count): Likewise.
17580 (change_zero_ext): Likewise.
17581 (simplify_comparison): Likewise.
17582 * dwarf2out.c (scompare_loc_descriptor): Likewise.
17583 (ucompare_loc_descriptor): Likewise.
17584 (minmax_loc_descriptor): Likewise.
17585 (mem_loc_descriptor): Likewise.
17586 (loc_descriptor): Likewise.
17587 * expmed.c (init_expmed_one_mode): Likewise.
17588 * lra-constraints.c (lra_constraint_offset): Likewise.
17589 * optabs.c (prepare_libcall_arg): Likewise.
17590 * postreload.c (move2add_note_store): Likewise.
17591 * reload.c (operands_match_p): Likewise.
17592 * rtl.h (load_extend_op): Likewise.
17593 * rtlhooks.c (gen_lowpart_general): Likewise.
17594 * simplify-rtx.c (simplify_truncation): Likewise.
17595 (simplify_unary_operation_1): Likewise.
17596 (simplify_binary_operation_1): Likewise.
17597 (simplify_const_binary_operation): Likewise.
17598 (simplify_const_relational_operation): Likewise.
17599 (simplify_subreg): Likewise.
17600 * stor-layout.c (bitwise_mode_for_mode): Likewise.
17601 * var-tracking.c (adjust_mems): Likewise.
17602 (prepare_call_arguments): Likewise.
17604 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17605 Alan Hayward <alan.hayward@arm.com>
17606 David Sherwood <david.sherwood@arm.com>
17608 * machmode.h (is_int_mode): New fuction.
17609 * combine.c (find_split_point): Use it.
17610 (combine_simplify_rtx): Likewise.
17611 (simplify_if_then_else): Likewise.
17612 (simplify_set): Likewise.
17613 (simplify_shift_const_1): Likewise.
17614 (simplify_comparison): Likewise.
17615 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
17616 * cse.c (notreg_cost): Likewise.
17617 (cse_insn): Likewise.
17618 * cselib.c (cselib_lookup_1): Likewise.
17619 * dojump.c (do_jump_1): Likewise.
17620 (do_compare_rtx_and_jump): Likewise.
17621 * dse.c (get_call_args): Likewise.
17622 * dwarf2out.c (rtl_for_decl_init): Likewise.
17623 (native_encode_initializer): Likewise.
17624 * expmed.c (emit_store_flag_1): Likewise.
17625 (emit_store_flag): Likewise.
17626 * expr.c (convert_modes): Likewise.
17627 (store_field): Likewise.
17628 (expand_expr_real_1): Likewise.
17629 * fold-const.c (fold_read_from_constant_string): Likewise.
17630 * gimple-ssa-sprintf.c (get_format_string): Likewise.
17631 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
17632 * optabs.c (expand_binop): Likewise.
17633 (expand_unop): Likewise.
17634 (expand_abs_nojump): Likewise.
17635 (expand_one_cmpl_abs_nojump): Likewise.
17636 * simplify-rtx.c (mode_signbit_p): Likewise.
17637 (val_signbit_p): Likewise.
17638 (val_signbit_known_set_p): Likewise.
17639 (val_signbit_known_clear_p): Likewise.
17640 (simplify_relational_operation_1): Likewise.
17641 * tree.c (vector_type_mode): Likewise.
17643 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17644 Alan Hayward <alan.hayward@arm.com>
17645 David Sherwood <david.sherwood@arm.com>
17647 * machmode.h (smallest_mode_for_size): Fix formatting.
17648 (smallest_int_mode_for_size): New function.
17649 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
17650 instead of smallest_mode_for_size.
17651 * combine.c (make_extraction): Likewise.
17652 * config/arc/arc.c (arc_expand_movmem): Likewise.
17653 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
17654 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
17655 * config/s390/s390.c (s390_expand_insv): Likewise.
17656 * config/sparc/sparc.c (assign_int_registers): Likewise.
17657 * config/spu/spu.c (spu_function_value): Likewise.
17658 (spu_function_arg): Likewise.
17659 * coverage.c (get_gcov_type): Likewise.
17660 (get_gcov_unsigned_t): Likewise.
17661 * dse.c (find_shift_sequence): Likewise.
17662 * expmed.c (store_bit_field_1): Likewise.
17663 * expr.c (convert_move): Likewise.
17664 (store_field): Likewise.
17665 * internal-fn.c (expand_arith_overflow): Likewise.
17666 * optabs-query.c (get_best_extraction_insn): Likewise.
17667 * optabs.c (expand_twoval_binop_libfunc): Likewise.
17668 * stor-layout.c (layout_type): Likewise.
17669 (initialize_sizetypes): Likewise.
17670 * targhooks.c (default_get_mask_mode): Likewise.
17671 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
17673 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17674 Alan Hayward <alan.hayward@arm.com>
17675 David Sherwood <david.sherwood@arm.com>
17677 * machmode.h (opt_mode::else_blk): New function.
17678 (int_mode_for_mode): Declare.
17679 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
17680 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
17682 * cfgexpand.c (expand_debug_expr): Likewise.
17683 * combine.c (gen_lowpart_or_truncate): Likewise.
17684 (gen_lowpart_for_combine): Likewise.
17685 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
17686 * config/avr/avr.c (avr_to_int_mode): Likewise.
17687 (avr_out_plus_1): Likewise.
17688 (avr_out_plus): Likewise.
17689 (avr_out_round): Likewise.
17690 * config/i386/i386.c (ix86_split_to_parts): Likewise.
17691 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
17692 (s390_expand_vcond): Likewise.
17693 * config/spu/spu.c (spu_split_immediate): Likewise.
17694 (spu_expand_mov): Likewise.
17695 * dse.c (get_stored_val): Likewise.
17696 * expmed.c (store_bit_field_1): Likewise.
17697 (convert_extracted_bit_field): Use int_mode_for_mode instead of
17699 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
17700 (extract_low_bits): Likewise.
17701 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
17702 handling rather than repeating the check.
17703 (emit_group_store): Likewise.
17704 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
17705 * optabs.c (expand_absneg_bit): Likewise.
17706 (expand_copysign_absneg): Likewise.
17707 (expand_copysign_bit): Likewise.
17708 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
17709 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
17710 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
17711 * var-tracking.c (prepare_call_arguments): Likewise.
17712 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
17713 int_mode_for_mode instead of mode_for_size.
17714 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
17716 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17717 Alan Hayward <alan.hayward@arm.com>
17718 David Sherwood <david.sherwood@arm.com>
17720 * machmode.h (int_mode_for_size): New function.
17721 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
17722 instead of mode_for_size.
17723 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
17725 * combine.c (expand_field_assignment): Use int_mode_for_size
17726 instead of mode_for_size.
17727 (make_extraction): Likewise.
17728 (simplify_shift_const_1): Likewise.
17729 (simplify_comparison): Likewise.
17730 * dojump.c (do_jump): Likewise.
17731 * dwarf2out.c (mem_loc_descriptor): Likewise.
17732 * emit-rtl.c (init_derived_machine_modes): Likewise.
17733 * expmed.c (flip_storage_order): Likewise.
17734 (convert_extracted_bit_field): Likewise.
17735 * expr.c (copy_blkmode_from_reg): Likewise.
17736 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
17737 * internal-fn.c (expand_mul_overflow): Likewise.
17738 * lower-subreg.c (simple_move): Likewise.
17739 * optabs-libfuncs.c (init_optabs): Likewise.
17740 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17741 * tree.c (vector_type_mode): Likewise.
17742 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
17743 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
17744 * tree-vect-generic.c (expand_vector_parallel): Likewise.
17745 * tree-vect-stmts.c (vectorizable_load): Likewise.
17746 (vectorizable_store): Likewise.
17748 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17749 Alan Hayward <alan.hayward@arm.com>
17750 David Sherwood <david.sherwood@arm.com>
17752 * coretypes.h (pod_mode): New type.
17753 (scalar_int_mode_pod): New typedef.
17754 * machmode.h (pod_mode): New class.
17755 (int_n_data_t::m): Change type to scalar_int_mode_pod.
17756 * genmodes.c (emit_mode_int_n): Update accordingly.
17757 * lower-subreg.h (target_lower_subreg): Change type to
17758 scalar_int_mode_pod.
17759 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
17760 scalar_int_mode_pod.
17762 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17763 Alan Hayward <alan.hayward@arm.com>
17764 David Sherwood <david.sherwood@arm.com>
17766 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
17767 machine_mode to scalar_int_mode.
17768 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
17769 (rs6000_option_override_internal): Remove cast to int.
17770 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
17771 machine_mode to scalar_int_mode.
17772 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
17773 (rs6000_option_override_internal): Remove cast to int.
17774 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
17775 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
17777 * config/s390/s390.c (s390_expand_builtin): Likewise.
17778 * coretypes.h (scalar_int_mode): New type.
17779 (opt_scalar_int_mode): New typedef.
17780 * machmode.h (scalar_int_mode): New class.
17781 (scalar_int_mode::includes_p): New function.
17782 (byte_mode): Change type to scalar_int_mode.
17783 (word_mode): Likewise.
17784 (ptr_mode): Likewise.
17785 * emit-rtl.c (byte_mode): Likewise.
17786 (word_mode): Likewise.
17787 (ptr_mode): Likewise.
17788 (init_derived_machine_modes): Update accordingly.
17789 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
17790 and MODE_PARTIAL_INT.
17791 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
17792 opt_scalar_int_mode.
17794 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17795 Alan Hayward <alan.hayward@arm.com>
17796 David Sherwood <david.sherwood@arm.com>
17798 * target.def (libgcc_floating_mode_supported_p): Take a
17800 * doc/tm.texi: Regenerate.
17801 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
17803 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
17804 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
17807 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17808 Alan Hayward <alan.hayward@arm.com>
17809 David Sherwood <david.sherwood@arm.com>
17811 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
17812 * doc/tm.texi: Regenerate.
17813 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
17814 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
17815 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
17816 * targhooks.h (default_floatn_mode): Likewise.
17817 * targhooks.c (default_floatn_mode): Likewise.
17818 * tree.c (build_common_tree_nodes): Update accordingly.
17820 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17821 Alan Hayward <alan.hayward@arm.com>
17822 David Sherwood <david.sherwood@arm.com>
17824 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
17825 (mode_iterator::iterate_p): Likewise.
17826 (mode_iterator::get_wider): Likewise.
17827 * expr.c (init_expr_target): Use opt_scalar_float_mode.
17829 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17830 Alan Hayward <alan.hayward@arm.com>
17831 David Sherwood <david.sherwood@arm.com>
17833 * coretypes.h (opt_scalar_float_mode): New typedef.
17834 * machmode.h (float_mode_for_size): New function.
17835 * emit-rtl.c (double_mode): Delete.
17836 (init_emit_once): Use float_mode_for_size.
17837 * stor-layout.c (layout_type): Likewise.
17838 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
17840 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17841 Alan Hayward <alan.hayward@arm.com>
17842 David Sherwood <david.sherwood@arm.com>
17844 * output.h (assemble_real): Take a scalar_float_mode.
17845 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
17846 * config/arm/arm.md (consttable_4): Likewise.
17847 (consttable_8): Likewise.
17848 (consttable_16): Likewise.
17849 * config/mips/mips.md (consttable_float): Likewise.
17850 * config/s390/s390.c (s390_output_pool_entry): Likewise.
17851 * varasm.c (assemble_real): Take a scalar_float_mode.
17852 (output_constant_pool_2): Update accordingly.
17853 (output_constant): Likewise.
17855 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17856 Alan Hayward <alan.hayward@arm.com>
17857 David Sherwood <david.sherwood@arm.com>
17859 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
17860 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
17861 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
17862 (native_encode_real): Likewise.
17863 (native_interpret_real): Likewise.
17864 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
17865 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
17867 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17868 Alan Hayward <alan.hayward@arm.com>
17869 David Sherwood <david.sherwood@arm.com>
17871 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
17872 <scalar_float_mode>. Simplify.
17873 (gen_extend_conv_libfunc): Likewise.
17875 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17876 Alan Hayward <alan.hayward@arm.com>
17877 David Sherwood <david.sherwood@arm.com>
17879 * coretypes.h (scalar_float_mode): New type.
17880 * machmode.h (mode_traits::from_int): Use machine_mode if
17881 USE_ENUM_MODES is defined.
17882 (is_a): New function.
17884 (dyn_cast): Likewise.
17885 (scalar_float_mode): New class.
17886 (scalar_float_mode::includes_p): New function.
17887 (is_float_mode): Likewise.
17888 * gdbhooks.py (MachineModePrinter): New class.
17889 (build_pretty_printer): Use it for scalar_float_mode.
17890 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
17891 (format_helper::format_helper): Turn into a template.
17892 * genmodes.c (get_mode_class): New function.
17893 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
17894 or machine_mode if none.
17895 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
17896 as_a <scalar_float_mode>.
17897 * dwarf2out.c (mem_loc_descriptor): Likewise.
17898 (insert_float): Likewise.
17899 (add_const_value_attribute): Likewise.
17900 * simplify-rtx.c (simplify_immed_subreg): Likewise.
17901 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
17902 (expand_unop): Update accordingly.
17903 (expand_abs_nojump): Likewise.
17904 (expand_copysign_absneg): Take a scalar_float_mode.
17905 (expand_copysign_bit): Likewise.
17906 (expand_copysign): Update accordingly.
17908 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17909 Alan Hayward <alan.hayward@arm.com>
17910 David Sherwood <david.sherwood@arm.com>
17912 * coretypes.h (opt_mode): New class.
17913 * machmode.h (opt_mode): Likewise.
17914 (opt_mode::else_void): New function.
17915 (opt_mode::require): Likewise.
17916 (opt_mode::exists): Likewise.
17917 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
17918 (GET_MODE_2XWIDER_MODE): Likewise.
17919 (mode_iterator::get_wider): Update accordingly.
17920 (mode_iterator::get_2xwider): Likewise.
17921 (mode_iterator::get_known_wider): Likewise, turning into a template.
17922 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
17923 forcing a wider mode to exist.
17924 * config/cr16/cr16.h (LONG_REG_P): Likewise.
17925 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
17926 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
17927 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
17928 * lower-subreg.c (init_lower_subreg): Likewise.
17929 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
17930 on the final iteration.
17931 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
17932 a wider mode exists before asking for a move pattern.
17933 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
17934 forcing a wider mode to exist.
17935 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
17936 returning false if no such mode exists.
17937 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
17938 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
17939 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
17940 Avoid checking for a MODE_INT if we already know the mode is not a
17942 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
17943 forcing a wider mode to exist.
17944 (expmed_mult_highpart_optab): Likewise.
17945 (expmed_mult_highpart): Likewise.
17946 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
17948 * lto-streamer-in.c (lto_input_mode_table): Likewise.
17949 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
17950 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
17951 * internal-fn.c (expand_mul_overflow): Update use of
17952 GET_MODE_2XWIDER_MODE.
17953 * omp-low.c (omp_clause_aligned_alignment): Likewise.
17954 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
17955 GET_MODE_WIDER_MODE.
17956 (convert_plusminus_to_widen): Likewise.
17957 * tree-switch-conversion.c (array_value_type): Likewise.
17958 * var-tracking.c (emit_note_insn_var_location): Likewise.
17959 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
17960 Return false inside rather than outside the loop if no wider mode
17962 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
17963 and GET_MODE_2XWIDER_MODE
17964 (can_compare_p): Use else_void.
17965 * gdbhooks.py (OptMachineModePrinter): New class.
17966 (build_pretty_printer): Use it for opt_mode.
17968 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17969 Alan Hayward <alan.hayward@arm.com>
17970 David Sherwood <david.sherwood@arm.com>
17972 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
17973 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
17975 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17976 Alan Hayward <alan.hayward@arm.com>
17977 David Sherwood <david.sherwood@arm.com>
17979 * machmode.h (mode_traits): New structure.
17980 (get_narrowest_mode): New function.
17981 (mode_iterator::start): Likewise.
17982 (mode_iterator::iterate_p): Likewise.
17983 (mode_iterator::get_wider): Likewise.
17984 (mode_iterator::get_known_wider): Likewise.
17985 (mode_iterator::get_2xwider): Likewise.
17986 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
17987 (FOR_EACH_MODE): Likewise.
17988 (FOR_EACH_MODE_FROM): Likewise.
17989 (FOR_EACH_MODE_UNTIL): Likewise.
17990 (FOR_EACH_WIDER_MODE): Likewise.
17991 (FOR_EACH_2XWIDER_MODE): Likewise.
17992 * builtins.c (expand_builtin_strlen): Use new mode iterators.
17993 * combine.c (simplify_comparison): Likewise
17994 * config/i386/i386.c (type_natural_mode): Likewise.
17995 * cse.c (cse_insn): Likewise.
17996 * dse.c (find_shift_sequence): Likewise.
17997 * emit-rtl.c (init_derived_machine_modes): Likewise.
17998 (init_emit_once): Likewise.
17999 * explow.c (hard_function_value): Likewise.
18000 * expmed.c (extract_fixed_bit_field_1): Likewise.
18001 (extract_bit_field_1): Likewise.
18002 (expand_divmod): Likewise.
18003 (emit_store_flag_1): Likewise.
18004 * expr.c (init_expr_target): Likewise.
18005 (convert_move): Likewise.
18006 (alignment_for_piecewise_move): Likewise.
18007 (widest_int_mode_for_size): Likewise.
18008 (emit_block_move_via_movmem): Likewise.
18009 (copy_blkmode_to_reg): Likewise.
18010 (set_storage_via_setmem): Likewise.
18011 (compress_float_constant): Likewise.
18012 * omp-low.c (omp_clause_aligned_alignment): Likewise.
18013 * optabs-query.c (get_best_extraction_insn): Likewise.
18014 * optabs.c (expand_binop): Likewise.
18015 (expand_twoval_unop): Likewise.
18016 (expand_twoval_binop): Likewise.
18017 (widen_leading): Likewise.
18018 (widen_bswap): Likewise.
18019 (expand_parity): Likewise.
18020 (expand_unop): Likewise.
18021 (prepare_cmp_insn): Likewise.
18022 (prepare_float_lib_cmp): Likewise.
18023 (expand_float): Likewise.
18024 (expand_fix): Likewise.
18025 (expand_sfix_optab): Likewise.
18026 * postreload.c (move2add_use_add2_insn): Likewise.
18027 * reg-stack.c (reg_to_stack): Likewise.
18028 * reginfo.c (choose_hard_reg_mode): Likewise.
18029 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
18030 * stor-layout.c (mode_for_size): Likewise.
18031 (smallest_mode_for_size): Likewise.
18032 (mode_for_vector): Likewise.
18033 (finish_bitfield_representative): Likewise.
18034 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
18035 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
18036 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
18037 * var-tracking.c (prepare_call_arguments): Likewise.
18039 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18040 Alan Hayward <alan.hayward@arm.com>
18041 David Sherwood <david.sherwood@arm.com>
18043 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
18044 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
18045 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
18046 * machmode.h (mode_size): Move earlier in file.
18047 (mode_precision): Likewise.
18048 (mode_inner): Likewise.
18049 (mode_nunits): Likewise.
18050 (mode_unit_size): Likewise.
18051 (unit_unit_precision): Likewise.
18052 (mode_wider): Likewise.
18053 (mode_2xwider): Likewise.
18054 (machine_mode): New class.
18055 (mode_to_bytes): New function.
18056 (mode_to_bits): Likewise.
18057 (mode_to_precision): Likewise.
18058 (mode_to_inner): Likewise.
18059 (mode_to_unit_size): Likewise.
18060 (mode_to_unit_precision): Likewise.
18061 (mode_to_nunits): Likewise.
18062 (GET_MODE_SIZE): Use mode_to_bytes.
18063 (GET_MODE_BITSIZE): Use mode_to_bits.
18064 (GET_MODE_PRECISION): Use mode_to_precision.
18065 (GET_MODE_INNER): Use mode_to_inner.
18066 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
18067 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
18068 (GET_MODE_NUNITS): Use mode_to_nunits.
18069 * system.h (ALWAYS_INLINE): New macro.
18070 * config/powerpcspe/powerpcspe-c.c
18071 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
18072 int for arg1_mode and arg2_mode.
18074 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18075 Alan Hayward <alan.hayward@arm.com>
18076 David Sherwood <david.sherwood@arm.com>
18078 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
18079 Prefix mode names with E_ in case statements.
18080 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18081 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
18082 (aarch64_split_simd_move): Likewise.
18083 (aarch64_gen_storewb_pair): Likewise.
18084 (aarch64_gen_loadwb_pair): Likewise.
18085 (aarch64_gen_store_pair): Likewise.
18086 (aarch64_gen_load_pair): Likewise.
18087 (aarch64_get_condition_code_1): Likewise.
18088 (aarch64_constant_pool_reload_icode): Likewise.
18089 (get_rsqrte_type): Likewise.
18090 (get_rsqrts_type): Likewise.
18091 (get_recpe_type): Likewise.
18092 (get_recps_type): Likewise.
18093 (aarch64_gimplify_va_arg_expr): Likewise.
18094 (aarch64_simd_container_mode): Likewise.
18095 (aarch64_emit_load_exclusive): Likewise.
18096 (aarch64_emit_store_exclusive): Likewise.
18097 (aarch64_expand_compare_and_swap): Likewise.
18098 (aarch64_gen_atomic_cas): Likewise.
18099 (aarch64_emit_bic): Likewise.
18100 (aarch64_emit_atomic_swap): Likewise.
18101 (aarch64_emit_atomic_load_op): Likewise.
18102 (aarch64_evpc_trn): Likewise.
18103 (aarch64_evpc_uzp): Likewise.
18104 (aarch64_evpc_zip): Likewise.
18105 (aarch64_evpc_ext): Likewise.
18106 (aarch64_evpc_rev): Likewise.
18107 (aarch64_evpc_dup): Likewise.
18108 (aarch64_gen_ccmp_first): Likewise.
18109 (aarch64_gen_ccmp_next): Likewise.
18110 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
18111 (alpha_emit_xfloating_libcall): Likewise.
18112 (emit_insxl): Likewise.
18113 (alpha_arg_type): Likewise.
18114 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
18115 (arc_preferred_simd_mode): Likewise.
18116 (arc_secondary_reload): Likewise.
18117 (get_arc_condition_code): Likewise.
18118 (arc_print_operand): Likewise.
18119 (arc_legitimate_constant_p): Likewise.
18120 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18121 * config/arc/arc.md (casesi_load): Likewise.
18122 (casesi_compact_jump): Likewise.
18123 * config/arc/predicates.md (proper_comparison_operator): Likewise.
18124 (cc_use_register): Likewise.
18125 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18126 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
18127 (arm_init_iwmmxt_builtins): Likewise.
18128 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
18129 (neon_expand_vector_init): Likewise.
18130 (arm_attr_length_move_neon): Likewise.
18131 (maybe_get_arm_condition_code): Likewise.
18132 (arm_emit_vector_const): Likewise.
18133 (arm_preferred_simd_mode): Likewise.
18134 (arm_output_iwmmxt_tinsr): Likewise.
18135 (thumb1_output_casesi): Likewise.
18136 (thumb2_output_casesi): Likewise.
18137 (arm_emit_load_exclusive): Likewise.
18138 (arm_emit_store_exclusive): Likewise.
18139 (arm_expand_compare_and_swap): Likewise.
18140 (arm_evpc_neon_vuzp): Likewise.
18141 (arm_evpc_neon_vzip): Likewise.
18142 (arm_evpc_neon_vrev): Likewise.
18143 (arm_evpc_neon_vtrn): Likewise.
18144 (arm_evpc_neon_vext): Likewise.
18145 (arm_validize_comparison): Likewise.
18146 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
18147 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
18148 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
18149 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
18150 (c6x_preferred_simd_mode): Likewise.
18151 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
18152 (epiphany_rtx_costs): Likewise.
18153 * config/epiphany/predicates.md (proper_comparison_operator):
18155 * config/frv/frv.c (condexec_memory_operand): Likewise.
18156 (frv_emit_move): Likewise.
18157 (output_move_single): Likewise.
18158 (output_condmove_single): Likewise.
18159 (frv_hard_regno_mode_ok): Likewise.
18160 (frv_matching_accg_mode): Likewise.
18161 * config/h8300/h8300.c (split_adds_subs): Likewise.
18162 (h8300_rtx_costs): Likewise.
18163 (h8300_print_operand): Likewise.
18164 (compute_mov_length): Likewise.
18165 (output_logical_op): Likewise.
18166 (compute_logical_op_length): Likewise.
18167 (compute_logical_op_cc): Likewise.
18168 (h8300_shift_needs_scratch_p): Likewise.
18169 (output_a_shift): Likewise.
18170 (compute_a_shift_length): Likewise.
18171 (compute_a_shift_cc): Likewise.
18172 (expand_a_rotate): Likewise.
18173 (output_a_rotate): Likewise.
18174 * config/i386/i386.c (classify_argument): Likewise.
18175 (function_arg_advance_32): Likewise.
18176 (function_arg_32): Likewise.
18177 (function_arg_64): Likewise.
18178 (function_value_64): Likewise.
18179 (ix86_gimplify_va_arg): Likewise.
18180 (ix86_legitimate_constant_p): Likewise.
18181 (put_condition_code): Likewise.
18182 (split_double_mode): Likewise.
18183 (ix86_avx256_split_vector_move_misalign): Likewise.
18184 (ix86_expand_vector_logical_operator): Likewise.
18185 (ix86_split_idivmod): Likewise.
18186 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
18187 (ix86_build_const_vector): Likewise.
18188 (ix86_build_signbit_mask): Likewise.
18189 (ix86_match_ccmode): Likewise.
18190 (ix86_cc_modes_compatible): Likewise.
18191 (ix86_expand_branch): Likewise.
18192 (ix86_expand_sse_cmp): Likewise.
18193 (ix86_expand_sse_movcc): Likewise.
18194 (ix86_expand_int_sse_cmp): Likewise.
18195 (ix86_expand_vec_perm_vpermi2): Likewise.
18196 (ix86_expand_vec_perm): Likewise.
18197 (ix86_expand_sse_unpack): Likewise.
18198 (ix86_expand_int_addcc): Likewise.
18199 (ix86_split_to_parts): Likewise.
18200 (ix86_vectorize_builtin_gather): Likewise.
18201 (ix86_vectorize_builtin_scatter): Likewise.
18202 (avx_vpermilp_parallel): Likewise.
18203 (inline_memory_move_cost): Likewise.
18204 (ix86_tieable_integer_mode_p): Likewise.
18205 (x86_maybe_negate_const_int): Likewise.
18206 (ix86_expand_vector_init_duplicate): Likewise.
18207 (ix86_expand_vector_init_one_nonzero): Likewise.
18208 (ix86_expand_vector_init_one_var): Likewise.
18209 (ix86_expand_vector_init_concat): Likewise.
18210 (ix86_expand_vector_init_interleave): Likewise.
18211 (ix86_expand_vector_init_general): Likewise.
18212 (ix86_expand_vector_set): Likewise.
18213 (ix86_expand_vector_extract): Likewise.
18214 (emit_reduc_half): Likewise.
18215 (ix86_emit_i387_round): Likewise.
18216 (ix86_mangle_type): Likewise.
18217 (ix86_expand_round_sse4): Likewise.
18218 (expand_vec_perm_blend): Likewise.
18219 (canonicalize_vector_int_perm): Likewise.
18220 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
18221 (expand_vec_perm_1): Likewise.
18222 (expand_vec_perm_interleave3): Likewise.
18223 (expand_vec_perm_even_odd_pack): Likewise.
18224 (expand_vec_perm_even_odd_1): Likewise.
18225 (expand_vec_perm_broadcast_1): Likewise.
18226 (ix86_vectorize_vec_perm_const_ok): Likewise.
18227 (ix86_expand_vecop_qihi): Likewise.
18228 (ix86_expand_mul_widen_hilo): Likewise.
18229 (ix86_expand_sse2_abs): Likewise.
18230 (ix86_expand_pextr): Likewise.
18231 (ix86_expand_pinsr): Likewise.
18232 (ix86_preferred_simd_mode): Likewise.
18233 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
18234 * config/i386/sse.md (*andnot<mode>3): Likewise.
18235 (<mask_codefor><code><mode>3<mask_name>): Likewise.
18236 (*<code><mode>3): Likewise.
18237 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
18238 (ia64_expand_atomic_op): Likewise.
18239 (ia64_arg_type): Likewise.
18240 (ia64_mode_to_int): Likewise.
18241 (ia64_scalar_mode_supported_p): Likewise.
18242 (ia64_vector_mode_supported_p): Likewise.
18243 (expand_vec_perm_broadcast): Likewise.
18244 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
18245 (iq2000_function_arg_advance): Likewise.
18246 (iq2000_function_arg): Likewise.
18247 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
18248 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
18249 (m68k_libcall_value): Likewise.
18250 (m68k_function_value): Likewise.
18251 (sched_attr_op_type): Likewise.
18252 * config/mcore/mcore.c (mcore_output_move): Likewise.
18253 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
18255 (microblaze_function_arg): Likewise.
18256 * config/mips/mips.c (mips16_build_call_stub): Likewise.
18257 (mips_print_operand): Likewise.
18258 (mips_mode_ok_for_mov_fmt_p): Likewise.
18259 (mips_vector_mode_supported_p): Likewise.
18260 (mips_preferred_simd_mode): Likewise.
18261 (mips_expand_vpc_loongson_even_odd): Likewise.
18262 (mips_expand_vec_unpack): Likewise.
18263 (mips_expand_vi_broadcast): Likewise.
18264 (mips_expand_vector_init): Likewise.
18265 (mips_expand_vec_reduc): Likewise.
18266 (mips_expand_msa_cmp): Likewise.
18267 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
18268 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
18269 (cc_flags_for_mode): Likewise.
18270 * config/msp430/msp430.c (msp430_print_operand): Likewise.
18271 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
18272 (nds32_output_casesi_pc_relative): Likewise.
18273 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18274 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
18275 (nvptx_gen_unpack): Likewise.
18276 (nvptx_gen_pack): Likewise.
18277 (nvptx_gen_shuffle): Likewise.
18278 (nvptx_gen_wcast): Likewise.
18279 (nvptx_preferred_simd_mode): Likewise.
18280 * config/pa/pa.c (pa_secondary_reload): Likewise.
18281 * config/pa/predicates.md (base14_operand): Likewise.
18282 * config/powerpcspe/powerpcspe-c.c
18283 (altivec_resolve_overloaded_builtin): Likewise.
18284 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
18286 (rs6000_preferred_simd_mode): Likewise.
18287 (output_vec_const_move): Likewise.
18288 (rs6000_expand_vector_extract): Likewise.
18289 (rs6000_split_vec_extract_var): Likewise.
18290 (reg_offset_addressing_ok_p): Likewise.
18291 (rs6000_legitimate_offset_address_p): Likewise.
18292 (rs6000_legitimize_address): Likewise.
18293 (rs6000_emit_set_const): Likewise.
18294 (rs6000_const_vec): Likewise.
18295 (rs6000_emit_move): Likewise.
18296 (spe_build_register_parallel): Likewise.
18297 (rs6000_darwin64_record_arg_recurse): Likewise.
18298 (swap_selector_for_mode): Likewise.
18299 (spe_init_builtins): Likewise.
18300 (paired_init_builtins): Likewise.
18301 (altivec_init_builtins): Likewise.
18302 (do_load_for_compare): Likewise.
18303 (rs6000_generate_compare): Likewise.
18304 (rs6000_expand_float128_convert): Likewise.
18305 (emit_load_locked): Likewise.
18306 (emit_store_conditional): Likewise.
18307 (rs6000_output_function_epilogue): Likewise.
18308 (rs6000_handle_altivec_attribute): Likewise.
18309 (rs6000_function_value): Likewise.
18310 (emit_fusion_gpr_load): Likewise.
18311 (emit_fusion_p9_load): Likewise.
18312 (emit_fusion_p9_store): Likewise.
18313 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
18314 (fusion_gpr_mem_load): Likewise.
18315 (fusion_addis_mem_combo_load): Likewise.
18316 (fusion_addis_mem_combo_store): Likewise.
18317 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
18318 (fusion_gpr_mem_load): Likewise.
18319 (fusion_addis_mem_combo_load): Likewise.
18320 (fusion_addis_mem_combo_store): Likewise.
18321 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18323 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
18324 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
18325 (rs6000_preferred_simd_mode): Likewise.
18326 (output_vec_const_move): Likewise.
18327 (rs6000_expand_vector_extract): Likewise.
18328 (rs6000_split_vec_extract_var): Likewise.
18329 (reg_offset_addressing_ok_p): Likewise.
18330 (rs6000_legitimate_offset_address_p): Likewise.
18331 (rs6000_legitimize_address): Likewise.
18332 (rs6000_emit_set_const): Likewise.
18333 (rs6000_const_vec): Likewise.
18334 (rs6000_emit_move): Likewise.
18335 (rs6000_darwin64_record_arg_recurse): Likewise.
18336 (swap_selector_for_mode): Likewise.
18337 (paired_init_builtins): Likewise.
18338 (altivec_init_builtins): Likewise.
18339 (rs6000_expand_float128_convert): Likewise.
18340 (emit_load_locked): Likewise.
18341 (emit_store_conditional): Likewise.
18342 (rs6000_output_function_epilogue): Likewise.
18343 (rs6000_handle_altivec_attribute): Likewise.
18344 (rs6000_function_value): Likewise.
18345 (emit_fusion_gpr_load): Likewise.
18346 (emit_fusion_p9_load): Likewise.
18347 (emit_fusion_p9_store): Likewise.
18348 * config/rx/rx.c (rx_gen_move_template): Likewise.
18349 (flags_from_mode): Likewise.
18350 * config/s390/predicates.md (s390_alc_comparison): Likewise.
18351 (s390_slb_comparison): Likewise.
18352 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
18353 (s390_vector_mode_supported_p): Likewise.
18354 (s390_cc_modes_compatible): Likewise.
18355 (s390_match_ccmode_set): Likewise.
18356 (s390_canonicalize_comparison): Likewise.
18357 (s390_emit_compare_and_swap): Likewise.
18358 (s390_branch_condition_mask): Likewise.
18359 (s390_rtx_costs): Likewise.
18360 (s390_secondary_reload): Likewise.
18361 (__SECONDARY_RELOAD_CASE): Likewise.
18362 (s390_expand_cs): Likewise.
18363 (s390_preferred_simd_mode): Likewise.
18364 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
18365 * config/sh/sh.c (sh_print_operand): Likewise.
18366 (dump_table): Likewise.
18367 (sh_secondary_reload): Likewise.
18368 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18369 * config/sh/sh.md (casesi_worker_1): Likewise.
18370 (casesi_worker_2): Likewise.
18371 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
18372 (fcc_comparison_operator): Likewise.
18373 * config/sparc/sparc.c (sparc_expand_move): Likewise.
18374 (emit_soft_tfmode_cvt): Likewise.
18375 (sparc_preferred_simd_mode): Likewise.
18376 (output_cbranch): Likewise.
18377 (sparc_print_operand): Likewise.
18378 (sparc_expand_vec_perm_bmask): Likewise.
18379 (vector_init_bshuffle): Likewise.
18380 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
18381 (spu_vector_mode_supported_p): Likewise.
18382 (spu_expand_insv): Likewise.
18383 (spu_emit_branch_or_set): Likewise.
18384 (spu_handle_vector_attribute): Likewise.
18385 (spu_builtin_splats): Likewise.
18386 (spu_builtin_extract): Likewise.
18387 (spu_builtin_promote): Likewise.
18388 (spu_expand_sign_extend): Likewise.
18389 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
18390 (tilegx_simd_int): Likewise.
18391 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
18392 (tilepro_simd_int): Likewise.
18393 * config/v850/v850.c (const_double_split): Likewise.
18394 (v850_print_operand): Likewise.
18395 (ep_memory_offset): Likewise.
18396 * config/vax/vax.c (vax_rtx_costs): Likewise.
18397 (vax_output_int_move): Likewise.
18398 (vax_output_int_add): Likewise.
18399 (vax_output_int_subtract): Likewise.
18400 * config/visium/predicates.md (visium_branch_operator): Likewise.
18401 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
18402 (visium_print_operand_address): Likewise.
18403 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18404 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
18405 (xtensa_expand_conditional_branch): Likewise.
18406 (xtensa_copy_incoming_a7): Likewise.
18407 (xtensa_output_literal): Likewise.
18408 * dfp.c (decimal_real_maxval): Likewise.
18409 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
18411 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18412 Alan Hayward <alan.hayward@arm.com>
18413 David Sherwood <david.sherwood@arm.com>
18415 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
18416 (mode_nunits_inline): Likewise.
18417 (mode_inner_inline): Likewise.
18418 (mode_unit_size_inline): Likewise.
18419 (mode_unit_precision_inline): Likewise.
18420 (emit_insn_modes_h): Likewise. Also emit a #define of the
18422 (emit_mode_wider): Add an E_ prefix to mode names.
18423 (emit_mode_complex): Likewise.
18424 (emit_mode_inner): Likewise.
18425 (emit_mode_adjustments): Likewise.
18426 (emit_mode_int_n): Likewise.
18427 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
18428 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
18429 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
18430 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
18431 (CRC32_BUILTIN, ENTRY): Likewise.
18432 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
18433 (aarch64_pop_regs): Likewise.
18434 (aarch64_process_components): Likewise.
18435 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
18436 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
18437 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
18438 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
18439 * config/arm/arm.c (arm_init_libfuncs): Likewise.
18440 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
18442 * config/i386/i386-builtin.def (pcmpestr): Likewise.
18443 (pcmpistr): Likewise.
18444 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
18445 * config/mmix/mmix.c (mmix_output_condition): Likewise.
18446 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
18448 * config/rl78/rl78.c (mduc_regs): Likewise.
18449 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
18450 (htm_expand_builtin): Likewise.
18451 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
18452 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
18453 * config/xtensa/xtensa.c (print_operand): Likewise.
18454 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
18455 (NUM_MODE_VECTOR_INT): Likewise.
18456 * genoutput.c (null_operand): Likewise.
18457 (output_operand_data): Likewise.
18458 * genrecog.c (print_parameter_value): Likewise.
18459 * lra.c (debug_operand_data): Likewise.
18461 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18463 * dwarf2out.c (scompare_loc_descriptor_wide)
18464 (scompare_loc_descriptor_narrow): New functions, split out from...
18465 (scompare_loc_descriptor): ...here.
18466 * expmed.c (emit_store_flag_int): New function, split out from...
18467 (emit_store_flag): ...here.
18469 2017-08-30 Richard Biener <rguenther@suse.de>
18471 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
18472 (dwarf2out_early_finish): Move setting of AT_pubnames from
18473 early debug output to early finish.
18475 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
18477 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
18478 and -mdata-region to the assembler.
18480 2017-08-30 Richard Biener <rguenther@suse.de>
18482 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
18484 (gen_subprogram_die): Add DW_AT_object_pointer only early.
18485 (dwarf2out_early_global_decl): Only generate a DIE for the
18486 abstract origin if it doesn't already exist or is a declaration DIE.
18487 (resolve_addr): Do not add the linkage name twice when
18488 generating a stub DIE for the DW_TAG_GNU_call_site target.
18490 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18492 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18493 Use machine_mode rather than int for arg1_mode.
18495 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
18498 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
18499 that the second argument of the built-in functions to unpack
18500 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
18501 switch statement instead a lot of if statements.
18502 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
18503 Allow 64-bit values to be in Altivec registers as well as
18504 traditional floating point registers.
18505 (pack<mode>, FMOVE128_VSX iterator): Likewise.
18507 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
18509 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
18510 MAX_REGS_PER_ADDRESS == 1.
18512 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
18514 * config/i386/i386.opt (flag_fentry): Do not init to -1.
18515 * config/i386/i386.c (ix86_option_override_internal): Simplify
18516 setting of opts->x_flag_entry.
18518 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18519 Jakub Jelinek <jakub@redhat.com>
18520 Richard Biener <rguenther@suse.de>
18522 PR tree-optimization/81503
18523 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
18524 folded constant fits in the target type; reorder tests for clarity.
18526 2017-08-29 Martin Liska <mliska@suse.cz>
18528 * passes.def: Include pass_lower_switch.
18529 * stmt.c (dump_case_nodes): Remove and move to
18530 tree-switch-conversion.
18531 (case_values_threshold): Likewise.
18532 (expand_switch_as_decision_tree_p): Likewise.
18533 (emit_case_decision_tree): Likewise.
18534 (expand_case): Likewise.
18535 (balance_case_nodes): Likewise.
18536 (node_has_low_bound): Likewise.
18537 (node_has_high_bound): Likewise.
18538 (node_is_bounded): Likewise.
18539 (emit_case_nodes): Likewise.
18540 (struct simple_case_node): New struct.
18541 (add_case_node): Remove.
18542 (emit_case_dispatch_table): Use vector instead of case_list.
18543 (reset_out_edges_aux): Remove.
18544 (compute_cases_per_edge): Likewise.
18545 (expand_case): Build list of simple_case_node.
18546 (expand_sjlj_dispatch_table): Use it.
18547 * tree-switch-conversion.c (struct case_node): Moved from
18548 stmt.c and adjusted.
18549 (emit_case_nodes): Likewise.
18550 (node_has_low_bound): Likewise.
18551 (node_has_high_bound): Likewise.
18552 (node_is_bounded): Likewise.
18553 (case_values_threshold): Likewise.
18554 (reset_out_edges_aux): Likewise.
18555 (compute_cases_per_edge): Likewise.
18556 (add_case_node): Likewise.
18557 (dump_case_nodes): Likewise.
18558 (balance_case_nodes): Likewise.
18559 (expand_switch_as_decision_tree_p): Likewise.
18560 (emit_jump): Likewise.
18561 (emit_case_decision_tree): Likewise.
18562 (try_switch_expansion): Likewise.
18563 (do_jump_if_equal): Likewise.
18564 (emit_cmp_and_jump_insns): Likewise.
18565 (fix_phi_operands_for_edge): New function.
18566 (record_phi_operand_mapping): Likewise.
18567 (class pass_lower_switch): New pass.
18568 (pass_lower_switch::execute): New function.
18569 (make_pass_lower_switch): Likewise.
18570 (conditional_probability):
18571 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
18572 * tree-pass.h: Add make_pass_lower_switch.
18574 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
18577 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
18580 2017-08-29 Richard Biener <rguenther@suse.de>
18582 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
18583 we do not add a DW_AT_inline attribute twice.
18584 (gen_subprogram_die): Remove code setting DW_AT_inline on
18585 DECL_ABSTRACT_P nodes.
18587 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
18589 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
18590 calls to internal functions.
18591 (gimplify_modify_expr): Likewise.
18592 * tree-call-cdce.c (use_internal_fn): Likewise.
18593 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
18594 (convert_to_divmod): Set the nothrow flag.
18595 * tree-if-conv.c (predicate_mem_writes): Likewise.
18596 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
18597 (vectorizable_call): Likewise.
18598 (vectorizable_store): Likewise.
18599 (vectorizable_load): Likewise.
18600 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
18601 (vect_recog_mask_conversion_pattern): Likewise.
18603 2017-08-29 Martin Liska <mliska@suse.cz>
18606 * gcc.c (driver_handle_option): Add new argument.
18607 * opts-common.c (handle_option): Pass
18608 target_option_override_hook.
18609 * opts-global.c (lang_handle_option): Add new option.
18610 (set_default_handlers): Add new argument.
18611 (decode_options): Likewise.
18612 * opts.c (target_handle_option): Likewise.
18613 (common_handle_option): Call target_option_override_hook.
18614 * opts.h (struct cl_option_handler_func): Add hook for
18615 target option override.
18616 (struct cl_option_handlers): Likewise.
18617 (set_default_handlers): Add new argument.
18618 (decode_options): Likewise.
18619 (common_handle_option): Likewise.
18620 (target_handle_option): Likewise.
18621 * toplev.c (toplev::main): Pass targetm.target_option.override
18624 2017-08-29 Richard Biener <rguenther@suse.de>
18625 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
18627 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
18628 life to the active subtree.
18630 2017-08-28 Jeff Law <law@redhat.com>
18632 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
18633 derive_equivalences.
18634 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
18635 Code moved into....
18636 (edge_info::derive_equivalences): New private member function
18638 * tree-ssa-dom.c (class edge_info): Changed from a struct
18639 to a class. Add ctor/dtor, methods and data members.
18640 (edge_info::edge_info): Renamed from allocate_edge_info.
18641 Initialize additional members.
18642 (edge_info::~edge_info): New.
18643 (free_dom_edge_info): Delete the edge info.
18644 (record_edge_info): Use new class & associated member functions.
18645 Tighten forms for testing for edge equivalences.
18646 (record_temporary_equivalences): Iterate over the simple
18647 equivalences rather than assuming there's only one per edge.
18648 (cprop_into_successor_phis): Iterate over the simple
18649 equivalences rather than assuming there's only one per edge.
18650 (optimize_stmt): Use operand_equal_p rather than pointer
18651 equality for mini-DSE code.
18653 2017-08-28 Nathan Sidwell <nathan@acm.org>
18655 * gcc.c (execute): Fold SIGPIPE handling into switch
18656 statement. Adjust internal error message.
18658 2017-08-28 Richard Biener <rguenther@suse.de>
18661 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
18662 Do nothing for removed DIEs.
18664 2017-08-28 Richard Biener <rguenther@suse.de>
18666 PR tree-optimization/81977
18667 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
18670 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
18673 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
18674 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
18675 using targetm.gen_mem_thread_fence.
18677 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
18680 * config/i386/i386.md (*<btsc><mode>): Change operand 2
18681 predicate to register_operand. Reorder operands.
18682 (*btr<mode>): Ditto.
18683 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
18684 (*btr<mode>_mask): Ditto.
18686 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
18688 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
18689 * config/rs6000/xmmintrin.h: New file.
18690 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
18692 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18695 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
18696 parameter and_insn and return it.
18697 (recombine_lvx_pattern): Insert a copy to ensure availability of
18698 the base register of the copied masking operation at the point of
18699 the instruction replacement.
18700 (recombine_stvx_pattern): Likewise.
18702 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
18704 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
18705 undocumented switches.
18706 (-mpower9-dform-vector): Likewise.
18707 (-mpower9-dform): Likewise.
18708 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
18709 comments to delete references to -mpower9-dform* switches.
18710 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
18711 Delete reference to -mpower9-dform* switches, test for
18712 -mpower9-vector instead.
18713 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
18714 (OTHER_P9_VECTOR_MASKS): Likewise.
18715 (POWERPC_MASKS): Likewise.
18716 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
18717 tests against -mpower9-dform* to -mpower9-vector. Delete code
18718 that checked for -mpower9-dform* consistancy with other options.
18719 Add test for -mpower9-misc to enable other power9 switches.
18720 (rs6000_init_hard_regno_mode_ok): Likewise.
18721 (rs6000_option_override_internal): Likewise.
18722 (rs6000_emit_prologue): Likewise.
18723 (rs6000_emit_epilogue): Likewise.
18724 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
18725 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
18726 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
18728 (emit_fusion_p9_store): Likewise.
18729 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
18730 resetting these macros if the assembler does not support ISA 3.0
18732 (TARGET_P9_DFORM_VECTOR): Likewise.
18733 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
18734 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
18736 2017-08-25 Alan Modra <amodra@gmail.com>
18738 PR rtl-optimization/81747
18739 * cse.c (cse_extended_basic_block): Don't attempt to record
18740 equivalences for degenerate conditional branches that branch
18741 to their fall-through.
18743 2017-08-24 Martin Sebor <msebor@redhat.com>
18745 PR middle-end/81908
18746 * gimple-fold.c (size_must_be_zero_p): New function.
18747 (gimple_fold_builtin_memory_op): Call it.
18749 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
18751 * config/rs6000/mm_malloc.h: New file.
18753 2017-08-24 Bin Cheng <bin.cheng@arm.com>
18755 PR tree-optimization/81913
18756 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
18757 analysis when either IVs in condition can wrap.
18759 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
18761 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
18762 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
18764 2017-08-24 Richard Biener <rguenther@suse.de>
18767 * targhooks.c (default_target_can_inline_p): Properly
18768 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
18769 is present and always compare.
18770 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
18771 infer -mfpmath=sse from TARGET_SSE_P.
18772 (ix86_can_inline_p): Properly use target_option_default_node when
18773 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
18775 2017-08-24 Richard Biener <rguenther@suse.de>
18778 * dwarf2out.c (output_die): Handle flag_generate_offload like
18780 (output_comp_unit): Likewise.
18781 (gen_array_type_die): Likewise.
18782 (dwarf2out_early_finish): Likewise.
18783 (note_variable_value_in_expr): Likewise.
18784 (dwarf2out_finish): Likewise. Adjust assert.
18785 * cgraphunit.c (symbol_table::compile): Move setting of
18786 flag_generate_offload earlier ...
18787 (symbol_table::finalize_compilation_unit): ... here, before
18788 early debug finalization.
18790 2017-08-24 Richard Biener <rguenther@suse.de>
18792 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
18793 and ipa-fnsummary.h.
18794 (ix86_can_inline_p): When ix86_fpmath flags do not match
18795 check whether the callee uses FP math at all.
18797 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
18799 PR middle-end/81931
18800 * tree-ssanames.c (get_nonzero_bits): Use element_precision
18801 instead of TYPE_PRECISION.
18803 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
18804 Alan Hayward <alan.hayward@arm.com>
18805 David Sherwood <david.sherwood@arm.com>
18807 * combine.c (make_extraction): Use subreg_offset_from_lsb.
18809 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
18811 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
18813 (ix86_frame::stack_realign_allocate): New field.
18814 (struct machine_frame_state): Modify comments.
18815 (machine_frame_state::sp_realigned_fp_end): New field.
18816 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
18817 layout calculation.
18818 (sp_valid_at): Add assertion to assure no attempt to access invalid
18819 offset of a realigned stack.
18820 (fp_valid_at): Likewise.
18821 (choose_baseaddr): Modify comments.
18822 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
18823 ix86_expand_prologue.
18824 (ix86_expand_prologue): Modify stack realignment and allocation.
18825 (ix86_expand_epilogue): Modify comments.
18826 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
18827 avx2_runtime, avx512f, and avx512f_runtime.
18829 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
18831 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
18832 (mstackrealign): Do not init to -1.
18833 * config/i386/i386.c (ix86_option_override_internal):
18834 Check opts_set, not opts when setting default value of
18835 opts->x_ix86_force_align_arg_pointer.
18837 2017-08-23 Richard Biener <rguenther@suse.de>
18839 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
18840 lang_hooks.decl_printable_name.
18841 * print-rtl-function.c (print_rtx_function): Likewise.
18842 * tree-pretty-print.c (dump_function_header): Likewise.
18844 2017-08-23 Richard Biener <rguenther@suse.de>
18847 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
18848 -g0 at compile-time.
18850 2017-08-23 Tamar Christina <tamar.christina@arm.com>
18852 PR middle-end/19706
18853 * doc/sourcebuild.texi (Other hardware attributes):
18856 2017-08-23 Tamar Christina <tamar.christina@arm.com>
18858 PR middle-end/19706
18859 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
18860 Fix single-use check.
18862 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
18864 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
18866 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
18868 * doc/install.texi: Modify to add more details on running selected
18871 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
18873 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
18874 is combined with -mabi=ms.
18875 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
18878 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18880 PR tree-optimization/81488
18881 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
18882 and cached_basis fields.
18883 (MAX_SPREAD): New constant.
18884 (alloc_cand_and_find_basis): Initialize new fields.
18885 (clear_visited): New function.
18886 (create_phi_basis_1): Rename from create_phi_basis, set visited
18887 and cached_basis fields.
18888 (create_phi_basis): New wrapper function.
18889 (phi_add_costs_1): Rename from phi_add_costs, add spread
18890 parameter, set visited field, short-circuit when limits reached.
18891 (phi_add_costs): New wrapper function.
18892 (record_phi_increments_1): Rename from record_phi_increments, set
18894 (record_phi_increments): New wrapper function.
18895 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
18896 (phi_incr_cost): New wrapper function.
18897 (all_phi_incrs_profitable_1): Rename from
18898 all_phi_incrs_profitable, set visited field.
18899 (all_phi_incrs_profitable): New wrapper function.
18901 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
18902 Alan Hayward <alan.hayward@arm.com>
18903 David Sherwood <david.sherwood@arm.com>
18905 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
18906 that takes the outer and inner modes.
18907 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
18908 comparison as the canonical test for a paradoxical subreg.
18909 * combine.c (simplify_set): Use paradoxical_subreg_p.
18910 (make_extraction): Likewise.
18911 (force_to_mode): Likewise.
18912 (rtx_equal_for_field_assignment_p): Likewise.
18913 (gen_lowpart_for_combine): Likewise.
18914 (simplify_comparison): Likewise.
18915 * cse.c (equiv_constant): Likewise.
18916 * expmed.c (store_bit_field_1): Likewise.
18917 * final.c (alter_subreg): Likewise.
18918 * fwprop.c (propagate_rtx): Likewise.
18919 (forward_propagate_subreg): Likewise.
18920 * ira-conflicts.c (ira_build_conflicts): Likewise.
18921 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
18922 * lra-constraints.c (curr_insn_transform): Likewise.
18923 (split_reg): Likewise.
18924 * lra-eliminations.c (move_plus_up): Likewise.
18925 (lra_eliminate_regs_1): Likewise.
18926 * recog.c (general_operand): Likewise.
18927 * ree.c (combine_reaching_defs): Likewise.
18928 * reload.c (push_reload): Likewise.
18929 (find_reloads): Likewise.
18930 * reload1.c (elimination_effects): Likewise.
18931 (compute_reload_subreg_offset): Likewise.
18932 (choose_reload_regs): Likewise.
18933 * rtlanal.c (subreg_lsb_1): Likewise.
18934 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18935 (simplify_subreg): Likewise.
18936 * var-tracking.c (track_loc_p): Likewise.
18937 * emit-rtl.c (byte_lowpart_offset): Likewise.
18938 (paradoxical_subreg_p): Delete out-of-line definition.
18940 2017-08-22 Jeff Law <law@redhat.com>
18942 PR tree-optimization/81741
18943 PR tree-optimization/71947
18944 * tree-ssa-dom.c: Include tree-inline.h.
18945 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
18946 equivalences if one is more expensive to compute than the other.
18947 * tree-ssa-scopedtables.h (class const_or_copies): Make
18948 record_const_or_copy_raw method private.
18949 (class avail_exprs_stack): New method simplify_binary_operation.
18950 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
18951 avail_exprs_stack::simplify_binary_operation as needed.
18952 (avail_exprs_stack::simplify_binary_operation): New function.
18954 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
18956 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
18957 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
18958 (DOT_SYMBOLS): Likewise.
18959 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
18960 (RELOCATABLE_NEEDS_FIXUP): Likewise.
18961 (RS6000_ABI_NAME): Likewise.
18962 (TARGET_CMODEL): Likewise.
18963 (TOC_SECTION_ASM_OP): Likewise.
18964 (SET_CMODEL): New macro.
18965 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
18967 2017-08-22 Richard Biener <rguenther@suse.de>
18969 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
18970 to allow for free-lang-data replacements similar to verify_type_variant.
18972 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
18973 Alan Hayward <alan.hayward@arm.com>
18974 David Sherwood <david.sherwood@arm.com>
18976 * config/aarch64/aarch64.md (casesi): Use DImode rather than
18977 VOIDmode for the LABEL_REF.
18979 2017-08-22 Richard Biener <rguenther@suse.de>
18981 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
18983 2017-08-22 Richard Biener <rguenther@suse.de>
18985 * common.opt (feliminate-dwarf2-dups): Ignore.
18986 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
18987 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
18988 same_die_p_wrap, compute_section_prefix,
18989 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
18990 (comdat_symbol_id, comdat_symbol_number): Likewise.
18991 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
18993 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
18994 (output_die): Mark unreachable path unreachable.
18995 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
18996 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
18997 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
18998 (dwarf2out_early_finish): Likewise.
19000 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
19002 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
19004 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
19007 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
19008 not VAR_P. Filter attribute warnings with OPT_Wattributes.
19009 (avr_attribute_table) <io, io_low, address>: Initialize
19010 .decl_required with true.
19012 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
19014 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
19015 undocumented debugging options.
19016 (-mvsx-scalar-double): Likewise.
19017 (-mallow-df-permute): Likewise.
19018 (-mvectorize-builtins): Likewise.
19019 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
19020 (rs6000_builtin_vectorized_function): Likewise.
19021 (rs6000_builtin_md_vectorized_function): Likewise.
19022 (rs6000_opt_vars): Likewise.
19024 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
19027 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
19028 (*btrq_imm): Rename from *btrq.
19029 (*btcq_imm): Rename from *btcq.
19030 (btsc): New code attribute.
19031 (*<btsc><mode>): New insn pattern.
19032 (*btr<mode>): Ditto.
19033 (*<btsc><mode>_mask): New insn_and_split pattern.
19034 (*btr<mode>_mask): Ditto.
19036 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19037 Alan Hayward <alan.hayward@arm.com>
19038 David Sherwood <david.sherwood@arm.com>
19040 * function.c (pad_below): Simplify padding calculation.
19042 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19043 Alan Hayward <alan.hayward@arm.com>
19044 David Sherwood <david.sherwood@arm.com>
19046 * target.def (function_prologue): Remove frame size argument.
19047 (function_epilogue): Likewise.
19048 * doc/tm.texi: Regenerate.
19049 * final.c (final_start_function): Update call to function_prologue.
19050 (final_end_function): Update call to function_epilogue.
19051 (default_function_pro_epilogue): Remove frame size argument.
19052 * output.h (default_function_pro_epilogue): Likewise.
19053 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
19054 (arm_output_function_prologue): Likewise.
19055 * config/frv/frv.c (frv_function_prologue): Likewise.
19056 (frv_function_epilogue): Likewise.
19057 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
19058 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
19059 (ia64_output_function_epilogue): Likewise.
19060 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
19061 (m32r_output_function_epilogue): Likewise.
19062 * config/microblaze/microblaze.c (microblaze_function_prologue)
19063 (microblaze_function_epilogue): Likewise.
19064 * config/mips/mips.c (mips_output_function_prologue): Likewise.
19065 (mips_output_function_epilogue): Likewise.
19066 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
19067 (mmix_target_asm_function_epilogue): Likewise.
19068 * config/msp430/msp430.c (msp430_start_function): Likewise.
19069 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
19070 (nds32_asm_function_epilogue): Likewise.
19071 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
19072 * config/pa/pa.c (pa_output_function_prologue): Likewise.
19073 (pa_output_function_epilogue): Likewise.
19074 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
19075 (rs6000_output_function_epilogue): Likewise.
19076 * config/rl78/rl78.c (rl78_start_function): Likewise.
19077 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
19078 (rs6000_output_function_epilogue): Likewise.
19079 * config/rx/rx.c (rx_output_function_prologue): Likewise.
19080 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
19081 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
19082 (sparc_asm_function_epilogue): Likewise.
19084 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19086 * tree.h (type_has_mode_precision_p): New function.
19087 * convert.c (convert_to_integer_1): Use it.
19088 * expr.c (expand_expr_real_2): Likewise.
19089 (expand_expr_real_1): Likewise.
19090 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
19091 * match.pd: Likewise.
19092 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
19093 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
19094 * tree-tailcall.c (process_assignment): Likewise.
19095 * tree-vect-loop.c (vectorizable_reduction): Likewise.
19096 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
19097 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
19098 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
19099 (vectorizable_assignment): Likewise.
19100 (vectorizable_shift): Likewise.
19101 (vectorizable_operation): Likewise.
19102 * tree-vrp.c (register_edge_assert_for_2): Likewise.
19104 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
19106 * match.pd: Add pow (C, x) simplification.
19108 2017-08-21 Richard Biener <rguenther@suse.de>
19110 PR tree-optimization/81900
19111 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
19112 for blocks with abnormal predecessors.
19113 (compute_antic): Do not set visited flag prematurely.
19115 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
19118 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
19120 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19122 * stor-layout.h (vector_type_mode): Move to...
19123 * tree.h (vector_type_mode): ...here.
19124 * stor-layout.c (vector_type_mode): Move to...
19125 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
19127 2017-08-21 Richard Biener <rguenther@suse.de>
19129 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
19130 register_external_die hooks.
19131 (debug_false_tree_charstarstar_uhwistar): Declare.
19132 (debug_nothing_tree_charstar_uhwi): Likewise.
19133 * debug.c (do_nothing_debug_hooks): Adjust.
19134 (debug_false_tree_charstarstar_uhwistar): New do nothing.
19135 (debug_nothing_tree_charstar_uhwi): Likewise.
19136 * dbxout.c (dbx_debug_hooks): Adjust.
19137 (xcoff_debug_hooks): Likewise.
19138 * sdbout.c (sdb_debug_hooks): Likewise.
19139 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
19140 * dwarf2out.c (macinfo_label_base): New global.
19141 (dwarf2out_register_external_die): New function for the
19142 register_external_die hook.
19143 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
19144 (dwarf2_debug_hooks): Use them.
19145 (dwarf2_lineno_debug_hooks): Adjust.
19146 (struct die_struct): Add with_offset flag.
19147 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
19148 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
19149 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
19150 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
19151 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
19152 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
19153 defining section names for the early LTO debug variants.
19154 (reset_indirect_string): New helper.
19155 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
19156 (print_dw_val): Add support for offsetted symbol references.
19157 (get_ultimate_context): Split out from is_cxx.
19158 (is_cxx): Use get_ultimate_context.
19159 (is_fortran): Add decl overload.
19160 (compute_comp_unit_symbol): Split out worker from
19161 compute_section_prefix.
19162 (compute_section_prefix): Call compute_comp_unit_symbol and
19163 set comdat_type_p here.
19164 (output_die): Skip DIE symbol output for the LTO added one.
19165 Handle DIE symbol references with offset.
19166 (output_comp_unit): Guard section name mangling properly.
19167 For LTO debug sections emit a symbol at the section beginning
19168 which we use to refer to its DIEs.
19169 (add_abstract_origin_attribute): For DIEs registered via
19170 dwarf2out_register_external_die directly refer to the early
19171 DIE rather than indirectly through the shadow one we created.
19172 Remove obsolete call to dwarf2out_abstract_function for
19173 non-function/block origins.
19174 (gen_array_type_die): When generating early LTO debug do
19175 not emit DW_AT_string_length.
19176 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
19177 late when in LTO. As suggested place a gcc_unreachable for
19178 the DECL_ABSTRACT_P case.
19179 (gen_subprogram_die): Avoid another specification DIE
19180 for early built declarations/definitions for the late LTO case.
19181 (gen_variable_die): Add type references for late duplicated VLA dies
19183 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
19184 we have the abstract instance already.
19185 (process_scope_var): Adjust decl DIE contexts in LTO which
19186 first puts them in limbo.
19187 (gen_decl_die): Do not generate type DIEs late apart from
19188 types for VLAs or for decls we do not yet have a DIE. Do not
19189 call dwarf2out_abstract_function late.
19190 (dwarf2out_early_global_decl): Make sure to create DIEs
19191 for abstract instances of a decl first.
19192 (dwarf2out_late_global_decl): Adjust comment.
19193 (output_macinfo_op): With multiple macro sections use
19194 macinfo_label_base to distinguish labels.
19195 (output_macinfo): Likewise. Update macinfo_label_base.
19196 Pass in the line info label.
19197 (note_variable_value_in_expr): When generating LTO resolve
19198 all variable values here by generating DIEs as needed.
19199 (init_sections_and_labels): Add early LTO debug flag parameter
19200 and generate different sections and names if set. Add generation
19201 counter for the labels so we can have multiple of them.
19202 (reset_dies): Helper to allow DIEs to be output multiple times.
19203 (dwarf2out_finish): When outputting DIEs to the fat part of an
19204 LTO object first reset DIEs.
19205 (dwarf2out_early_finish): Output early DIEs when generating LTO.
19206 (modified_type_die): Check for decl_ultimate_origin being self
19208 (gen_type_die_with_usage): Likewise.
19209 (gen_typedef_die): Allow decl_ultimate_origin being self.
19210 (set_decl_abstract_flags): Remove.
19211 (set_block_abstract_flags): Likewise.
19212 (dwarf2out_abstract_function): Treat the early generated DIEs
19213 as the abstract copy and only add DW_AT_inline and
19214 DW_AT_artificial here and call set_decl_origin_self.
19215 If the DIE has an abstract origin don't do anything.
19216 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
19217 if we have none yet (Go fails to build one, PR78628).
19218 (variably_modified_type_p): Prevent endless recursion for Ada
19219 cyclic pointer types.
19220 * lto-streamer-in.c: Include debug.h.
19221 (dref_queue): New global.
19222 (lto_read_tree_1): Stream in DIE references.
19223 (lto_input_tree): Register DIE references.
19224 (input_function): Stream DECL_DEBUG_ARGS.
19225 * lto-streamer-out.c: Include debug.h.
19226 (lto_write_tree_1): Output DIE references.
19227 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
19228 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
19229 (output_function): Stream DECL_DEBUG_ARGS.
19230 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
19231 Stream DECL_ABSTRACT_ORIGIN.
19232 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
19233 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
19234 DECL_CONTEXT for file-scope decls.
19235 * lto-streamer.h (struct dref_entry): Declare.
19236 (dref_queue): Likewise.
19237 * cfgexpand.c (pass_expand::execute): Do not call the
19238 outlining_inline_function hook here.
19239 * lto-wrapper.c (debug_obj): New global.
19240 (tool_cleanup): Unlink it if required.
19241 (debug_objcopy): New function.
19242 (run_gcc): Handle early debug sections in the IL files by
19243 extracting them to separate files, partially linkin them and
19244 feeding the result back as result to the linker.
19245 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
19246 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
19247 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
19248 sections into a separate segment.
19249 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
19251 (darwin_asm_dwarf_section): Likewise.
19252 (darwin_asm_output_dwarf_offset): Likewise.
19253 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
19255 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19256 Alan Hayward <alan.hayward@arm.com>
19257 David Sherwood <david.sherwood@arm.com>
19259 * read-md.h (md_reader::record_potential_iterator_use): Replace
19260 pointer argument with an rtx and an index.
19261 * read-rtl.c (iterator_group::apply_iterator): Likewise.
19262 (apply_mode_iterator): Likewise.
19263 (apply_code_iterator): Likewise.
19264 (apply_int_iterator): Likewise.
19265 (apply_subst_iterator): Likewise.
19266 (record_iterator_use): Likewise.
19267 (record_attribute_use): Likewise.
19268 (md_reader::record_potential_iterator_use): Likewise. Update calls
19269 to record_iterator_use and apply_iterator.
19270 (iterator_use): Replace ptr with x and index.
19271 (attribute_use): Likewise.
19272 (apply_attribute_uses): Update calls to apply_iterator.
19273 (apply_iterators): Likewise. Update initialization of iterator_use.
19274 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
19275 and record_potential_iterator_use.
19276 (rtx_reader::read_rtx_operand): Likewise.
19278 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19279 Alan Hayward <alan.hayward@arm.com>
19280 David Sherwood <david.sherwood@arm.com>
19282 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
19285 2017-08-21 Richard Biener <rguenther@suse.de>
19287 PR middle-end/81884
19288 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
19289 at struct end conservatively when comparing common bases.
19291 2017-08-21 Richard Biener <rguenther@suse.de>
19293 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
19294 (mem_ref_in_stmt): Remove.
19295 (determine_max_movement): Use ref index to get at the reference.
19296 (invariantness_dom_walker::before_dom_children): Deal with
19297 lim data already initialized.
19298 (gather_mem_refs_stmt): Initialize lim data and record ref index.
19300 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
19302 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
19303 (TARGET_ISA_ROUND): Ditto.
19304 (TARGET_ROUND): Ditto.
19305 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
19306 * config/i386/i386.md: Ditto.
19307 * config/i386/sse.md: Ditto.
19308 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
19309 with OPTION_MASK_ISA_SSE4_1.
19311 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
19314 * doc/extend.texi (x86 Built-in Functions): Correct the name of
19315 __builtin_ia32_lzcnt_u16.
19317 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
19320 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
19321 (rs6000_set_current_function): Rewrite function to use it.
19323 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
19326 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
19327 and TYPE_WARN_IF_NOT_ALIGN.
19328 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
19329 (handle_warn_if_not_align): New.
19330 (place_union_field): Call handle_warn_if_not_align.
19331 (place_field): Call handle_warn_if_not_align.
19332 Copy TYPE_WARN_IF_NOT_ALIGN.
19333 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
19334 (layout_type): Likewise.
19335 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
19337 (tree_decl_common): Add warn_if_not_align.
19338 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
19339 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
19340 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
19341 (DECL_WARN_IF_NOT_ALIGN): Likewise.
19342 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
19343 * doc/extend.texi: Document warn_if_not_aligned attribute.
19344 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
19346 2017-08-17 Martin Liska <mliska@suse.cz>
19349 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
19350 (get_data_dependence): Use it as pointer type.
19351 (distribute_loop): Likewise.
19353 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19355 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
19356 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
19357 (p8_vmrgow_<mode>_direct): New define_insn.
19358 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
19359 handle endianness for vmrgew and vmrgow permute patterns.
19361 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
19363 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
19364 * config/rs6000/rs6000-cpus.def: Remove comment.
19365 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
19366 (POWERPC_MASKS): Likewise.
19367 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
19368 use of TARGET_VSX_TIMODE.
19369 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
19370 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
19371 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
19372 (rs6000_option_override_internal): Remove dead code.
19373 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
19374 (rs6000_legitimize_reload_address): Likewise.
19375 (rs6000_legitimate_address_p): Likewise.
19376 (rs6000_opt_masks): Delete "vsx-timode".
19377 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
19378 from function comment.
19379 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
19380 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
19381 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
19383 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
19384 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
19385 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
19386 (VSX_TI): Likewise.
19388 (define_peephole2): Likewise.
19390 2017-08-17 Martin Sebor <msebor@redhat.com>
19393 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
19394 past the end of an array.
19395 (test_pp_format): Add test cases.
19397 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
19399 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
19400 missing ECF_NOTHROW flags.
19402 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
19405 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
19406 operands residing in integer registers.
19407 (*vsx_le_perm_load_<mode>): Likewise.
19408 (*vsx_le_perm_store_<mode>): Likewise.
19409 (define_peephole2): Add peepholes to optimize the above.
19411 2017-08-17 Marek Polacek <polacek@redhat.com>
19413 PR middle-end/81814
19414 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
19415 to mimic what shorten_compare did. Change the return type to bool.
19416 (fold_cond_expr_with_comparison): Update call to
19417 operand_equal_for_comparison_p.
19418 (fold_ternary_loc): Likewise.
19420 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
19422 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
19424 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
19426 2017-08-17 Richard Biener <rguenther@suse.de>
19428 * tree-ssa-structalias.c (solve_graph): When propagating
19429 to successors update the graphs succ edges and avoid duplicate work.
19431 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
19434 * config/i386/i386.c (ix86_option_override_internal): Save target
19435 specific options after ix86_stack_protector_guard_reg was changed.
19437 2017-08-17 Richard Biener <rguenther@suse.de>
19439 PR tree-optimization/81827
19440 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
19441 (new_var_info): Initialize it conservatively.
19442 (get_call_vi): Mark register vars.
19443 (new_scalar_tmp_constraint_exp): Likewise.
19444 (handle_rhs_call): Likewise.
19445 (handle_const_call): Likewise.
19446 (create_function_info_for): Likewise.
19447 (solve_constraints): Sort varinfos to separate register from
19448 non-register vars to pack points-to solution bitmaps during
19451 2017-08-17 Marek Polacek <polacek@redhat.com>
19453 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
19455 2017-08-17 Richard Biener <rguenther@suse.de>
19457 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
19458 to true when overflow is undefined and we saturated the result.
19460 2017-08-17 Alan Modra <amodra@gmail.com>
19463 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
19464 Don't use store multiple if only one reg needs saving.
19465 (interesting_frame_related_regno): New function.
19466 (rs6000_frame_related): Don't emit frame info for regs that
19468 (rs6000_emit_epilogue): Likewise.
19470 2017-08-16 Nathan Sidwell <nathan@acm.org>
19472 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
19473 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
19474 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
19475 * tree.c (free_lang_data_in_type): Use else-if chain. Always
19476 clear TYPE_LANG_1. Remove obsolete member-function stripping.
19477 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
19478 (verify_type): Adjust for TYPE_BINFO move.
19479 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
19480 process TYPE_BINFO directly.
19481 (hash_tree): Likewise.
19482 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
19484 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
19487 2017-08-16 David Malcolm <dmalcolm@redhat.com>
19489 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
19491 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
19494 * config/i386/i386.md (*anddi_1_btr): Change predicates of
19495 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
19496 Add ix86_binary_operator_ok to insn constraint.
19497 (*iordi_1_bts): Ditto.
19498 (*xordi_1_btc): Ditto.
19499 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
19500 Update corresponding peephole2 pattern.
19504 2017-08-16 Bin Cheng <bin.cheng@arm.com>
19506 PR tree-optimization/81832
19507 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
19508 copy loop header which has IFN_LOOP_DIST_ALIAS call.
19510 2017-08-16 Marek Polacek <polacek@redhat.com>
19513 * fold-const.c (fold_indirect_ref_1): Restore original behavior
19514 regarding size_zero_node.
19516 2017-08-16 Martin Liska <mliska@suse.cz>
19519 * config.gcc: Respect previously set extra_objs in case
19522 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
19524 PR tree-optimization/81835
19525 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
19526 the phi SSA_NAME. Check that the condition in a COND_EXPR does
19527 not depend on the phi.
19529 2017-08-16 Alan Modra <amodra@gmail.com>
19531 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
19534 2017-08-16 Alan Modra <amodra@gmail.com>
19536 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
19537 (save_reg_p): ..into this. Update all callers.
19538 (first_reg_to_save): Simplify.
19540 2017-08-16 Alan Modra <amodra@gmail.com>
19542 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
19545 2017-08-15 Joseph Myers <joseph@codesourcery.com>
19549 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
19550 constant count if that count is less than 32.
19552 2017-08-15 Nathan Sidwell <nathan@acm.org>
19554 * gcc.c (execute): Emit friendlier message if inferior is killed
19555 by an external cause.
19557 2017-08-15 Richard Biener <rguenther@suse.de>
19559 PR tree-optimization/81790
19560 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
19561 CONSTRUCTORs from simplifying and VN.
19563 2017-08-14 Martin Sebor <msebor@redhat.com>
19565 * builtin-attrs.def: Add comments.
19567 2017-08-14 Martin Sebor <msebor@redhat.com>
19570 * doc/extend.texi (attribute nonstring): Document new attribute.
19572 2017-08-14 Martin Sebor <msebor@redhat.com>
19575 * tree-diagnostic.c (default_tree_printer): Handle %G.
19576 * gimple-pretty-print.h (percent_G_format): Declare new function.
19577 * gimple-pretty-print.c (percent_G_format): Define.
19578 * tree-pretty-print.c (percent_K_format): Add argument.
19580 2017-08-14 Martin Sebor <msebor@redhat.com>
19582 PR translation/79998
19583 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
19584 Remove a stray space.
19586 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
19589 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
19590 (*iordi_1_bts): Ditto.
19591 (*xordi_1_btc): Ditto.
19593 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19596 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
19597 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19599 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
19600 quoted strings, and make more translator-friendly.
19601 (darwin_rs6000_override_options): Likewise.
19602 (rs6000_option_override_internal): Likewise.
19603 (rs6000_return_in_memory): Fix overlong line.
19604 (init_cmulative_args): Use quoted strings, and make more
19605 translator-friendly.
19606 (rs6000_pass_by_reference): Fix overlong line.
19607 (def_builtin): Use quoted strings.
19608 (altivec_expand_predicate_builtin): Use quoted strings, and make
19609 more translator-friendly.
19610 (htm_expand_builtin): Use quoted strings.
19611 (cpu_expand_builtin): Use quoted strings, and make more
19612 translator-friendly.
19613 (altivec_expand_builtin): Likewise.
19614 (paired_expand_predicate_builtin): Likewise.
19615 (rs6000_invalid_builtin): Likewise.
19616 (builtin_function_type): Use quoted strings.
19617 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
19618 more translator-friendly.
19619 (rs6000_trampoline_init): Likewise.
19620 (rs6000_handle_altivec_attribute): Likewise.
19621 (rs6000_inner_target_options): Use quoted strings.
19622 (rs6000_disable_incompatible_switches): Likewise.
19623 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
19624 strings, and make more translator-friendly.
19625 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
19627 2017-08-14 Bin Cheng <bin.cheng@arm.com>
19629 PR tree-optimization/81799
19630 * tree-loop-distribution.c (version_loop_by_alias_check): Force
19631 cond_expr to simple gimple operand.
19633 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
19635 PR middle-end/46932
19636 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
19638 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
19642 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
19643 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
19644 TARGET_GASISR_PROLOGUES.
19645 * config/avr/avr.c (avr_option_override): Same.
19646 (avr_pass_pre_proep::execute): Same.
19648 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
19651 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
19652 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
19654 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
19656 * config/i386/i386.md (*load_tp_<mode>): Redefine as
19657 define_insn_and_split. Split to a memory load from 0 in
19658 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
19659 using PTR mode iterator.
19660 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
19661 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
19662 (*add_tp_<mode>): Redefine as define_insn_and_split.
19663 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
19664 address space. Merge with *add_tp_x32 using PTR mode iterator.
19665 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
19666 Split to an add with a memory load from 0 in
19667 DEFAULT_TLS_SEG_REG address space.
19669 2017-08-12 Andrew Pinski <apinski@cavium.com>
19671 * config/aarch64/aarch64-option-extensions.def (rdma):
19672 Fix feature string to what Linux prints out in /proc/cpuinfo.
19674 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
19677 * dwarf2out.c (modified_type_die): For C typedef types that have
19678 an ultimate origin, process the ultimate origin instead of the
19680 (gen_typedef_die): Assert that input DECLs have no ultimate
19682 (gen_type_die_with_usage): For typedef variants that have an
19683 ultimate origin, just call gen_decl_die on the original DECL.
19684 (process_scope_var): Avoid creating DIEs for local typedefs and
19685 concrete static variables.
19687 2017-08-12 Alan Modra <amodra@gmail.com>
19691 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
19692 match gnu-user.h startfile.
19693 (ENDFILE_LINUX_SPEC): Similarly.
19695 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
19698 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
19700 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
19702 2017-08-11 Tamar Christina <tamar.christina@arm.com>
19703 * config/aarch64/aarch64.md (mov<mode>): Change.
19704 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
19705 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
19706 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
19708 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
19710 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
19711 for storage order barriers.
19713 2017-08-11 Martin Liska <mliska@suse.cz>
19716 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
19717 variables of void type.
19719 2017-08-11 Martin Liska <mliska@suse.cz>
19721 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
19722 TARGET_SUPPORTS_ALIASES.
19723 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
19724 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
19725 (optimize_weakref): Likewise.
19726 * symtab.c (symtab_node::noninterposable_alias): Likewise.
19727 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
19728 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
19730 2017-08-11 Martin Liska <mliska@suse.cz>
19733 * config/i386/i386.c (make_resolver_func): Do complete
19734 refactoring of the function.
19736 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
19739 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
19740 * config/i386/i386.c (ix86_stack_protect_guard): Use
19741 ix86_stack_protect_guard_symbol_str to generate varible declaration.
19742 * doc/invoke.texi (x86 Options): Document
19743 -mstack-protector-guard-symbol= option.
19745 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
19747 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
19748 * config/i386/i386.c (ix86_split_stack_guard): New function.
19749 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
19750 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
19751 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
19752 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
19753 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
19754 (split_stack_space_check): Call ix86_split_stack_guard.
19756 2017-08-10 Martin Sebor <msebor@redhat.com>
19758 * print-tree.c (print_node): Print location using the established
19760 Replace spaces with colons.
19761 (debug_raw, debug): Ditto.
19763 2017-08-10 Martin Sebor <msebor@redhat.com>
19766 * pretty-print.c (pp_format): Correct the handling of %s precision.
19768 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
19771 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
19773 (ix86_finalize_stack_frame_flags): This. Also clear
19774 frame_pointer_needed if -fno-omit-frame-pointer is used without
19776 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
19777 with ix86_finalize_stack_frame_flags.
19778 (ix86_expand_epilogue): Likewise.
19779 (ix86_expand_split_stack_prologue): Likewise.
19780 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
19782 2017-08-10 Martin Liska <mliska@suse.cz>
19785 * c-attribs.c (handle_target_attribute):
19786 Report warning for an empty string argument of target attribute.
19788 2017-08-09 Jakub Jelinek <jakub@redhat.com>
19791 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
19793 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
19794 or DECL_NONLOCAL labels.
19795 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
19796 or DECL_NONLOCAL labels here.
19798 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
19800 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
19801 to indicate when early gimple folding has been disabled.
19802 (rs6000_gimple_fold_builtin): Add debug content.
19803 (rs6000_invalid_builtin): Fix whitespace.
19804 (rs6000_expand_builtin): Fix whitespace.
19805 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
19807 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
19810 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
19811 SAVE_MULTIPLE if not all the registers that saves, should be saved.
19813 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
19815 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
19816 (qdf24xx): Likewise.
19817 * config/aarch64/aarch64.md: Include falkor.md.
19818 * config/aarch64/falkor.md: New.
19820 2017-08-09 Marek Polacek <polacek@redhat.com>
19823 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
19824 * diagnostic.c (emit_diagnostic): Add a comment.
19825 (emit_diagnostic_valist): New function.
19827 2017-08-09 Marek Polacek <polacek@redhat.com>
19830 * input.c (make_location): New overload.
19831 * input.h (make_location): Declare.
19833 2017-08-08 Alan Modra <amodra@gmail.com>
19834 H.J. Lu <hongjiu.lu@intel.com>
19837 * gcc.c (NO_PIE_SPEC): Delete.
19838 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
19840 (LINK_PIE_SPEC): ..to here.
19841 (LINK_COMMAND_SPEC): Support -no-pie.
19842 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
19843 chain of crtbegin*.o selection, update for PIE_SPEC changes and
19845 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
19846 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
19847 (ENDFILE_CRTEND_SPEC): Similarly.
19849 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
19852 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
19853 (mstack-protector-guard-offset=): Ditto.
19854 * config/i386/i386.c (ix86_option_override): Handle
19855 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
19857 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
19858 ix86_stack_protect_guard_offset variables.
19859 (TARGET_STACK_PROTECT_GUARD): Always define.
19860 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
19861 and -mstack-protector-guard-offset= options.
19863 2017-08-08 Bin Cheng <bin.cheng@arm.com>
19865 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
19866 boundary case for the last candidate.
19868 2017-08-08 Bin Cheng <bin.cheng@arm.com>
19870 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
19871 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
19873 2017-08-08 Tamar Christina <tamar.christina@arm.com>
19875 PR middle-end/19706
19876 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
19877 * config/aarch64/aarch64-builtins.c
19878 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
19879 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
19880 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
19882 2017-08-08 Tamar Christina <tamar.christina@arm.com>
19883 Andrew Pinski <pinskia@gmail.com>
19885 PR middle-end/19706
19886 * internal-fn.def (XORSIGN): New.
19887 * optabs.def (xorsign_optab): New.
19888 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
19889 (convert_expand_mult_copysign): New.
19890 (pass_optimize_widening_mul::execute): Call
19891 convert_expand_mult_copysign.
19893 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19895 PR tree-optimization/81354
19896 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
19897 Insert on edges rather than explicitly creating landing pads.
19898 (analyze_candidates_and_replace): Commit edge inserts.
19900 2017-08-08 Richard Biener <rguenther@suse.de>
19902 PR middle-end/81719
19903 * tree-ssa-loop-niter.c: Include tree-dfa.h.
19904 (expand_simple_operations): Also look through ADDR_EXPRs with
19905 MEM_REF bases treating them as POINTER_PLUS_EXPR.
19907 2017-08-08 Richard Biener <rguenther@suse.de>
19909 PR tree-optimization/81723
19910 * tree-vect-slp.c (struct bst_traits): New hash traits.
19911 (bst_fail): New global.
19912 (vect_build_slp_tree_2): New worker, split out from ...
19913 (vect_build_slp_tree): ... this now wrapping it with using
19914 bst_fail set to cache SLP tree build fails. Properly handle
19916 (vect_analyze_slp_instance): Allocate and free bst_fail.
19918 2017-08-08 Martin Liska <mliska@suse.cz>
19921 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
19922 LABEL_DECLs that can be from a different function.
19924 2017-08-08 Bin Cheng <bin.cheng@arm.com>
19926 PR tree-optimization/81744
19927 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
19928 loop's number of iterations.
19930 2017-08-08 Martin Liska <mliska@suse.cz>
19932 * asan.c: Include header files.
19933 * attribs.c (build_decl_attribute_variant): New function moved
19935 (build_type_attribute_qual_variant): Likewise.
19936 (cmp_attrib_identifiers): Likewise.
19937 (simple_cst_list_equal): Likewise.
19938 (omp_declare_simd_clauses_equal): Likewise.
19939 (attribute_value_equal): Likewise.
19940 (comp_type_attributes): Likewise.
19941 (build_type_attribute_variant): Likewise.
19942 (lookup_ident_attribute): Likewise.
19943 (remove_attribute): Likewise.
19944 (merge_attributes): Likewise.
19945 (merge_type_attributes): Likewise.
19946 (merge_decl_attributes): Likewise.
19947 (merge_dllimport_decl_attributes): Likewise.
19948 (handle_dll_attribute): Likewise.
19949 (attribute_list_equal): Likewise.
19950 (attribute_list_contained): Likewise.
19951 * attribs.h (lookup_attribute): New function moved from tree.[ch].
19952 (lookup_attribute_by_prefix): Likewise.
19953 * bb-reorder.c: Include header files.
19954 * builtins.c: Likewise.
19955 * calls.c: Likewise.
19956 * cfgexpand.c: Likewise.
19957 * cgraph.c: Likewise.
19958 * cgraphunit.c: Likewise.
19959 * convert.c: Likewise.
19960 * dwarf2out.c: Likewise.
19961 * final.c: Likewise.
19962 * fold-const.c: Likewise.
19963 * function.c: Likewise.
19964 * gimple-expr.c: Likewise.
19965 * gimple-fold.c: Likewise.
19966 * gimple-pretty-print.c: Likewise.
19967 * gimple.c: Likewise.
19968 * gimplify.c: Likewise.
19969 * hsa-common.c: Likewise.
19970 * hsa-gen.c: Likewise.
19971 * internal-fn.c: Likewise.
19972 * ipa-chkp.c: Likewise.
19973 * ipa-cp.c: Likewise.
19974 * ipa-devirt.c: Likewise.
19975 * ipa-fnsummary.c: Likewise.
19976 * ipa-inline.c: Likewise.
19977 * ipa-visibility.c: Likewise.
19979 * lto-cgraph.c: Likewise.
19980 * omp-expand.c: Likewise.
19981 * omp-general.c: Likewise.
19982 * omp-low.c: Likewise.
19983 * omp-offload.c: Likewise.
19984 * omp-simd-clone.c: Likewise.
19985 * opts-global.c: Likewise.
19986 * passes.c: Likewise.
19987 * predict.c: Likewise.
19988 * sancov.c: Likewise.
19989 * sanopt.c: Likewise.
19990 * symtab.c: Likewise.
19991 * toplev.c: Likewise.
19992 * trans-mem.c: Likewise.
19993 * tree-chkp.c: Likewise.
19994 * tree-eh.c: Likewise.
19995 * tree-into-ssa.c: Likewise.
19996 * tree-object-size.c: Likewise.
19997 * tree-parloops.c: Likewise.
19998 * tree-profile.c: Likewise.
19999 * tree-ssa-ccp.c: Likewise.
20000 * tree-ssa-live.c: Likewise.
20001 * tree-ssa-loop.c: Likewise.
20002 * tree-ssa-sccvn.c: Likewise.
20003 * tree-ssa-structalias.c: Likewise.
20004 * tree-ssa.c: Likewise.
20005 * tree-streamer-in.c: Likewise.
20006 * tree-vectorizer.c: Likewise.
20007 * tree-vrp.c: Likewise.
20008 * tsan.c: Likewise.
20009 * ubsan.c: Likewise.
20010 * varasm.c: Likewise.
20011 * varpool.c: Likewise.
20012 * tree.c: Remove functions moved to attribs.[ch].
20013 * tree.h: Likewise.
20014 * config/aarch64/aarch64.c: Add attrs.h header file.
20015 * config/alpha/alpha.c: Likewise.
20016 * config/arc/arc.c: Likewise.
20017 * config/arm/arm.c: Likewise.
20018 * config/avr/avr.c: Likewise.
20019 * config/bfin/bfin.c: Likewise.
20020 * config/c6x/c6x.c: Likewise.
20021 * config/cr16/cr16.c: Likewise.
20022 * config/cris/cris.c: Likewise.
20023 * config/darwin.c: Likewise.
20024 * config/epiphany/epiphany.c: Likewise.
20025 * config/fr30/fr30.c: Likewise.
20026 * config/frv/frv.c: Likewise.
20027 * config/ft32/ft32.c: Likewise.
20028 * config/h8300/h8300.c: Likewise.
20029 * config/i386/winnt.c: Likewise.
20030 * config/ia64/ia64.c: Likewise.
20031 * config/iq2000/iq2000.c: Likewise.
20032 * config/lm32/lm32.c: Likewise.
20033 * config/m32c/m32c.c: Likewise.
20034 * config/m32r/m32r.c: Likewise.
20035 * config/m68k/m68k.c: Likewise.
20036 * config/mcore/mcore.c: Likewise.
20037 * config/microblaze/microblaze.c: Likewise.
20038 * config/mips/mips.c: Likewise.
20039 * config/mmix/mmix.c: Likewise.
20040 * config/mn10300/mn10300.c: Likewise.
20041 * config/moxie/moxie.c: Likewise.
20042 * config/msp430/msp430.c: Likewise.
20043 * config/nds32/nds32-isr.c: Likewise.
20044 * config/nds32/nds32.c: Likewise.
20045 * config/nios2/nios2.c: Likewise.
20046 * config/nvptx/nvptx.c: Likewise.
20047 * config/pa/pa.c: Likewise.
20048 * config/pdp11/pdp11.c: Likewise.
20049 * config/powerpcspe/powerpcspe.c: Likewise.
20050 * config/riscv/riscv.c: Likewise.
20051 * config/rl78/rl78.c: Likewise.
20052 * config/rx/rx.c: Likewise.
20053 * config/s390/s390.c: Likewise.
20054 * config/sh/sh.c: Likewise.
20055 * config/sol2.c: Likewise.
20056 * config/sparc/sparc.c: Likewise.
20057 * config/spu/spu.c: Likewise.
20058 * config/stormy16/stormy16.c: Likewise.
20059 * config/tilegx/tilegx.c: Likewise.
20060 * config/tilepro/tilepro.c: Likewise.
20061 * config/v850/v850.c: Likewise.
20062 * config/vax/vax.c: Likewise.
20063 * config/visium/visium.c: Likewise.
20064 * config/xtensa/xtensa.c: Likewise.
20066 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
20069 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
20070 constraints since the -mupper-regs-* switches have been
20072 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
20073 into a vector from a double word element that was extracted from
20074 another vector, and eliminate extra XXPERMDI instructions.
20075 (vsx_concat_<mode>_2): Likewise.
20076 (vsx_concat_<mode>_3): Likewise.
20077 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
20078 concat to allow optimizing inserts from previous extracts.
20080 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
20082 * config/i386/i386.c (ix86_stack_protect_guard): Generate
20083 memory reference to a SSP offset in TLS address space.
20084 (ix86_print_operand) <case '@'>: Remove.
20085 (ix86_print_operand_punct_valid_p): Remove '@' code.
20086 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
20087 UNSPEC_SP_TLS_TEST.
20088 (stack_tls_protect_set_<mode>): Remove.
20089 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
20090 (stack_tls_protect_test_<mode>): Remove.
20091 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
20093 2017-08-07 Olivier Hainque <hainque@adacore.com>
20096 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
20098 2017-08-07 Douglas Rupp <rupp@adacore.com>
20100 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
20101 variable was referenced as multidir in command.
20103 2017-08-07 Jakub Jelinek <jakub@redhat.com>
20106 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
20109 2017-08-07 Martin Liska <mliska@suse.cz>
20111 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
20112 * config/rl78/rl78.c: Add include of attribs.h.
20113 * config/sh/sh.c: Likewise.
20114 * config/v850/v850.c: Likewise.
20116 2017-08-07 Tom de Vries <tom@codesourcery.com>
20118 PR middle-end/78266
20119 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
20121 2017-08-07 Martin Liska <mliska@suse.cz>
20123 * config/mips/mips.c: Include attribs.h.
20125 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
20128 * doc/invoke.texi: Document change in behvaior for -Ofast for
20131 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
20133 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
20135 (aarch64_pop_regs): Likewise.
20136 (aarch64_gen_load_pair): Likewise.
20137 (aarch64_save_callee_saves): Likewise.
20138 (aarch64_restore_callee_saves): Likewise.
20140 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
20142 * config/i386/i386.c: Revert the last change.
20144 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
20147 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
20149 (ix86_finalize_stack_frame_flags): This. Also clear
20150 frame_pointer_needed if -fno-omit-frame-pointer is used without
20152 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
20153 with ix86_finalize_stack_frame_flags.
20154 (ix86_expand_epilogue): Likewise.
20155 (ix86_expand_split_stack_prologue): Likewise.
20157 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
20160 * config/i386/i386.c (get_builtin_code_for_version): Set priority
20161 to P_AES for Westmere.
20163 2017-08-07 Jonathan Yong <10walls@gmail.com>
20165 * config/i386/mingw.opt (fset-stack-executable): Removed.
20166 * config/i386/cygming.opt (fset-stack-executable): Moved
20168 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
20170 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
20172 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
20174 2017-08-07 Marek Polacek <polacek@redhat.com>
20176 PR middle-end/81737
20177 * fold-const.c (fold_indirect_ref_1): Check type_domain.
20179 2017-08-07 Martin Liska <mliska@suse.cz>
20181 * attribs.h (canonicalize_attr_name): New function.
20182 (cmp_attribs): Move from c-format.c and adjusted.
20183 (is_attribute_p): Moved from tree.h.
20184 * tree-inline.c: Add new includes.
20185 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
20186 (private_is_attribute_p): Remove.
20187 (private_lookup_attribute): Likewise.
20188 (private_lookup_attribute_by_prefix): Simplify.
20189 (remove_attribute): Use is_attribute_p.
20190 * tree.h: Remove removed declarations.
20192 2017-08-07 Jakub Jelinek <jakub@redhat.com>
20194 PR middle-end/81698
20195 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
20196 instead of computing it in the function. Formatting fix.
20197 (expand_case): Don't rely on default_edge being the first edge,
20198 clear it if removing it, pass default_edge to
20199 emit_case_dispatch_table.
20200 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
20203 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
20205 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
20206 insn in the function, emit NOP after the insn.
20208 2017-08-06 Tom de Vries <tom@codesourcery.com>
20210 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
20213 2017-08-06 Tom de Vries <tom@codesourcery.com>
20215 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
20218 2017-08-04 Yury Gribov <tetra2005@gmail.com>
20220 PR tree-optimization/57371
20221 * match.pd: New pattern.
20223 2017-08-04 Marek Polacek <polacek@redhat.com>
20225 PR middle-end/81695
20226 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
20227 perform the computation in offset_int.
20229 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
20231 PR tree-optimization/81136
20232 * tree-vectorizer.h: Include tree-hash-traits.h.
20233 (vec_base_alignments): New typedef.
20234 (vec_info): Add a base_alignments field.
20235 (vect_record_base_alignments): Declare.
20236 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
20238 (DR_IS_CONDITIONAL_IN_STMT): New macro.
20239 (create_data_ref): Add an is_conditional_in_stmt argument.
20240 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
20241 the is_conditional_in_stmt field.
20242 (data_ref_loc): Add an is_conditional_in_stmt field.
20243 (get_references_in_stmt): Set the is_conditional_in_stmt field.
20244 (find_data_references_in_stmt): Update call to create_data_ref.
20245 (graphite_find_data_references_in_stmt): Likewise.
20246 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
20247 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
20248 (vect_record_base_alignment): New function.
20249 (vect_record_base_alignments): Likewise.
20250 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
20251 for nested statements even if we fail to compute a misalignment.
20252 Use pooled base alignments for unconditional references.
20253 (vect_find_same_alignment_drs): Compare base addresses instead
20255 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
20256 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
20258 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
20260 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
20261 Add an explicit name for the enum. Use auto_vec for slp_instances
20262 and grouped_stores.
20263 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
20265 (_bb_vec_info): Add a constructor and destructor.
20266 (vinfo_for_stmt): Return NULL for uids of -1 as well.
20267 (destroy_loop_vec_info): Delete.
20268 (vect_destroy_datarefs): Likewise.
20269 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
20270 (vec_info::vec_info): New function.
20271 (vec_info::~vec_info): Likewise.
20272 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
20273 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
20274 destroy_loop_vec_info.
20275 * tree-vect-loop.c (new_loop_vec_info): Replace with...
20276 (_loop_vec_info::_loop_vec_info): ...this.
20277 (destroy_loop_vec_info): Replace with...
20278 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
20279 the stmt_vec_infos. Leave handling of vec_info information to its
20280 destructor. Remove explicit vector releases.
20281 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
20282 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
20283 * tree-vect-slp.c (new_bb_vec_info): Replace with...
20284 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
20285 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
20286 (destroy_bb_vec_info): Replace with...
20287 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
20288 information to its destructor.
20289 (vect_slp_analyze_bb_1): Use new and delete instead of
20290 new_bb_vec_info and destroy_bb_vec_info.
20291 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
20294 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
20296 * tree-data-ref.h (subscript): Add access_fn field.
20297 (data_dependence_relation): Add could_be_independent_p.
20298 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
20299 (same_access_functions): Move to tree-data-ref.c.
20300 * tree-data-ref.c (ref_contains_union_access_p): New function.
20301 (access_fn_component_p): Likewise.
20302 (access_fn_components_comparable_p): Likewise.
20303 (dr_analyze_indices): Add a reference to access_fn_component_p.
20304 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
20306 (constant_access_functions): Likewise.
20307 (add_other_self_distances): Likewise.
20308 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
20309 (initialize_data_dependence_relation): Use XCNEW and remove
20310 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
20311 of access functions that have the same type. Allow the
20312 subsequence to end with different bases in some circumstances.
20313 Record the chosen access functions in SUB_ACCESS_FN.
20314 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
20315 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
20316 (subscript_dependence_tester_1): Likewise dra and drb.
20317 (build_classic_dist_vector): Update calls accordingly.
20318 (subscript_dependence_tester): Likewise.
20319 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
20320 DDR_COULD_BE_INDEPENDENT_P.
20321 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
20322 comp_alias_ddrs instead of may_alias_ddrs.
20323 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
20325 (vect_analyze_data_ref_dependence): Use it if
20326 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
20327 distance vectors if that fails.
20328 (dependence_distance_ge_vf): New function.
20329 (vect_prune_runtime_alias_test_list): Use it. Don't clear
20330 LOOP_VINFO_MAY_ALIAS_DDRS.
20332 2017-08-04 Richard Biener <rguenther@suse.de>
20334 PR middle-end/81705
20335 * fold-const.c (fold_binary_loc): Properly restrict
20336 minus_var0 && minus_var1 case when associating undefined overflow
20339 2017-08-04 Tom de Vries <tom@codesourcery.com>
20341 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
20343 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20345 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20346 Don't start diagnostic messages with a capital letter.
20347 * config/rs6000/rs6000.c (rs6000_option_override_internal):
20349 (rs6000_invalid_builtin): Likewise.
20350 (rs6000_trampoline_init): Likewise.
20352 2017-08-03 Jakub Jelinek <jakub@redhat.com>
20355 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
20356 after setting changeable df flags.
20358 2017-08-03 Richard Biener <rguenther@suse.de>
20360 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
20361 up if the use is in USE - X.
20363 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
20365 * toplev.c (dumpfile.h): New include.
20366 (internal_error_reentered): New static function. Use it...
20367 (internal_error_function): ...here to handle reentered internal_error.
20369 2017-08-03 Richard Biener <rguenther@suse.de>
20371 PR middle-end/81148
20372 * fold-const.c (split_tree): Add minus_var and minus_con
20373 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
20374 here but always use minus_*.
20375 (associate_trees): Assert we never associate with MINUS_EXPR
20376 and NULL first operand. Do not recurse for PLUS_EXPR operands
20377 when associating as MINUS_EXPR either.
20378 (fold_binary_loc): Track minus_var and minus_con.
20380 2017-08-03 Tom de Vries <tom@codesourcery.com>
20383 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
20384 ACCEL_COMPILER, apply finish_options on
20385 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
20387 2017-08-03 Tom de Vries <tom@codesourcery.com>
20390 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
20391 function_entry_patch_area_size > 0.
20393 2017-08-03 Jakub Jelinek <jakub@redhat.com>
20396 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
20397 instead of 10??LU, perform unit multiplication in wide_int,
20398 don't change alloc_object_size_limit if the limit is larger
20401 PR tree-optimization/81655
20402 PR tree-optimization/81588
20403 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
20404 the case when ranges[i].low and high are 1 for unsigned type with
20407 PR middle-end/81052
20408 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
20409 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
20411 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20413 * tree-vrp.h: Add include guard.
20415 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
20418 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
20419 (ud2): New insn pattern.
20420 * config/i386/i386.c (ix86_expand_epilogue):
20421 For naked functions, generate ud2 instead of trap insn.
20423 2017-08-02 Marek Polacek <polacek@redhat.com>
20426 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
20428 2017-08-02 Tom de Vries <tom@codesourcery.com>
20429 Cesar Philippidis <cesar@codesourcery.com>
20431 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
20432 Add missing edge probabilities.
20434 2017-08-02 Tamar Christina <tamar.christina@arm.com>
20436 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
20437 Correct endianness.
20439 2017-08-02 Jakub Jelinek <jakub@redhat.com>
20441 PR middle-end/79499
20442 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
20443 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
20444 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
20445 prologue_seq sequences - if any.
20447 2017-08-02 Richard Biener <rguenther@suse.de>
20449 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
20450 via vectors if supported, integer extracts via punning if supported
20451 or otherwise vector extracts.
20453 2017-08-02 Richard Biener <rguenther@suse.de>
20455 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
20457 (bitmap_insert_into_set): ... this.
20459 2017-08-02 Richard Biener <rguenther@suse.de>
20461 PR tree-optimization/81633
20463 2015-08-17 Alan Hayward <alan.hayward@arm.com>
20465 PR tree-optimization/71752
20466 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
20468 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
20470 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
20471 (machine_function::call_ms2sysv_pad_out): Remove field.
20472 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
20473 (ix86_compute_frame_layout): Likewise.
20475 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
20478 * config/i386/i386.c (ix86_set_func_type): Disallow naked
20479 attribute with interrupt attribute.
20481 2017-08-01 Andrew Pinski <apinski@cavium.com>
20483 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
20484 BIT_INSERT_EXPR's operand 1
20485 to see if the types precision matches.
20487 2017-08-01 Martin Liska <mliska@suse.cz>
20489 PR middle-end/70140
20490 * builtins.c (expand_builtin_memcpy_args): Remove.
20491 (expand_builtin_memcpy): Call newly added function
20492 expand_builtin_memory_copy_args.
20493 (expand_builtin_memcpy_with_bounds): Likewise.
20494 (expand_builtin_mempcpy): Remove last argument.
20495 (expand_builtin_mempcpy_with_bounds): Likewise.
20496 (expand_builtin_memory_copy_args): New function created from
20497 expand_builtin_mempcpy_args with small modifications.
20498 (expand_builtin_mempcpy_args): Remove.
20499 (expand_builtin_stpcpy): Remove unused argument.
20500 (expand_builtin): Likewise.
20501 (expand_builtin_with_bounds): Likewise.
20503 2017-08-01 Martin Liska <mliska@suse.cz>
20506 Make mempcpy more optimal (PR middle-end/70140).
20508 2017-08-01 Jakub Jelinek <jakub@redhat.com>
20511 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
20512 __builtin_vec_cmpne verify both arguments are compatible vectors
20513 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
20514 verify arg1_type is a pointer or array type. For __builtin_vec_st,
20515 move computation of aligned to after checking the argument types.
20519 * config/rs6000/vsx.md (vextract_fp_from_shorth,
20520 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
20523 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
20524 Doug Rupp <rupp@adacore.com>
20525 Olivier Hainque <hainque@adacore.com>
20527 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
20528 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
20530 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
20531 for TARGET_OS_CPP_BUILTIN.
20532 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
20533 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
20535 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
20536 passing required abi options to the assembler for EABI configurations.
20537 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
20538 of .text.hot and .text.unlikely sections for kernel modules when
20539 using ARM style exceptions.
20540 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
20541 options. Add EXTRA_CC1_SPEC.
20542 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
20544 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
20546 (ARM_TARGET2_DWARF_FORMAT): Define.
20547 * config/arm/t-vxworks: Adjust multilib control to removal of the
20548 Diab command line options.
20550 2017-08-01 Martin Liska <mliska@suse.cz>
20552 PR gcov-profile/81561
20553 * gcov.c (unblock): Make unblocking safe as we need to preserve
20554 index correspondence of blocks and block_lists.
20556 2017-08-01 Richard Biener <rguenther@suse.de>
20558 PR tree-optimization/81181
20559 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
20560 (compute_antic): ... end of iteration here.
20562 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
20564 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
20565 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
20566 (ftree-slp-vectorize): Likewise.
20567 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
20568 can no longer be set independent of flag_tree_loop_vectorize.
20569 * omp-general.c (emp_max_vf): Likewise.
20570 * opts.c (enable_fdo_optimizations): Remove references to
20571 flag_tree_vectorize, these are now implicit.
20572 (common_handle_option): Remove handling for OPT_ftree_vectorize,
20573 and leave it for the options machinery.
20575 2017-08-01 Martin Liska <mliska@suse.cz>
20577 PR middle-end/70140
20578 * builtins.c (expand_builtin_memcpy_args): Remove.
20579 (expand_builtin_memcpy): Call newly added function
20580 expand_builtin_memory_copy_args.
20581 (expand_builtin_memcpy_with_bounds): Likewise.
20582 (expand_builtin_mempcpy): Remove last argument.
20583 (expand_builtin_mempcpy_with_bounds): Likewise.
20584 (expand_builtin_memory_copy_args): New function created from
20585 expand_builtin_mempcpy_args with small modifications.
20586 (expand_builtin_mempcpy_args): Remove.
20587 (expand_builtin_stpcpy): Remove unused argument.
20588 (expand_builtin): Likewise.
20589 (expand_builtin_with_bounds): Likewise.
20591 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
20594 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
20595 print "ds:" only for immediates in generic address space.
20597 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
20600 * config/i386/i386.c (ix86_funciton_naked): New prototype.
20601 (ix86_function_ok_for_sibcall): Return false for naked functions.
20603 2017-08-01 Richard Biener <rguenther@suse.de>
20605 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
20606 (compute_antic): Seed worklist with exit block predecessors.
20607 * cfganal.c (dfs_find_deadend): For a cycle return the source
20608 of the edge closing it.
20610 2017-08-01 Tamar Christina <tamar.christina@arm.com>
20612 * config/aarch64/aarch64.c
20613 (aarch64_can_const_movi_rtx_p): Move 0 check.
20615 2017-08-01 Bin Cheng <bin.cheng@arm.com>
20617 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
20618 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
20620 * match.pd: Ditto in address comparison pattern.
20622 2017-08-01 Bin Cheng <bin.cheng@arm.com>
20624 PR tree-optimization/81627
20625 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
20626 closed ssa form for store-store chain.
20628 2017-08-01 Bin Cheng <bin.cheng@arm.com>
20630 PR tree-optimization/81620
20631 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
20632 for store-store chain.
20634 2017-08-01 Jakub Jelinek <jakub@redhat.com>
20636 PR tree-optimization/81588
20637 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
20638 ranges[i].in_p, invert comparison code ccode. For >/>=,
20639 swap rhs1 and rhs2 and comparison code unconditionally,
20640 for </<= don't do that. Don't swap rhs1/rhs2 again if
20641 ranges[i].in_p, instead invert comparison code ccode if
20642 opcode or oe->rank is BIT_IOR_EXPR.
20645 * optabs.def (vec_extract_optab, vec_init_optab): Change from
20646 a direct optab to conversion optab.
20647 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
20648 with GET_MODE_INNER as last argument instead of optab_handler.
20649 * expmed.c (extract_bit_field_1): Likewise. Use vector from
20650 vector extraction if possible and optab is available.
20651 * expr.c (store_constructor): Use convert_optab_handler instead
20652 of optab_handler. Use vector initialization from smaller
20653 vectors if possible and optab is available.
20654 * tree-vect-stmts.c (vectorizable_load): Likewise.
20655 * doc/md.texi (vec_extract, vec_init): Document that the optabs
20656 now have two modes.
20657 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
20658 of vec_init from half-sized vectors with the same element mode.
20659 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
20660 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
20661 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
20662 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
20663 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
20664 after mode in gen_vec_extract* calls.
20665 (vec_extract<mode>): Renamed to ...
20666 (vec_extract<mode><ssescalarmodelower>): ... this.
20667 (vec_extract<mode><ssehalfvecmodelower>): New expander.
20668 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
20669 element mode after mode in gen_vec_init* calls.
20670 (VEC_INIT_HALF_MODE): New mode iterator.
20671 (vec_init<mode>): Renamed to ...
20672 (vec_init<mode><ssescalarmodelower>): ... this.
20673 (vec_init<mode><ssehalfvecmodelower>): New expander.
20674 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
20675 (vec_extractv2sfsf): ... this.
20676 (vec_initv2sf): Renamed to ...
20677 (vec_initv2sfsf): ... this.
20678 (vec_extractv2si): Renamed to ...
20679 (vec_extractv2sisi): ... this.
20680 (vec_initv2si): Renamed to ...
20681 (vec_initv2sisi): ... this.
20682 (vec_extractv4hi): Renamed to ...
20683 (vec_extractv4hihi): ... this.
20684 (vec_initv4hi): Renamed to ...
20685 (vec_initv4hihi): ... this.
20686 (vec_extractv8qi): Renamed to ...
20687 (vec_extractv8qiqi): ... this.
20688 (vec_initv8qi): Renamed to ...
20689 (vec_initv8qiqi): ... this.
20690 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
20691 (vec_init<mode>): Renamed to ...
20692 (vec_init<mode><VEC_base_l>): ... this.
20693 (vec_extract<mode>): Renamed to ...
20694 (vec_extract<mode><VEC_base_l>): ... this.
20695 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
20696 (vec_initv2sfsf): ... this.
20697 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
20698 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
20699 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
20700 element mode after mode in gen_vec_init* calls.
20701 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
20702 (vec_init<mode><Vel>): ... this.
20703 (vec_extract<mode>): Renamed to ...
20704 (vec_extract<mode><Vel>): ... this.
20705 * config/aarch64/iterators.md (Vel): New mode attribute.
20706 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
20707 Add element mode after mode in gen_vec_extract* calls.
20708 * config/s390/vector.md (non_vec_l): New mode attribute.
20709 (vec_extract<mode>): Renamed to ...
20710 (vec_extract<mode><non_vec_l>): ... this.
20711 (vec_init<mode>): Renamed to ...
20712 (vec_init<mode><non_vec_l>): ... this.
20713 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
20714 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
20716 * config/arm/iterators.md (V_elem_l): New mode attribute.
20717 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
20718 (vec_extract<mode><V_elem_l>): ... this.
20719 (vec_extractv2di): Renamed to ...
20720 (vec_extractv2didi): ... this.
20721 (vec_init<mode>): Renamed to ...
20722 (vec_init<mode><V_elem_l>): ... this.
20723 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
20724 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
20725 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
20726 Add element mode after gen_vec_extract* calls.
20727 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
20728 (vec_init<mode><unitmode>): ... this.
20729 (vec_extract<mode>): Renamed to ...
20730 (vec_extract<mode><unitmode>): ... this.
20731 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
20732 (vec_init<mode><unitmode>): ... this.
20733 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
20734 (vec_initv2sfsf): ... this.
20735 (vec_extractv2sf): Renamed to ...
20736 (vec_extractv2sfsf): ... this.
20737 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
20738 Add element mode after gen_vec_extract* calls.
20739 * config/mips/mips.md (unitmode): New mode iterator.
20740 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
20741 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
20742 * config/spu/spu.md (inner_l): New mode attribute.
20743 (vec_init<mode>): Renamed to ...
20744 (vec_init<mode><inner_l>): ... this.
20745 (vec_extract<mode>): Renamed to ...
20746 (vec_extract<mode><inner_l>): ... this.
20747 * config/sparc/sparc.md (veltmode): New mode iterator.
20748 (vec_init<VMALL:mode>): Renamed to ...
20749 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
20750 * config/ia64/vect.md (vec_initv2si): Renamed to ...
20751 (vec_initv2sisi): ... this.
20752 (vec_initv2sf): Renamed to ...
20753 (vec_initv2sfsf): ... this.
20754 (vec_extractv2sf): Renamed to ...
20755 (vec_extractv2sfsf): ... this.
20756 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
20757 (vec_init<mode>): Renamed to ...
20758 (vec_init<mode><VEC_base_l>): ... this.
20759 (vec_extract<mode>): Renamed to ...
20760 (vec_extract<mode><VEC_base_l>): ... this.
20761 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
20762 (vec_initv2sfsf): ... this.
20763 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
20764 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
20765 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
20766 gen_vec_init* calls.
20768 2017-08-01 Richard Biener <rguenther@suse.de>
20770 PR tree-optimization/81297
20771 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
20772 TREE_OVERFLOW from INTEGER_CSTs.
20774 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
20776 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
20778 2017-07-31 Carl Love <cel@us.ibm.com>
20780 * config/rs6000/rs6000-c: Add support for built-in functions
20781 vector signed char vec_xl_be (signed long long, signed char *);
20782 vector unsigned char vec_xl_be (signed long long, unsigned char *);
20783 vector signed int vec_xl_be (signed long long, signed int *);
20784 vector unsigned int vec_xl_be (signed long long, unsigned int *);
20785 vector signed long long vec_xl_be (signed long long, signed long long *);
20786 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
20787 vector signed short vec_xl_be (signed long long, signed short *);
20788 vector unsigned short vec_xl_be (signed long long, unsigned short *);
20789 vector double vec_xl_be (signed long long, double *);
20790 vector float vec_xl_be (signed long long, float *);
20791 * config/rs6000/altivec.h (vec_xl_be): Add #define.
20792 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
20793 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
20795 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
20796 (altivec_expand_builtin): Add switch statement to call
20797 altivec_expand_xl_be for each builtin.
20798 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
20799 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
20800 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
20801 __builtin_vsx_le_be_v16qi.
20802 * doc/extend.texi: Update the built-in documentation file for the
20803 new built-in functions.
20805 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
20808 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
20810 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
20812 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20814 * config.gcc: Add z14.
20815 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
20816 CPU model numbers for z13s and z14.
20817 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
20819 * config/s390/s390-opts.h (enum processor_type): Rename
20820 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
20821 * config/s390/s390.c (processor_table): Add field for CPU name to
20822 be passed to Binutils.
20823 (s390_asm_output_machine_for_arch): Use the new field in
20824 processor_table for Binutils.
20825 (s390_expand_builtin): Replace arch12 with z14.
20826 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
20827 (s390_get_sched_attrmask): Likewise.
20828 (s390_get_unit_mask): Likewise.
20829 * config/s390/s390.opt: Add z14 to processor_type enum.
20831 2017-07-31 Martin Jambor <mjambor@suse.cz>
20834 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
20835 regardless of optimization level.
20837 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
20838 Martin Liska <mliska@suse.cz>
20840 * predict.def: Remove old comment and adjust probability.
20841 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
20842 PREDICT statements.
20844 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
20847 * config/i386/i386.c (ix86_function_naked): New function.
20848 (ix86_can_use_return_insn_p): Return false for naked functions.
20849 (ix86_expand_prologue): Skip prologue for naked functions.
20850 (ix86_expand_epilogue): Skip epilogue for naked functions
20851 and emit trap instruction.
20852 (ix86_warn_func_return): New function.
20853 (ix86_attribute_table): Add "naked" attribute specification.
20854 (TARGET_WARN_FUNC_RETURN): Define.
20855 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
20857 2017-07-31 Martin Liska <mliska@suse.cz>
20859 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
20860 (dump_gimple_bb_header): Always dump BB info.
20861 (pp_cfg_jump): Do not append info about BB when dumping a jump.
20863 2017-07-31 Martin Liska <mliska@suse.cz>
20866 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
20867 also with current_function_decl non-null equality.
20869 2017-07-31 Jakub Jelinek <jakub@redhat.com>
20872 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
20873 change type to the element type, instead add eltype variable and
20874 use it where we are interested in the element type.
20876 PR tree-optimization/81603
20877 * ipa-polymorphic-call.c
20878 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
20879 offset arithmetic in offset_int, bail out if the resulting bit offset
20880 doesn't fit into shwi.
20882 2017-07-31 Martin Liska <mliska@suse.cz>
20884 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
20885 (gimplify_save_expr): Fix comment.
20887 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
20890 * config/i386/i386.c (ix86_function_arg): Update arguments for
20892 (ix86_compute_frame_layout): Set the initial stack offset to
20893 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
20894 INCOMING_FRAME_SP_OFFSET.
20895 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
20896 stack before exception handler returns.
20897 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
20898 the 'ERROR_CODE' for exception handler.
20900 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
20902 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
20903 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
20904 (ASM_OUTPUT_REG_POP): Ditto.
20905 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
20906 instead of asm_fprintf to output pure string.
20908 2017-07-29 Jakub Jelinek <jakub@redhat.com>
20910 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
20911 to imported_module_or_decl hook.
20912 (debug_nothing_tree_tree_tree_bool): Remove.
20913 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
20914 * debug.c (do_nothing_debug_hooks): Use
20915 debug_nothing_tree_tree_tree_bool_bool instead of
20916 debug_nothing_tree_tree_tree_bool.
20917 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
20918 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
20919 * sdbout.c (sdb_debug_hooks): Likewise.
20920 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
20921 (gen_namespace_die): Add DW_AT_export_symbols attribute if
20923 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
20924 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
20925 attribute, don't add anything.
20927 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20929 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
20930 (fold_build2_stat_loc): Likewise.
20931 (fold_build3_stat_loc): Likewise.
20932 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
20933 (fold_build1_loc): Remove macro.
20934 (fold_build2_loc): Likewise.
20935 (fold_build3_loc): Likewise.
20937 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20939 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
20940 (gimple_build_debug_bind_source_stat): Likewise.
20941 * gimple.h (gimple_build_debug_bind): Remove macro.
20942 (gimple_build_debug_bind_source): Likewise.
20944 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20946 * bitmap.c (bitmap_alloc): Adjust.
20947 (bitmap_gc_alloc): Likewise.
20948 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
20950 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20952 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
20953 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
20954 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
20955 (bitmap_gc_alloc_stat): Likewise.
20956 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
20958 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20960 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
20961 * rtl.h (shallow_copy_rtx): Remove macro.
20963 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20965 * emit-rtl.c (gen_raw_REG): Adjust.
20966 * gengenrtl.c (gendef): Likewise.
20967 * rtl.c (rtx_alloc_stat): Remove _stat from name.
20968 * rtl.h (rtx_alloc): Remove macro.
20970 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20972 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
20973 (build_tree_list_stat): Likewise.
20974 * tree.h (build_tree_list): Remove macro.
20975 (build_tree_list_vec): Likewise.
20977 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20979 * tree.c (make_vector_stat): Remove _stat from name.
20980 (build_vector_stat): Likewise.
20981 * tree.h (make_vector_stat): Remove macro.
20982 (build_vector_stat): Likewise.
20984 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20986 * tree.h (build_var_debug_value): Remove prototype.
20988 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20990 * tree.c (tree_cons_stat): Remove _stat from name.
20991 * tree.h (tree_cons): Remove macro.
20993 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20995 * tree.c (build_vl_exp_stat): Remove _stat from name.
20996 * tree.h (build_vl_exp): Remove macro.
20998 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21000 * tree.c (build_decl_stat): Remove _stat from name.
21001 * tree.h (build_decl): Remove macro.
21003 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21005 * gimple.c (gimple_build_with_ops_stat): Adjust.
21006 (gimple_alloc_stat): Remove _stat from name.
21007 * gimple.h (gimple_alloc): Remove macro.
21009 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21011 * tree.c (make_tree_vec_stat): Remove _stat from name.
21012 (grow_tree_vec_stat): Likewise.
21013 * tree.h (make_tree_vec_stat): Adjust prototype.
21014 (grow_tree_vec_stat): Likewise.
21015 (make_tree_vec): Remove macro.
21016 (grow_tree_vec): Likewise.
21018 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21020 * fold-const.c (fold_build1_stat_loc): Adjust.
21021 (fold_build2_stat_loc): Likewise.
21022 (fold_build3_stat_loc): Likewise.
21023 * tree.c (build0_stat): Remove _stat from name.
21024 (build1_stat): Likewise.
21025 (build2_stat): Likewise.
21026 (build3_stat): Likewise.
21027 (build4_stat): Likewise.
21028 (build5_stat): Likewise.
21029 * tree.h (build1_loc): Remove macro, and rename _stat function
21031 (build2_loc): Likewise.
21032 (build3_loc): Likewise.
21033 (build4_loc): Likewise.
21034 (build5_loc): Likewise.
21036 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21038 * tree.c (make_int_cst_stat): Remove _stat from name.
21039 * tree.h (make_int_cst_stat): Adjust prototype.
21040 (make_int_cst): Remove macro.
21042 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21044 * tree.c (make_tre_binfo_stat): Remove _stat from name.
21045 * tree.h (make_tree_binfo_stat): Adjust prototype.
21046 (make_tree_binfo): Remove.
21048 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21050 * tree.c (copy_node_stat): Rename to copy_node.
21051 (build_distinct_type_copy): Adjust.
21052 * tree.h (copy_node_stat): Adjust prototype.
21053 (copy_node): Remove macro.
21055 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21057 * tree.c (make_node_stat): rename to make_node.
21058 (build_tree_list_stat): Adjust.
21059 (build0_stat): Likewise.
21060 (build2_stat): Likewise.
21061 (build3_stat): Likewise.
21062 (build4_stat): Likewise.
21063 (build5_stat): Likewise.
21064 (build_decl_stat): Likewise.
21065 * tree.h (make_node_stat): Adjust prototype.
21066 (make_node): remove macro.
21068 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
21070 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
21071 (PPC_FEATURE2_SCV): Likewise.
21072 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
21074 2017-07-28 Tamar Christina <tamar.christina@arm.com>
21076 * config/aarch64/aarch64.c
21077 (aarch64_internal_mov_immediate): Add new special pattern.
21078 * config/aarch64/aarch64.md (*movdi_aarch64):
21079 Add reg/32bit const mov case.
21081 2017-07-28 Tamar Christina <tamar.christina@arm.com>
21082 Richard Sandiford <richard.sandiford@linaro.org>
21084 * config/aarch64/aarch64.md (mov<mode>): Generalize.
21085 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
21086 Add integer and movi cases.
21087 (movi-split-hf-df-sf split, fp16): New.
21088 (enabled): Added TARGET_FP_F16INST.
21089 * config/aarch64/iterators.md (GPF_HF): New.
21090 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
21092 2017-07-28 Tamar Christina <tamar.christina@arm.com>
21094 * config/aarch64/aarch64.c
21095 (aarch64_simd_container_mode): Add prototype.
21096 (aarch64_expand_mov_immediate): Add HI support.
21097 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
21098 (aarch64_can_const_movi_rtx_p): New.
21099 (aarch64_preferred_reload_class):
21100 Remove restrictions of using FP registers for certain SIMD operations.
21101 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
21102 (aarch64_valid_floating_const): Add integer move validation.
21103 (aarch64_simd_imm_scalar_p): Remove.
21104 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
21105 (aarch64_legitimate_constant_p): Expand list of supported cases.
21106 * config/aarch64/aarch64-protos.h
21107 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
21108 (aarch64_reinterpret_float_as_int): New.
21109 (aarch64_simd_imm_scalar_p): Remove.
21110 * config/aarch64/constraints.md (Uvi): New.
21111 (Dd): Split into Ds and new Dd.
21112 * config/aarch64/aarch64.md (*movsi_aarch64):
21114 (*movdi_aarch64): Add SIMD mov case.
21116 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21118 * tree-predcom.c: (struct chain): Handle store-store chain in which
21119 stores for elimination only store loop invariant values.
21120 (execute_pred_commoning_chain): Ditto.
21121 (prepare_initializers_chain_store_elim): Ditto.
21122 (prepare_finalizers): Ditto.
21123 (is_inv_store_elimination_chain): New function.
21124 (initialize_root_vars_store_elim_1): New function.
21126 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21128 * tree-predcom.c: Revise general description of the pass.
21129 (enum chain_type): New enum type for store elimination.
21130 (struct chain): New field supporting store elimination.
21131 (struct component): Ditto.
21132 (dump_chain): Dump store-stores chain.
21133 (release_chain): Release resources.
21134 (split_data_refs_to_components): Compute and create component
21135 contains only stores for elimination.
21136 (get_chain_last_ref_at): New function.
21137 (make_invariant_chain): Initialization.
21138 (make_rooted_chain): Specify chain type in parameter and record it.
21139 (add_looparound_copies): Skip for store-stores chain.
21140 (determine_roots_comp): Compute type of chain and pass it to
21142 (initialize_root_vars_store_elim_2): New function.
21143 (finalize_eliminated_stores): New function.
21144 (remove_stmt): Handle store for elimination.
21145 (execute_pred_commoning_chain): Execute predictive commoning on
21146 store-store chains.
21147 (determine_unroll_factor): Skip unroll for store-stores chain.
21148 (prepare_initializers_chain_store_elim): New function.
21149 (prepare_initializers_chain): Hanlde store-store chain.
21150 (prepare_finalizers_chain, prepare_finalizers): New function.
21151 (tree_predictive_commoning_loop): Return integer value indicating
21152 if loop is unrolled or lcssa form is corrupted.
21153 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
21155 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21157 * tree-predcom.c (initialize_root): Delete.
21158 (execute_pred_commoning_chain): Initialize root vars and replace
21159 reference of non-combined chain directly, rather than call above
21162 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21164 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
21165 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
21167 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21169 * tree-predcom.c (struct chain): New field init_seq.
21170 (release_chain): Release init_seq.
21171 (prepare_initializers_chain): Record intialization stmts in above
21173 (insert_init_seqs): New function.
21174 (tree_predictive_commoning_loop): Call insert_init_seqs.
21176 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21178 * tree-predcom.c (determine_roots_comp): Skip trivial components.
21180 2017-07-28 Richard Biener <rguenther@suse.de>
21182 * match.pd: Remove superfluous :c.
21183 * genmatch.c (simplify::id): Add member.
21184 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
21186 (current_id): New global.
21187 (dt_node::parent): Move from ...
21188 (dt_operand::parent): ... here. Add for_id member.
21189 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
21190 (decision_tree::find_node): Relax order requirement when
21191 merging DT_TRUE nodes to ones inbetween the current simplify
21192 and the one we try to merge with. Add diagnostic whenever
21193 we need to enforce pattern order by not merging.
21194 (decision_tree::insert): Set current_id.
21195 (decision_tree::print_node): Dump parent node and for_id.
21196 (parser::last_id): Add member.
21197 (parser::push_simplify): Assign unique id.
21198 (parser::parser): Initialize last_id.
21200 2017-07-28 Martin Liska <mliska@suse.cz>
21203 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
21204 gimple_build_debug_bind.
21206 2017-07-28 Richard Biener <rguenther@suse.de>
21208 PR tree-optimization/81502
21209 * match.pd: Add pattern combining BIT_INSERT_EXPR with
21211 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
21213 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
21214 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
21215 for BIT_FIELD_REF args.
21216 * fold-const.c (make_bit_field_ref): Likewise.
21217 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
21219 2017-07-28 Jakub Jelinek <jakub@redhat.com>
21222 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
21223 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
21224 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
21225 Or it into SANITIZER_UNDEFINED.
21226 * ubsan.c: Include gimple-fold.h and varasm.h.
21227 (ubsan_expand_ptr_ifn): New function.
21228 (instrument_pointer_overflow): New function.
21229 (maybe_instrument_pointer_overflow): New function.
21230 (instrument_object_size): Formatting fix.
21231 (pass_ubsan::execute): Call instrument_pointer_overflow
21232 and maybe_instrument_pointer_overflow.
21233 * internal-fn.c (expand_UBSAN_PTR): New function.
21234 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
21235 * sanitizer.def (__ubsan_handle_pointer_overflow,
21236 __ubsan_handle_pointer_overflow_abort): New builtins.
21237 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
21238 * internal-fn.def (UBSAN_PTR): New internal function.
21239 * opts.c (sanitizer_opts): Add pointer-overflow.
21240 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
21241 * fold-const.c (build_range_check): Compute pointer range check in
21242 integral type if pointer arithmetics would be needed. Formatting
21245 2017-07-28 Martin Liska <mliska@suse.cz>
21248 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
21249 parameters that are of a variable-length.
21251 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
21253 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
21255 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
21256 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
21257 (CRT_CALL_STATIC_FUNCTION): Likewise.
21258 (ASM_DEFAULT_SPEC): New define.
21259 (ASM_SPEC32): Likewise.
21260 (ASM_SPEC64): Likewise.
21261 (ASM_SPEC_COMMON): Likewise.
21262 (ASM_SPEC): Likewise.
21263 (INVALID_64BIT): Likewise.
21264 (LINK_OS_DEFAULT_SPEC): Likewise.
21265 (LINK_OS_SPEC32): Likewise.
21266 (LINK_OS_SPEC64): Likewise.
21267 (POWERPC_LINUX): Likewise.
21268 (PTRDIFF_TYPE): Likewise.
21269 (RESTORE_FP_PREFIX): Likewise.
21270 (RESTORE_FP_SUFFIX): Likewise.
21271 (SAVE_FP_PREFIX): Likewise.
21272 (SAVE_FP_SUFFIX): Likewise.
21273 (SIZE_TYPE): Likewise.
21274 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
21275 (TARGET_64BIT): Likewise.
21276 (TARGET_64BIT): Likewise.
21277 (TARGET_AIX): Likewise.
21278 (WCHAR_TYPE_SIZE): Likewise.
21279 (WCHAR_TYPE): Undefine.
21280 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
21281 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
21282 (CPP_OS_RTEMS_SPEC): Delete.
21283 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
21284 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
21286 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
21288 2017-07-28 Jakub Jelinek <jakub@redhat.com>
21290 PR tree-optimization/81578
21291 * tree-parloops.c (build_new_reduction): Bail out if
21292 reduction_code isn't one of the standard OpenMP reductions.
21293 Move the details printing after that decision.
21295 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
21297 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
21298 related to reload_in_progress.
21299 (splat_input_operand): Likewise.
21300 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
21302 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
21304 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
21305 (TARGET_INSTANTIATE_DECLS): Likewise.
21306 (legitimate_indexed_address_p): Delete reload_in_progress code.
21307 (rs6000_debug_legitimate_address_p): Likewise.
21308 (rs6000_eliminate_indexed_memrefs): Likewise.
21309 (rs6000_emit_le_vsx_store): Likewise.
21310 (rs6000_emit_move_si_sf_subreg): Likewise.
21311 (rs6000_emit_move): Likewise.
21312 (register_to_reg_type): Likewise.
21313 (rs6000_pre_atomic_barrier): Likewise.
21314 (rs6000_machopic_legitimize_pic_address): Likewise.
21315 (rs6000_allocate_stack_temp): Likewise.
21316 (rs6000_address_for_fpconvert): Likewise.
21317 (rs6000_address_for_altivec): Likewise.
21318 (rs6000_secondary_memory_needed_rtx): Delete function.
21319 (rs6000_check_sdmode): Likewise.
21320 (rs6000_alloc_sdmode_stack_slot): Likewise.
21321 (rs6000_instantiate_decls): Likewise.
21322 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
21323 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
21324 Delete reload_in_progress.
21325 (*vec_reload_and_plus_<mptrsize>): Likewise.
21326 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
21327 (vsx_div_v2di): Likewise.
21328 (vsx_udiv_v2di): Likewise.
21330 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
21332 * config/rs6000/rs6000.opt (mlra): Replace with stub.
21333 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
21334 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
21335 (rs6000_debug_reg_global): Delete print of LRA status.
21336 (rs6000_option_override_internal): Delete dead LRA related code.
21337 (rs6000_lra_p): Delete function.
21338 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
21340 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
21342 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
21343 * config/riscv/rtems.h: New file.
21345 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21346 Sudakshina Das <sudi.das@arm.com>
21348 * config/aarch64/aarch64.md
21349 (define_split for and<mode>3nr_compare): Move
21350 non aarch64_logical_operand to a register.
21351 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
21352 register immediate operand to a register.
21353 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
21355 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
21357 PR middle-end/81564
21358 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
21360 2017-07-27 Richard Biener <rguenther@suse.de>
21362 PR tree-optimization/81573
21363 PR tree-optimization/81494
21364 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
21365 multi defuse cycle case.
21367 2017-07-27 Richard Biener <rguenther@suse.de>
21369 PR tree-optimization/81571
21370 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
21373 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
21375 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
21376 earlier and only if MASK_FPU is set. Adjust formatting.
21378 2017-07-27 Martin Liska <mliska@suse.cz>
21380 * opt-functions.awk: Add validation of value of Init.
21381 * optc-gen.awk: Pass new argument.
21383 2017-07-27 Martin Liska <mliska@suse.cz>
21385 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
21386 Fix wrong condition.
21388 2017-07-27 Martin Liska <mliska@suse.cz>
21390 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
21391 BBs and edges seen by autoFDO.
21393 2017-07-27 Richard Biener <rguenther@suse.de>
21395 PR tree-optimization/81502
21396 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
21397 with incompatible but same sized type.
21398 (execute_update_addresses_taken): Likewise.
21400 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
21402 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
21403 flag_tree_loop_vectorize rather than flag_tree_vectorize.
21405 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21408 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
21409 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
21410 Change s_operand to memory_operand.
21412 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
21414 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
21415 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
21416 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
21417 Emit instructions rather than returning an expression. Handle TFmode
21418 and KFmode by casting to TImode.
21419 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
21420 (rs6000_emit_le_vsx_store): Likewise.
21421 * config/rs6000/vsx.md (VSX_TI): New iterator.
21422 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
21423 (*vsx_le_undo_permute_<mode>): Likewise.
21424 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
21425 emit the split sequence.
21426 (*vsx_le_perm_store_<mode>): Likewise.
21428 2017-07-27 Jakub Jelinek <jakub@redhat.com>
21430 PR tree-optimization/81555
21431 PR tree-optimization/81556
21432 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
21433 if true, force CHANGED for the recursive invocation.
21434 (reassociate_bb): Remember original length of ops array, pass
21435 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
21437 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
21438 attributes for noipa attribute. For naked attribute use
21439 lookup_attribute first before lookup_attribute_spec.
21440 * final.c (rest_of_handle_final): Disable IPA RA for functions with
21442 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
21443 for functions with noipa attribute.
21444 (cgraph_externally_visible_p): Return true for functions with noipa
21446 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
21447 for functions with noipa attribute.
21448 * doc/extend.texi: Document noipa function attribute.
21449 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
21450 also for functions with noipa attribute.
21451 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
21453 2017-07-26 Andrew Pinski <apinski@cavium.com>
21455 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
21456 vec_unalign_load_cost and vec_unalign_store_cost.
21458 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
21460 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
21461 -mvsx-small-integer option.
21462 (ISA_3_0_MASKS_IEEE): Likewise.
21463 (OTHER_VSX_VECTOR_MASKS): Likewise.
21464 (POWERPC_MASKS): Likewise.
21465 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
21466 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
21467 code, only testing for DImode being allowed in non-VSX floating
21469 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
21470 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
21472 (rs6000_option_override_internal): Delete -mvsx-small-integer.
21473 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
21474 TARGET_P8_VECTOR test.
21475 (rs6000_secondary_reload_simple_move): Likewise.
21476 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
21477 since TARGET_P9_VECTOR was already tested.
21478 (rs6000_opt_masks): Remove -mvsx-small-integer.
21479 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
21480 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
21482 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
21483 test for TARGET_VEXTRACTUB was used, and that uses
21485 (p9 extract splitter): Likewise.
21486 (vsx_extract_<mode>_di_p9): Likewise.
21487 (vsx_extract_<mode>_store_p9): Likewise.
21488 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
21489 for TARGET_P9_VECTOR was used. Delete code that is now dead with
21490 the elimination of TARGET_VSX_SMALL_INTEGER.
21491 (vsx_extract_<mode>_p8): Likewise.
21492 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
21493 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
21494 (vsx_set_<mode>_p9): Likewise.
21495 (vsx_set_v4sf_p9): Likewise.
21496 (vsx_set_v4sf_p9_zero): Likewise.
21497 (vsx_insert_extract_v4sf_p9): Likewise.
21498 (vsx_insert_extract_v4sf_p9_2): Likewise.
21499 * config/rs6000/rs6000.md (sign extend splitter): Change
21500 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
21501 (floatsi<mode>2_lfiwax_mem): Likewise.
21502 (floatunssi<mode>2_lfiwzx_mem): Likewise.
21503 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
21504 since a test for TARGET_P9_VECTOR was used.
21505 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
21506 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
21507 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
21508 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
21509 TARGET_P8_VECTOR test.
21510 (fix_trunc<mode>si2_stfiwx): Likewise.
21511 (fix_trunc<mode>si2_internal): Likewise.
21512 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
21513 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
21515 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
21516 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
21517 TARGET_P8_VECTOR test.
21518 (fixuns_trunc<mode>si2_stfiwx): Likewise.
21519 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
21520 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
21522 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
21523 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
21524 since a test for TARGET_P9_VECTOR was used.
21525 (splitter for loading small constants): Likewise.
21527 2017-07-26 Andrew Pinski <apinski@cavium.com>
21529 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
21532 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
21535 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
21536 (fp_valid_at): Likewise.
21538 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
21540 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
21541 (qdf24xx_addrcost_table): Likewise.
21542 (cortexa57_tunings): Update to use generic_branch_cost.
21543 (cortexa72_tunings): Likewise.
21544 (cortexa73_tunings): Likewise.
21545 (qdf24xx_tunings): Likewise.
21547 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
21549 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
21550 (thunderx2t99_branch_cost): Likewise.
21551 (cortexa35_tunings): Update to use generic_branch_cost.
21552 (cortexa53_tunings): Likewise.
21553 (cortexa57_tunings): Likewise.
21554 (cortexa72_tunings): Likewise.
21555 (cortexa73_tunings): Likewise.
21556 (thunderx2t99_tunings): Likewise.
21558 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
21560 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
21561 (sparc_option_override): Honour MASK_FSMULD.
21562 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
21563 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
21564 * config/sparc/sparc.opt (mfsmuld): New option.
21565 * doc/invoke.texi (mfsmuld): Document option.
21567 2017-07-26 Marek Polacek <polacek@redhat.com>
21569 PR middle-end/70992
21570 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
21572 2017-07-26 Richard Biener <rguenther@suse.de>
21574 * gimple-match-head.c (do_valueize): Return OP if valueize
21576 (get_def): New helper to get at the def stmt of a SSA name
21577 if valueize allows.
21578 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
21579 do_valueize to get at the def stmt.
21580 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
21582 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
21584 PR middle-end/46932
21585 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
21587 2017-07-26 Martin Liska <mliska@suse.cz>
21590 * function.c (expand_function_start): Make expansion of
21591 nonlocal_goto_save_area after parm_birth_insn.
21593 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
21595 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
21596 from all CPU target flags enable members.
21598 2017-07-26 Richard Biener <rguenther@suse.de>
21600 * genmatch.c (dt_simplify::gen): Make iterator vars const.
21601 (decision_tree::gen): Make 'type' const.
21602 (write_predicate): Likewise.
21604 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
21606 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
21607 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
21608 (rs6000_option_override_internal): Likewise.
21609 (rs6000_expand_vector_set): Likewise.
21610 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
21611 (TARGET_UPPER_REGS_SF): Likewise.
21612 (TARGET_UPPER_REGS_DI): Likewise.
21613 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
21614 (TARGET_DIRECT_MOVE_64BIT): Likewise.
21615 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
21616 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
21617 (Splitters for DI constants in Altivec registers): Likewise.
21618 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
21619 (vsx_set_v4sf_p9): Likewise.
21620 (vsx_set_v4sf_p9_zero): Likewise.
21621 (vsx_insert_extract_v4sf_p9): Likewise.
21622 (vsx_insert_extract_v4sf_p9_2): Likewise.
21624 2017-07-25 Carl Love <cel@us.ibm.com>
21626 * doc/extend.texi: Update the built-in documentation file for the
21627 existing built-in functions
21628 vector signed char vec_cnttz (vector signed char);
21629 vector unsigned char vec_cnttz (vector unsigned char);
21630 vector signed short vec_cnttz (vector signed short);
21631 vector unsigned short vec_cnttz (vector unsigned short);
21632 vector signed int vec_cnttz (vector signed int);
21633 vector unsigned int vec_cnttz (vector unsigned int);
21634 vector signed long long vec_cnttz (vector signed long long);
21635 vector unsigned long long vec_cnttz (vector unsigned long long);
21637 2017-07-25 Andrew Pinski <apinski@cavium.com>
21639 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
21640 accesses where the use is for the first operand of a BIT_INSERT.
21642 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
21645 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
21646 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
21648 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
21650 * config/i386/gstabs.h: Delete.
21651 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
21653 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
21655 * config/i386/i386.c (ix86_decompose_address): Do not check for
21656 register RTX when looking at index_reg or base_reg.
21657 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
21659 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
21661 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
21662 to update EH info here.
21664 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
21666 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
21668 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
21670 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
21672 2017-07-25 Torsten Duwe <duwe@suse.de>
21674 * common.opt: Introduce -fpatchable-function-entry
21675 command line option, and its variables function_entry_patch_area_size
21676 and function_entry_patch_area_start.
21677 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
21678 including a two-value parser.
21679 * target.def (print_patchable_function_entry): New target hook.
21680 * targhooks.h (default_print_patchable_function_entry): New function.
21681 * targhooks.c (default_print_patchable_function_entry): Likewise.
21682 * toplev.c (process_options): Switch off IPA-RA if
21683 patchable function entries are being generated.
21684 * varasm.c (assemble_start_function): Look at the
21685 patchable-function-entry command line switch and current
21686 function attributes and maybe generate NOP instructions by
21687 calling the print_patchable_function_entry hook.
21688 * doc/extend.texi: Document patchable_function_entry attribute.
21689 * doc/invoke.texi: Document -fpatchable_function_entry
21690 command line option.
21691 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
21693 * doc/tm.texi: Re-generate.
21695 2017-07-25 Jakub Jelinek <jakub@redhat.com>
21698 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
21699 TARGET_AVX512DQ rather than TARGET_AVX512BW.
21701 2017-07-25 Tamar Christina <tamar.christina@arm.com>
21703 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
21705 2017-07-25 Richard Biener <rguenther@suse.de>
21707 PR tree-optimization/81455
21708 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
21709 not walk in cycles when looking for guards.
21711 2017-07-25 Richard Biener <rguenther@suse.de>
21713 PR tree-optimization/81529
21714 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
21715 when optimizing backedge uses.
21717 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
21719 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
21721 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
21722 to dl_section_ref. On AIX, append an expression to subtract
21723 the size of the section length to dl_section_ref.
21725 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
21727 * configure.ac: If any of the config.* scripts fail, exit 1.
21728 * configure: Regenerate.
21730 2017-07-25 Richard Biener <rguenther@suse.de>
21732 PR middle-end/81546
21733 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
21734 of immediate uses, be more verbose on errors.
21736 2017-07-25 Richard Biener <rguenther@suse.de>
21738 PR tree-optimization/81510
21739 * tree-vect-loop.c (vect_is_simple_reduction): When the
21740 reduction stmt is not inside the loop bail out.
21742 2017-07-25 Richard Biener <rguenther@suse.de>
21744 PR tree-optimization/81303
21745 * tree-vect-loop-manip.c (vect_loop_versioning): Build
21746 profitability check against LOOP_VINFO_NITERSM1.
21748 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
21750 * domwalk.c (cmp_bb_postorder): Simplify.
21751 (sort_bbs_postorder): New function. Use it...
21752 (dom_walker::walk): ...here to optimize common cases.
21754 2017-07-25 Martin Liska <mliska@suse.cz>
21757 * ipa-visibility.c (function_and_variable_visibility): Make the
21758 redirection just on target that supports aliasing.
21759 Fix GNU coding style.
21761 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
21764 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
21766 * config/arm/rtems.h: Likewise.
21767 * config/bfin/rtems.h: Likewise.
21768 * config/i386/rtemself.h: Likewise.
21769 * config/lm32/rtems.h: Likewise.
21770 * config/m32c/rtems.h: Likewise.
21771 * config/m68k/rtemself.h: Likewise.
21772 * config/microblaze/rtems.h: Likewise.
21773 * config/mips/rtems.h: Likewise.
21774 * config/moxie/rtems.h: Likewise.
21775 * config/nios2/rtems.h: Likewise.
21776 * config/powerpcspe/rtems.h: Likewise.
21777 * config/rs6000/rtems.h: Likewise.
21778 * config/rtems.h: Likewise.
21779 * config/sh/rtems.h: Likewise.
21780 * config/sh/rtemself.h: Likewise.
21781 * config/sparc/rtemself.h: Likewise.
21783 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
21786 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
21787 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
21788 * tree-ssa-structalias.c (alias_get_name): Same.
21790 2017-07-25 Bin Cheng <bin.cheng@arm.com>
21793 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
21794 instructions if no du chain is found.
21796 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
21798 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
21800 2017-07-25 Richard Biener <rguenther@suse.de>
21802 PR middle-end/81505
21803 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
21806 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
21808 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
21809 upper-regs options.
21810 (ISA_2_7_MASKS_SERVER): Likewise.
21811 (ISA_3_0_MASKS_IEEE): Likewise.
21812 (OTHER_P8_VECTOR_MASKS): Likewise.
21813 (OTHER_VSX_VECTOR_MASKS): Likewise.
21814 (POWERPC_MASKS): Likewise.
21815 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
21816 duplicate list of options.
21817 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
21818 explicit -mupper-regs options.
21819 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
21820 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
21821 alias for -mupper-regs-df.
21822 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
21823 (rs6000_init_hard_regno_mode_ok): Likewise.
21824 (rs6000_option_override_internal): Likewise.
21825 (rs6000_opt_masks): Likewise.
21826 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
21827 options in terms of whether -mvsx or -mpower8-vector was used.
21828 (TARGET_UPPER_REGS_DI): Likewise.
21829 (TARGET_UPPER_REGS_SF): Likewise.
21830 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
21831 -mupper-regs-* options.
21833 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
21835 * passes.c (emergency_dump_function): Print some empty lines and a
21836 header before the RTL dump.
21838 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
21840 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
21842 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
21845 * config/aarch64/aarch64.c (aarch64_classify_symbol):
21846 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
21848 2017-07-24 Carl Love <cel@us.ibm.com>
21850 * config/rs6000/rs6000-c.c: Add support for built-in functions
21851 vector float vec_extract_fp32_from_shorth (vector unsigned short);
21852 vector float vec_extract_fp32_from_shortl (vector unsigned short);
21853 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
21854 vec_extract_fp_from_shortl): Add defines for the two builtins.
21855 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
21856 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
21858 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
21859 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
21860 * doc/extend.texi: Update the built-in documentation file for the
21861 new built-in function.
21863 2017-07-24 Jakub Jelinek <jakub@redhat.com>
21866 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
21868 * doc/generic.texi: Likewise.
21869 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
21870 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
21872 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
21874 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
21875 (aarch64_mls_elt_merge<mode>): Likewise.
21877 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
21879 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
21880 having __cxa_atexit.
21882 2017-07-23 Michael Collison <michael.collison@arm.com>
21884 * config/arm/arm.c (arm_option_override): Deprecate
21885 use of -mstructure-size-boundary.
21886 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
21887 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
21889 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21892 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
21893 Reduce cost estimate for direct moves.
21895 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
21898 * config/i386/i386.c (ix86_option_override_internal): Disable
21899 BMI, BMI2 and TBM instructions for -m16.
21901 2017-07-21 Carl Love <cel@us.ibm.com>
21903 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21904 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
21905 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
21906 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
21907 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
21908 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
21909 VMULOSW): New enum "unspec" values.
21910 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
21911 altivec_vmulosw): New patterns.
21912 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
21913 VMULOSW): Add definitions.
21915 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
21917 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
21918 (qdf24xx): Likewise.
21919 * config/aarch64/aarch64-options-extensions.def (rdma); New.
21920 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
21921 (AARCH64_FL_V8_1): Renumber.
21922 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
21923 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
21924 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
21925 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
21926 rdma to feature modifiers list.
21928 2017-07-21 Yury Gribov <tetra2005@gmail.com>
21930 PR middle-end/56727
21931 * ipa-visibility (function_and_variable_visibility): Convert
21932 recursive PLT call to direct call if appropriate.
21934 2017-07-21 Andrew Pinski <apinski@cavium.com>
21936 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
21937 operand 1 to see if the types precision matches.
21938 * fold-const.c (operand_equal_p): Likewise.
21940 2017-07-21 Richard Biener <rguenther@suse.de>
21942 PR tree-optimization/81303
21943 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
21944 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
21945 (vect_peeling_hash_get_lowest_cost): Adjust.
21946 (vect_enhance_data_refs_alignment): Likewise. Use
21947 vect_get_peeling_costs_all_drs to compute the penalty for no
21948 peeling to match up costs.
21950 2017-07-21 Richard Biener <rguenther@suse.de>
21952 PR tree-optimization/81500
21953 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
21954 we didn't identify a reduction path.
21956 2017-07-21 Tom de Vries <tom@codesourcery.com>
21957 Cesar Philippidis <cesar@codesourcery.com>
21959 PR gcov-profile/81442
21960 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
21963 2017-07-21 Tom de Vries <tom@codesourcery.com>
21966 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
21968 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
21969 nvptx_override_options_after_change.
21971 2017-07-21 Ulrich Drepper <drepper@redhat.com>
21973 * dwarf2out.c (output_file_names): Avoid double testing for
21974 dwarf_version >= 5.
21976 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
21978 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
21980 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
21982 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
21984 (try_crossjump_to_edge): Do not punt on partitioned functions.
21986 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
21988 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
21989 Put all BBs reachable only via paths crossing cold region to cold
21991 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
21993 2016-07-21 Richard Biener <rguenther@suse.de>
21995 PR tree-optimization/81303
21996 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
21997 into account prologue and epilogue iterations when raising
21998 min_profitable_iters to sth at least covering one vector iteration.
22000 2017-07-21 Tamar Christina <tamar.christina@arm.com>
22002 * config/arm/arm.c (arm_test_cpu_arch_dat):
22005 2017-07-20 Nathan Sidwell <nathan@acm.org>
22007 Remove TYPE_METHODS.
22008 * tree.h (TYPE_METHODS): Delete.
22009 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
22010 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
22011 (dbxout_type_methods): Scan TYPE_FIELDS.
22012 (dbxout_type): Don't check TYPE_METHODS here.
22013 * function.c (use_register_for_decl): Always ignore register for
22014 class types when not optimizing.
22015 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
22016 * tree.c (free_lang_data_in_type): Stitch out member functions and
22017 templates from TYPE_FIELDS.
22018 (build_distinct_type_copy, verify_type_variant,
22019 verify_type): Member fns are on TYPE_FIELDS.
22020 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
22021 * tree-pretty-print.c (dump_generic_node): Likewise.
22023 2017-07-20 Jakub Jelinek <jakub@redhat.com>
22026 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
22027 V2TImode and V4TImode.
22028 (ix86_expand_vector_extract): Likewise.
22029 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
22030 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
22031 (ssescalarmode): Handle V4TImode and V2TImode.
22032 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
22033 (*vec_extractv2ti, *vec_extractv4ti): New insns.
22034 (VEXTRACTI128_MODE): New mode iterator.
22035 (splitter for *vec_extractv?ti first element): New.
22036 (VEC_INIT_MODE): New mode iterator.
22037 (vec_init<mode>): Consolidate 3 expanders into one using
22038 VEC_INIT_MODE mode iterator.
22040 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
22042 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
22043 non_spilled_static_chain_regno_p.
22045 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
22047 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
22049 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
22051 * bb-reorder.c (connect_traces): Allow copying of blocks within
22054 2017-07-20 Richard Biener <rguenther@suse.de>
22056 * gimple.h (gimple_phi_result): Add gphi * overload.
22057 (gimple_phi_result_ptr): Likewise.
22058 (gimple_phi_arg): Likewise. Adjust index assert to only
22059 allow actual argument accesses rather than all slots available
22061 (gimple_phi_arg_def): Add gphi * overload.
22062 * tree-phinodes.c (make_phi_node): Initialize only actual
22064 (resize_phi_node): Clear memory not covered by old node,
22065 do not initialize excess argument slots.
22066 (reserve_phi_args_for_new_edge): Initialize new argument slot
22069 2017-07-20 Bin Cheng <bin.cheng@arm.com>
22071 PR tree-optimization/81388
22073 2016-07-21 Bin Cheng <bin.cheng@arm.com>
22075 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
22076 by removing computation of may_be_zero.
22078 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
22079 Tom de Vries <tom@codesourcery.com>
22081 PR middle-end/81030
22082 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
22083 when gimple level profile disagrees with what RTL expander did.
22085 2017-07-20 Richard Biener <rguenther@suse.de>
22087 PR tree-optimization/61171
22088 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
22089 (vect_analyze_stmt): Add slp instance parameter.
22090 (vectorizable_reduction): Likewise.
22091 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
22092 (vect_is_simple_reduction): Deal with chains not detected
22093 as SLP reduction chain, specifically not properly associated
22094 chains containing a mix of plus/minus.
22095 (get_reduction_op): Remove.
22096 (get_initial_defs_for_reduction): Simplify, pass in whether
22097 this is a reduction chain, pass in the SLP node for the PHIs.
22098 (vect_create_epilog_for_reduction): Get the SLP instance as
22100 (vectorizable_reduction): Get the SLP instance as arg.
22101 During analysis remember the SLP node with the PHIs in the
22102 instance. Simplify getting at the vectorized reduction PHIs.
22103 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
22104 through SLP instance.
22105 (vect_slp_analyze_operations): Likewise.
22106 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
22107 (vect_transform_stmt): Likewise.
22109 2017-07-20 Tom de Vries <tom@codesourcery.com>
22111 PR tree-optimization/81489
22112 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
22113 read of phi arg location to before loop that modifies phi.
22115 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22117 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
22120 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
22122 PR middle-end/81331
22123 * except.c (execute): Fix ordering issue.
22125 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
22127 PR rtl-optimization/81423
22128 * combine.c (make_compound_operation_int): Don't try to optimize
22129 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
22131 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
22133 PR rtl-optimization/81423
22134 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
22135 with a constant that is -1 in the truncated to mode.
22137 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
22139 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
22140 (determine_unlikely_bbs): ... here.
22141 * predict.h (propagate_unlikely_bbs_forward): Declare.
22142 * cfgexpand.c (pass_expand::execute): Use it.
22143 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
22145 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
22146 propagate_unlikely_bbs_forward.
22148 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
22150 PR middle-end/81331
22151 * except.c (maybe_add_nop_after_section_switch): New function.
22154 2017-07-19 Tom de Vries <tom@codesourcery.com>
22156 * gimple.h (gimple_phi_set_arg): Make assert more strict.
22158 2017-07-19 Tom de Vries <tom@codesourcery.com>
22160 * gimple.h (gimple_phi_arg): Make assert more strict.
22162 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
22164 * config.gcc (powerpc*-*-*): Add mmintrin.h.
22165 * config/rs6000/mmintrin.h: New file.
22166 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
22168 2017-07-19 Jakub Jelinek <jakub@redhat.com>
22170 PR tree-optimization/81346
22171 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
22173 2017-07-19 Tom de Vries <tom@codesourcery.com>
22175 * config/nvptx/nvptx.md (VECIM): Add V2DI.
22177 2017-07-19 Tom de Vries <tom@codesourcery.com>
22179 * config/nvptx/nvptx-modes.def: Add V2DImode.
22180 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
22181 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
22182 (nvptx_output_mov_insn): Handle lack of mov.b128.
22183 (nvptx_print_operand): Handle 'H' and 'L' codes.
22184 (nvptx_vector_mode_supported): Allow V2DImode.
22185 (nvptx_preferred_simd_mode): New function.
22186 (nvptx_data_alignment): New function.
22187 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
22188 nvptx_preferred_simd_mode.
22189 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
22191 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
22193 2017-07-19 Tom de Vries <tom@codesourcery.com>
22195 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
22196 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
22197 (nvptx_vector_mode_supported): New function. Allow V2SImode.
22198 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
22199 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
22200 (mov<VECIM>_insn): New define_insn.
22201 (define_expand "mov<VECIM>): New define_expand.
22203 2017-07-19 Tom de Vries <tom@codesourcery.com>
22205 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
22207 2017-07-19 Jakub Jelinek <jakub@redhat.com>
22209 PR tree-optimization/81346
22210 * fold-const.h (fold_div_compare, range_check_type): Declare.
22211 * fold-const.c (range_check_type): New function.
22212 (build_range_check): Use range_check_type.
22213 (fold_div_compare): No longer static, rewritten into
22214 a match.pd helper function.
22215 (fold_comparison): Don't call fold_div_compare here.
22216 * match.pd (X / C1 op C2): New optimization using fold_div_compare
22217 as helper function.
22219 2017-07-19 Nathan Sidwell <nathan@acm.org>
22221 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
22222 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
22223 * tree.c (find_decls_types_r, verify_type): Use
22224 TYPE_{MIN,MAX}_VALUE_RAW.
22225 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
22226 (hash_tree): Likewise.
22227 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
22229 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
22232 2017-07-18 Tom de Vries <tom@codesourcery.com>
22234 PR middle-end/81464
22235 * omp-expand.c (expand_omp_for_static_chunk): Handle
22236 equal-argument loop exit phi.
22238 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
22241 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
22242 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
22243 operand 2 predicate.
22244 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
22245 operand 2 predicate.
22246 (ror,rol -> rorx splitters): Use const_int_operand as
22247 operand 2 predicate.
22249 2017-06-18 Richard Biener <rguenther@suse.de>
22251 PR tree-optimization/81410
22252 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
22253 the gap in the ! slp_perm SLP case after each group.
22255 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
22257 PR middle-end/81463
22258 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
22261 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
22263 PR middle-end/81462
22264 * predict.c (set_even_probabilities): Cleanup; do not affect
22265 probabilities that are already known.
22266 (combine_predictions_for_bb): Call even when count is set.
22268 2017-07-18 Nathan Sidwell <nathan@acm.org>
22270 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
22273 2017-07-18 Bin Cheng <bin.cheng@arm.com>
22276 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
22277 optimization for loop niter analysis.
22279 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
22282 * config/avr/avr.c (avr_optimize_casesi): Don't use
22283 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
22285 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
22287 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
22288 body_cost_vec from _vect_peel_extended_info.
22289 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
22290 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
22293 2017-07-18 Bin Cheng <bin.cheng@arm.com>
22295 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
22297 2017-07-18 Richard Biener <rguenther@suse.de>
22299 PR tree-optimization/80620
22300 PR tree-optimization/81403
22301 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
22302 info when re-using a VN table entry.
22304 2017-07-18 Richard Biener <rguenther@suse.de>
22306 PR tree-optimization/81418
22307 * tree-vect-loop.c (vectorizable_reduction): Properly compute
22308 vectype_in. Verify that with lane-reducing reduction operations
22309 we have a single def-use cycle.
22311 2017-07-17 Carl Love <cel@us.ibm.com>
22313 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
22315 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22316 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
22317 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
22318 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
22319 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
22320 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
22321 VMULOSW): New enum "unspec" values.
22322 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
22323 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
22324 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
22325 altivec_vmulosw): New patterns.
22326 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
22327 VMULOSW): Add definitions.
22329 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
22331 * config/alpha/alpha.c: Include predict.h.
22333 2017-07-17 Yury Gribov <tetra2005@gmail.com>
22335 * tree-vrp.c (compare_assert_loc): Fix comparison function
22336 to return predictable results.
22338 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
22340 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
22342 (subdi3): Likewise.
22343 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
22344 * doc/invoke.texi (mexpand-adddi): Update text.
22346 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
22348 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
22349 that also clobbers the CC register. The old expand code is moved
22351 (*arc_clzsi2): ... here.
22352 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
22353 the CC register. The old expand code is moved to ...
22354 (arc_ctzsi2): ... here.
22356 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
22358 * config/arc/arc.opt (mindexed-loads): Use initial value
22359 TARGET_INDEXED_LOADS_DEFAULT.
22360 (mauto-modify-reg): Use initial value
22361 TARGET_AUTO_MODIFY_REG_DEFAULT.
22362 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
22363 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
22364 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
22365 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
22367 2017-07-17 Martin Liska <mliska@suse.cz>
22370 * opts.c (finish_options): Do not allow -fgnu-tm
22371 w/ -fsanitize={kernel-,}address. Say sorry.
22373 2017-07-17 Bin Cheng <bin.cheng@arm.com>
22376 * tree-loop-distribution.c (classify_partition): Only assert on
22377 numer of iterations.
22378 (merge_dep_scc_partitions): Delete prameter. Update function call.
22379 (distribute_loop): Remove code handling loop with unknown niters.
22380 (pass_loop_distribution::execute): Skip loop with unknown niters.
22382 2017-07-17 Bin Cheng <bin.cheng@arm.com>
22385 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
22386 function sort_partitions_by_post_order.
22388 2017-07-17 Bin Cheng <bin.cheng@arm.com>
22390 PR tree-optimization/81374
22391 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
22392 the max index of basic blocks, rather than number of basic blocks.
22394 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
22396 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
22398 (arc_legitimate_pic_operand_p): Likewise.
22399 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
22401 (arc_needs_pcl_p): Likewise.
22402 (arc_legitimate_pc_offset_p): Likewise.
22403 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
22404 function is also used in constrains.md.
22405 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
22406 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
22407 PLUS. Only return true/false in known cases, otherwise assert.
22408 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
22409 is already called in arc_legitimate_constant_p.
22410 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
22412 (LEGITIMATE_PIC_OPERAND_P): Use
22413 arc_raw_symbolic_reference_mentioned_p function.
22414 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
22419 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
22420 Andrew Burgess <andrew.burgess@embecosm.com>
22422 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
22423 (arc_return_address_register): New function.
22424 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
22425 (arc_handle_fndecl_attribute): Add naked attribute.
22426 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
22427 (TARGET_WARN_FUNC_RETURN): Likewise.
22428 (arc_allocate_stack_slots_for_args): New function.
22429 (arc_warn_func_return): Likewise.
22430 (machine_function): Change type fn_type.
22431 (arc_compute_function_type): Consider new naked function type,
22432 change function return type.
22433 (arc_must_save_register): Adapt to handle new
22434 arc_compute_function_type's return type.
22435 (arc_expand_prologue): Likewise.
22436 (arc_expand_epilogue): Likewise.
22437 (arc_return_address_regs): Delete.
22438 (arc_return_address_register): New function.
22439 (arc_epilogue_uses): Use above function.
22440 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
22441 (arc_function_type): Change encoding, add naked type.
22442 (ARC_INTERRUPT_P): Change to handle the new encoding.
22443 (ARC_FAST_INTERRUPT_P): Likewise.
22444 (ARC_NORMAL_P): Define.
22445 (ARC_NAKED_P): Likewise.
22446 (arc_compute_function_type): Delete prototype.
22447 * config/arc/arc.md (in_ret_delay_slot): Use
22448 arc_return_address_register function.
22449 (simple_return): Likewise.
22450 (p_return_i): Likewise.
22452 2017-07-17 Jakub Jelinek <jakub@redhat.com>
22454 PR tree-optimization/81428
22455 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
22456 can't be built for those types.
22458 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
22460 Remove stuff dead since r239246.
22462 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
22463 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
22464 (avr_inform_devices): Remove dead stuff.
22466 2017-07-17 Tamar Christina <tamar.christina@arm.com>
22468 * config/arm/arm_neon.h: Fix softp typo.
22470 2017-07-17 Jakub Jelinek <jakub@redhat.com>
22472 PR tree-optimization/81365
22473 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
22474 aggregate moves onto bb predecessor edges, make sure there are no
22475 loads that could alias the lhs in between the start of bb and the
22478 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
22481 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
22482 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
22483 [LSHIFTRT, outer_code = TRUNCATE]: Same.
22485 2017-07-17 Jakub Jelinek <jakub@redhat.com>
22487 PR tree-optimization/81396
22488 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
22489 (init_symbolic_number): Initialize it to 1.
22490 (perform_symbolic_merge): Add n_ops from both operands into the new
22492 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
22493 without base_addr as useless if they need more than one operation.
22494 (bswap_replace): Handle !bswap case for NULL base_addr.
22496 2017-07-17 Tom de Vries <tom@codesourcery.com>
22499 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
22502 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
22504 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
22505 conditional builtin define __FIX_LEON3FT_B2BST.
22507 2017-07-17 Daniel Cederman <cederman@gaisler.com>
22509 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
22510 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
22513 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
22515 PR rtl-optimization/81424
22516 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
22517 to remove potential trapping from operands if -fnon-call-exceptions.
22519 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
22521 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
22522 profile_proability for scalling.
22523 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
22525 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
22527 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
22529 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
22531 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
22532 fixpoint arithmetics.
22534 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
22536 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
22537 fixpoint arithmetics.
22539 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
22541 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
22542 fixpoint arithmetics.
22544 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
22546 * profile-count.h (profile_probability::from_reg_br_prob_note,
22547 profile_probability::to_reg_br_prob_note): New functions.
22548 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
22549 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
22550 * predict.c (probability_reliable_p): Update.
22551 (edge_probability_reliable_p): Update.
22552 (br_prob_note_reliable_p): Update.
22553 (invert_br_probabilities): Update.
22554 (add_reg_br_prob_note): New function.
22555 (combine_predictions_for_insn): Update.
22556 * asan.c (asan_clear_shadow): Update.
22557 * cfgbuild.c (compute_outgoing_frequencies): Update.
22558 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
22559 (update_br_prob_note): Update.
22560 (rtl_verify_edges): Update.
22561 (purge_dead_edges): Update.
22562 (fixup_reorder_chain): Update.
22563 * emit-rtl.c (try_split): Update.
22564 * ifcvt.c (cond_exec_process_insns): Update.
22565 (cond_exec_process_if_block): Update.
22566 (dead_or_predicable): Update.
22567 * internal-fn.c (expand_addsub_overflow): Update.
22568 (expand_neg_overflow): Update.
22569 (expand_mul_overflow): Update.
22570 * loop-doloop.c (doloop_modify): Update.
22571 * loop-unroll.c (compare_and_jump_seq): Update.
22572 * optabs.c (emit_cmp_and_jump_insn_1): Update.
22573 * predict.h: Update.
22574 * reorg.c (mostly_true_jump): Update.
22576 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
22577 * config/alpha/alpha.c (emit_unlikely_jump): Update.
22578 * config/arc/arc.c: (emit_unlikely_jump): Update.
22579 * config/arm/arm.c: (emit_unlikely_jump): Update.
22580 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
22581 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
22582 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
22583 (ix86_print_operand): Update.
22584 (ix86_split_fp_branch): Update.
22585 (predict_jump): Update.
22586 * config/ia64/ia64.c (ia64_print_operand): Update.
22587 * config/mmix/mmix.c (mmix_print_operand): Update.
22588 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
22589 (rs6000_expand_split_stack_prologue): Update.
22590 * config/rs6000/rs6000.c: Update.
22591 * config/s390/s390.c (s390_expand_vec_strlen): Update.
22592 (s390_expand_vec_movstr): Update.
22593 (s390_expand_cs_tdsi): Update.
22594 (s390_expand_split_stack_prologue): Update.
22595 * config/sh/sh.c (sh_print_operand): Update.
22596 (expand_cbranchsi4): Update.
22597 (expand_cbranchdi4): Update.
22598 * config/sparc/sparc.c (output_v9branch): Update.
22599 * config/spu/spu.c (get_branch_target): Update.
22600 (ea_load_store_inline): Update.
22601 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
22602 * config/tilepro/tilepro.c: Update.
22604 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
22606 * gimplify.c (mostly_copy_tree_r): Revert latest change.
22607 (gimplify_save_expr): Likewise.
22609 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
22611 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
22613 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
22615 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
22617 * timevar.def (TV_IPA_FNSUMMARY): Define.
22619 2017-07-16 Daniel Cederman <cederman@gaisler.com>
22621 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
22622 to back store errata sensitive sequence from being generated.
22623 (sqrtdf2_fix): Likewise.
22625 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
22627 * tree-ssa-threadupdate.c (compute_path_counts,
22628 update_joiner_offpath_counts): Use profile_probability.
22630 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
22633 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
22635 * config/arm/arm-c.c (arm_cpu_builtins): Define
22636 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
22638 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
22640 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22641 array entries to represent __ieee128 versions of the
22642 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
22643 scalar_extract_sig, and scalar_insert_exp built-in functions.
22644 (altivec_resolve_overloaded_builtin): Add special case handling
22645 for the __builtin_scalar_insert_exp function, as represented by
22646 the P9V_BUILTIN_VEC_VSIEDP constant.
22647 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
22648 exponent support for __ieee128 argument.
22649 (VSESQP): Add scalar extract signature support for __ieee128
22651 (VSTDCNQP): Add scalar test negative support for __ieee128
22653 (VSIEQP): Add scalar insert exponent support for __int128 argument
22654 with __ieee128 result.
22655 (VSIEQPF): Add scalar insert exponent support for __ieee128
22656 argument with __ieee128 result.
22657 (VSTDCQP): Add scalar test data class support for __ieee128
22659 (VSTDCNQP): Add overload support for scalar test negative with
22660 __ieee128 argument.
22661 (VSTDCQP): Add overload support for scalar test data class
22662 __ieee128 argument.
22663 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
22664 UNSPEC_VSX_SXSIGDP.
22665 (UNSPEC_VSX_SIEXPQP): New constant.
22666 (xsxexpqp): New insn for VSX scalar extract exponent quad
22668 (xsxsigqp): New insn for VSX scalar extract significand quad
22670 (xsiexpqpf): New insn for VSX scalar insert exponent quad
22671 precision with floating point argument.
22672 (xststdcqp): New expand for VSX scalar test data class quad
22674 (xststdcnegqp): New expand for VSX scalar test negative quad
22676 (xststdcqp): New insn to match expansions for VSX scalar test data
22677 class quad precision and VSX scalar test negative quad precision.
22678 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
22679 special case operand checking to enforce that second operand of
22680 VSX scalar test data class with quad precision argument is a 7-bit
22682 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
22683 prototypes and descriptions of __ieee128 versions of
22684 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
22685 scalar_test_data_class, and scalar_test_neg built-in functions.
22687 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22689 PR tree-optimization/81162
22690 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
22691 replace a negate with an add.
22693 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
22695 * doc/invoke.texi (arm/-mcpu): Document +crypto.
22697 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
22699 * config/arm/arm-c.c (arm_cpu_builtins): Define
22700 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
22702 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
22704 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
22705 (armv8-r): Set ARM Cortex-R52 as default CPU.
22706 * config/arm/arm-tables.opt: Regenerate.
22707 * config/arm/arm-tune.md: Regenerate.
22708 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
22710 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
22711 extension for -mcpu=cortex-r52.
22713 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
22715 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
22716 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
22717 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
22718 (fp-armv8): Define it as FP_ARMv8 only.
22719 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
22720 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
22722 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
22723 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
22724 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
22725 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
22726 than TARGET_FPU_ARMV8.
22727 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
22728 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
22729 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
22731 * config/arm/neon.md (neon_vrint): Likewise.
22732 (neon_vcvt): Likewise.
22733 (neon_<fmaxmin_op><mode>): Likewise.
22734 (<fmaxmin><mode>3): Likewise.
22735 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
22736 * config/arm/predicates.md (arm_cond_move_operator): Check against
22737 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
22739 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
22741 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
22742 to top of function.
22744 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22746 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
22747 loop in comment with memset.
22749 2017-07-14 Martin Liska <mliska@suse.cz>
22751 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
22752 * dwarf2out.c (is_java): Remove the function.
22753 (output_pubname): Remove usage of the function.
22754 (lower_bound_default): Remove usage of DW_LANG_Java.
22755 (gen_compile_unit_die): Likewise.
22756 * gcc.c: Remove compiler defaults for .java and .zip files.
22757 * gimple-expr.c (remove_suffix): Change as there's no longer
22758 extension than 4-letter one.
22759 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
22760 (gimplify_save_expr): Likewise.
22761 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
22762 as it's possible even for other languages than Java.
22763 * langhooks.h (struct lang_hooks): Remove Java from a comment.
22764 * lto-opts.c (lto_write_options): Remove reference to Java.
22765 * opts.c (strip_off_ending): Update file extension handling.
22766 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
22767 * tree-eh.c (lower_resx): Likewise.
22768 * tree.c (free_lang_data_in_type): Remove dead code.
22769 (find_decls_types_r): Likewise.
22770 (build_common_builtin_nodes): Remove Java from a comment.
22771 (verify_type): Remove dead code.
22772 * varasm.c (assemble_external): Remove Java from a comment.
22774 2017-07-14 Martin Liska <mliska@suse.cz>
22776 * opts.c (finish_options): Add quotes.
22777 (common_handle_option): Likewise.
22779 2017-07-14 Martin Liska <mliska@suse.cz>
22781 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
22782 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
22783 Remove N_SO_PASCAL.
22784 * dwarf2out.c (lower_bound_default): Do not handle
22786 (gen_compile_unit_die): Likewise.
22787 * gcc.c: Remove default extension binding for GNU Pascal.
22788 * stmt.c: Remove Pascal language from a comment.
22789 * xcoffout.c: Likewise.
22791 2017-07-13 David Malcolm <dmalcolm@redhat.com>
22794 * diagnostic-show-locus.c (fixit_cmp): New function.
22795 (layout::layout): Sort m_fixit_hints.
22796 (column_range::column_range): Assert that the values are valid.
22797 (struct char_span): New struct.
22798 (correction::overwrite): New method.
22799 (struct source_line): New struct.
22800 (line_corrections::add_hint): Add assertions. Reimplement memcpy
22801 calls in terms of classes source_line and char_span, and
22802 correction::overwrite.
22803 (selftest::test_overlapped_fixit_printing_2): New function.
22804 (selftest::diagnostic_show_locus_c_tests): Call it.
22806 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
22808 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
22809 early if there is no lhs.
22811 2017-07-13 Martin Liska <mliska@suse.cz>
22813 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
22814 (gen_reference_type_die): Likewise.
22815 * stor-layout.c: Remove Pascal-related comment.
22817 2017-07-13 Martin Liska <mliska@suse.cz>
22819 * opts.c (finish_options): Add quotes to error messages.
22820 (parse_sanitizer_options): Likewise.
22822 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
22824 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
22826 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
22828 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
22830 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
22832 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
22834 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
22836 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
22839 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
22840 provides the hardware capability bits, define the macro
22841 __BUILTIN_CPU_SUPPORTS__.
22842 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
22843 if GLIBC does not provide the hardware capability bits. Add a
22844 gcc_unreachable call if the built-in cpu function is neither
22845 __builtin_cpu_is nor __builtin_cpu_supports.
22846 (rs6000_get_function_versions_dispatcher): Change the warning
22847 that an old GLIBC is used which does not export the capability
22848 bits to be an error.
22849 * doc/extend.texi (target_clones attribute): Document the
22850 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
22851 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
22852 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
22853 the macros defined by GCC if the newer GLIBC is available.
22855 2017-07-12 Jeff Law <law@redhat.com>
22857 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
22858 remaining includes slightly.
22859 * config/riscv/riscv-builtins.c: Include profile-count.h.
22861 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
22864 * config/avr/avr.c (avr_set_current_function): In diagnostic
22865 messages: Quote keywords and (parts of) identifiers.
22866 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
22869 2017-07-12 Carl Love <cel@us.ibm.com>
22871 * config/rs6000/rs6000-c.c: Add support for built-in functions
22872 vector bool char vec_revb (vector bool char);
22873 vector bool short vec_revb (vector short char);
22874 vector bool int vec_revb (vector bool int);
22875 vector bool long long vec_revb (vector bool long long);
22876 * doc/extend.texi: Update the built-in documentation file for the
22877 new built-in functions.
22879 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22881 * config/s390/s390.md: Remove movcc splitter.
22883 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22885 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
22886 load/store on condition.
22888 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
22891 * config/avr/avr.c (avr_encode_section_info)
22892 [progmem && !TREE_READONLY]: Error if progmem object needs
22895 2017-07-11 Michael Collison <michael.collison@arm.com>
22897 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
22900 2017-07-11 Carl Love <cel@us.ibm.com>
22902 * config/rs6000/rs6000-c.c: Add support for builtins
22903 vector unsigned int vec_parity_lsbb (vector signed int);
22904 vector unsigned int vec_parity_lsbb (vector unsigned int);
22905 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
22906 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
22907 vector unsigned long long vec_parity_lsbb (vector signed long long);
22908 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
22909 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
22910 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
22911 * doc/extend.texi: Update the built-in documentation file for the
22912 new built-in functions.
22914 2017-07-11 David Malcolm <dmalcolm@redhat.com>
22916 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
22917 (layout::m_primary_loc): New field.
22918 (layout::layout): Initialize new field. Move location filtering
22919 logic from here to...
22920 (layout::maybe_add_location_range): ...this new method. Add
22921 support for filtering to just the lines already specified by other
22923 (layout::will_show_line_p): New method.
22924 (gcc_rich_location::add_location_if_nearby): New method.
22925 (selftest::test_add_location_if_nearby): New test function.
22926 (selftest::diagnostic_show_locus_c_tests): Call it.
22927 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
22930 2017-07-11 Tom de Vries <tom@codesourcery.com>
22932 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
22933 (bb_first_real_insn): New function.
22934 (nvptx_single): Add extra initialization of broadcasted condition
22937 2017-07-11 Nathan Sidwell <nathan@acm.org>
22939 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
22941 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
22943 * doc/extend.texi (AVR Function Attributes): Remove weblink to
22944 Binutils doc as TEXI will mess them up.
22945 * doc/invoke.texi (AVR Options): Same here.
22947 2017-07-11 Daniel Cederman <cederman@gaisler.com>
22949 * config/sparc/sparc.opt (mfix-ut700): New option.
22950 (mfix-gr712rc): Likewise.
22951 (sparc_fix_b2bst): New variable.
22952 * doc/invoke.texi (SPARC options): Document them.
22953 (ARM options): Fix warnings.
22954 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
22955 instructions to prevent sequences that can trigger the store-store
22956 errata for certain LEON3FT processors.
22957 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
22958 (sparc_option_override): Set sparc_fix_b2bst appropriately.
22959 * config/sparc/sparc.md (fix_b2bst): New attribute.
22960 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
22962 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
22965 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
22967 (*rsqrtsf2_sse): Ditto.
22969 (div<mode>3): Macroize insn from divdf3 and divsf3
22970 using MODEF mode iterator.
22972 2017-07-10 Martin Sebor <msebor@redhat.com>
22974 PR tree-optimization/80397
22975 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
22976 instead of testing for equality to INTEGER_TYPE.
22978 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
22980 * config.gcc: Remove uclibc from arc target spec.
22982 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
22984 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
22986 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
22989 * lto-wrapper.c (remove_option): New function.
22990 (merge_and_complain): Merge PIC/PIE options more realistically.
22992 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
22994 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
22998 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
22999 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
23000 * config.in: Regenerate.
23001 * configure: Regenerate.
23002 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
23003 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
23004 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
23005 (TARGET_GASISR_PROLOGUES): ...target mask.
23006 * common/config/avr/avr-common.c
23007 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
23008 Set -mgas-isr-prologues.
23009 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
23010 INSERT_PASS_BEFORE for it.
23011 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
23012 * config/avr/avr.c (avr_option_override)
23013 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
23014 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
23015 (avr_attribute_table) <no_gccisr>: Add new function attribute.
23016 (avr_set_current_function) <is_no_gccisr>: Init machine field.
23017 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
23019 (make_avr_pass_pre_proep): New function.
23020 (emit_push_sfr) <treg>: Add argument to function and use it
23021 instead of TMP_REG.
23022 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
23023 and set machine->gasisr.yes.
23024 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
23025 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
23026 __gcc_isr.n_pushed to .L__stack_usage.
23027 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
23028 (avr_asm_final_postscan_insn): ...this new static function.
23029 * config/avr/avr.h (machine_function)
23030 <is_no_gccisr, use_L__stack_usage>: New fields.
23031 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
23032 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
23033 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
23034 (gasisr, *gasisr): New expander and insn.
23035 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
23036 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
23037 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
23039 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
23041 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
23044 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
23046 Move jump-tables out of .text again.
23049 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
23050 (ASM_OUTPUT_ADDR_VEC): New function.
23051 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
23052 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
23053 INSN_ADDRESSes as asm comment.
23054 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
23055 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
23056 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
23057 * config/avr/avr.md (*tablejump): Adjust comment.
23058 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
23059 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
23061 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
23062 (avr_output_addr_vec): New proto.
23063 (avr_log_t) <insn_addresses>: New field.
23065 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
23068 * config/i386/i386.c (ix86_function_arg_advance): Set
23069 outgoing_args_on_stack to true if there are outgoing arguments
23071 (ix86_function_arg): Likewise.
23072 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
23073 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
23074 * config/i386/i386.h (machine_function): Add
23075 outgoing_args_on_stack.
23077 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
23079 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
23080 supporting pthreds.
23081 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
23083 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
23085 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
23086 (REAL_H): Remove $(MACHMODE_H).
23087 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
23089 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
23090 $(MACHMODE_H) and double-int.h.
23091 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
23093 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
23096 2017-07-07 Andrew Pinski <apinski@cavium.com>
23098 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
23099 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
23101 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
23103 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
23104 Add warning if GCC was not configured to link against a GLIBC that
23105 exports the hardware capability bits.
23106 (make_resolver_func): Make resolver function private and not a
23107 COMDAT function. Create the name with clone_function_name instead
23108 of make_unique_name.
23111 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
23112 correct operand in doing the split.
23114 2017-07-07 Carl Love <cel@us.ibm.com>
23116 * config/rs6000/rs6000-c: Add support for built-in function
23117 vector unsigned short vec_pack_to_short_fp32 (vector float,
23119 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
23120 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
23121 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
23122 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
23123 (convert_4f32_8i16): Add define_expand.
23124 * doc/extend.texi: Update the built-in documentation file for the
23125 new built-in function.
23127 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23129 * config/sparc/m8.md: New file.
23130 * config/sparc/sparc.md: Include m8.md.
23132 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23134 * config/sparc/sparc.opt: New option -mvis4b.
23135 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
23136 (sparc_option_override): Handle VIS4B.
23137 (enum sparc_builtins): Define
23138 SPARC_BUILTIN_DICTUNPACK{8,16,32},
23139 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
23140 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
23141 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
23142 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
23143 (check_constant_argument): New function.
23144 (sparc_vis_init_builtins): Define builtins
23145 __builtin_vis_dictunpack{8,16,32},
23146 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
23147 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
23148 __builtin_vis_fpcmpde{8,16,32}shl and
23149 __builtin_vis_fpcmpur{8,16,32}shl.
23150 (sparc_expand_builtin): Check that the constant operands to
23151 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
23152 constant and in range.
23153 * config/sparc/sparc-c.c (sparc_target_macros): Handle
23155 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
23156 (SPARC_IMM5_P): Likewise.
23157 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
23158 (enabled): Handle vis4b.
23159 (UNSPEC_DICTUNPACK): New unspec.
23160 (UNSPEC_FPCMPSHL): Likewise.
23161 (UNSPEC_FPUCMPSHL): Likewise.
23162 (UNSPEC_FPCMPDESHL): Likewise.
23163 (UNSPEC_FPCMPURSHL): Likewise.
23164 (cpu_feature): New CPU feature `vis4b'.
23165 (dictunpack{8,16,32}): New insns.
23166 (FPCSMODE): New mode iterator.
23167 (fpcscond): New code iterator.
23168 (fpcsucond): Likewise.
23169 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
23170 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
23171 (fpcmpde{8,16,32}{si,di}shl): Likewise.
23172 (fpcmpur{8,16,32}{si,di}shl): Likewise.
23173 * config/sparc/constraints.md: Define constraints `q' for unsigned
23174 2-bit integer constants and `t' for unsigned 5-bit integer
23176 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
23178 (imm5_operand_dictunpack16): Likewise.
23179 (imm5_operand_dictunpack32): Likewise.
23180 (imm2_operand): Likewise.
23181 * doc/invoke.texi (SPARC Options): Document -mvis4b.
23182 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
23183 ditunpack* and fpcmp*shl builtins.
23185 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23187 * config.gcc: Handle m8 in --with-{cpu,tune} options.
23188 * config.in: Add HAVE_AS_SPARC6 define.
23189 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
23191 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
23193 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
23194 (CPP_CPU_SPEC): Handle m8.
23195 (ASM_CPU_SPEC): Likewise.
23196 * config/sparc/sparc-opts.h (enum processor_type): Add
23198 * config/sparc/sparc.c (m8_costs): New struct.
23199 (sparc_option_override): Handle TARGET_CPU_m8.
23200 (sparc32_initialize_trampoline): Likewise.
23201 (sparc64_initialize_trampoline): Likewise.
23202 (sparc_issue_rate): Likewise.
23203 (sparc_register_move_cost): Likewise.
23204 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
23205 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
23206 (ASM_CPU64_DEFAULT_SPEC): Likewise.
23207 (CPP_CPU_SPEC): Handle M8.
23208 (ASM_CPU_SPEC): Likewise.
23209 (AS_M8_FLAG): Define.
23210 * config/sparc/sparc.md: Add m8 to the cpu attribute.
23211 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
23212 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
23214 * configure: Regenerate.
23215 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
23218 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23220 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
23222 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
23223 ("*movdi_insn_sp32"): Do not set v3pipe.
23224 ("*movsi_insn"): Likewise.
23225 ("*movdi_insn_sp64"): Likewise.
23226 ("*movsf_insn"): Likewise.
23227 ("*movdf_insn_sp32"): Likewise.
23228 ("*movdf_insn_sp64"): Likewise.
23229 ("*zero_extendsidi2_insn_sp64"): Likewise.
23230 ("*sign_extendsidi2_insn"): Likewise.
23231 ("*mov<VM32:mode>_insn"): Likewise.
23232 ("*mov<VM64:mode>_insn_sp64"): Likewise.
23233 ("*mov<VM64:mode>_insn_sp32"): Likewise.
23234 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
23235 ("<vlop:code><VL:mode>3"): Likewise.
23236 ("*not_<vlop:code><VL:mode>3"): Likewise.
23237 ("*nand<VL:mode>_vis"): Likewise.
23238 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
23239 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
23240 ("one_cmpl<VL:mode>2"): Likewise.
23241 ("faligndata<VM64:mode>_vis"): Likewise.
23242 ("alignaddrsi_vis"): Likewise.
23243 ("alignaddrdi_vis"): Likweise.
23244 ("alignaddrlsi_vis"): Likewise.
23245 ("alignaddrldi_vis"): Likewise.
23246 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
23247 ("bmaskdi_vis"): Likewise.
23248 ("bmasksi_vis"): Likewise.
23249 ("bshuffle<VM64:mode>_vis"): Likewise.
23250 ("cmask8<P:mode>_vis"): Likewise.
23251 ("cmask16<P:mode>_vis"): Likewise.
23252 ("cmask32<P:mode>_vis"): Likewise.
23253 ("pdistn<P:mode>_vis"): Likewise.
23254 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
23256 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23258 * config/sparc/sparc.md ("subtype"): New insn attribute.
23259 ("*wrgsr_sp64"): Set insn subtype.
23260 ("*rdgsr_sp64"): Likewise.
23261 ("alignaddrsi_vis"): Likewise.
23262 ("alignaddrdi_vis"): Likewise.
23263 ("alignaddrlsi_vis"): Likewise.
23264 ("alignaddrldi_vis"): Likewise.
23265 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
23266 ("fexpand_vis"): Likewise.
23267 ("fpmerge_vis"): Likewise.
23268 ("faligndata<VM64:mode>_vis"): Likewise.
23269 ("bshuffle<VM64:mode>_vis"): Likewise.
23270 ("cmask8<P:mode>_vis"): Likewise.
23271 ("cmask16<P:mode>_vis"): Likewise.
23272 ("cmask32<P:mode>_vis"): Likewise.
23273 ("fchksm16_vis"): Likewise.
23274 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
23275 ("fmean16_vis"): Likewise.
23276 ("fp<plusminus_insn>64_vis"): Likewise.
23277 ("<plusminus_insn>v8qi3"): Likewise.
23278 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
23279 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
23280 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
23281 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
23282 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
23283 ("*movqi_insn"): Likewise.
23284 ("*movhi_insn"): Likewise.
23285 ("*movsi_insn"): Likewise.
23286 ("movsi_pic_gotdata_op"): Likewise.
23287 ("*movdi_insn_sp32"): Likewise.
23288 ("*movdi_insn_sp64"): Likewise.
23289 ("movdi_pic_gotdata_op"): Likewise.
23290 ("*movsf_insn"): Likewise.
23291 ("*movdf_insn_sp32"): Likewise.
23292 ("*movdf_insn_sp64"): Likewise.
23293 ("*zero_extendhisi2_insn"): Likewise.
23294 ("*zero_extendqihi2_insn"): Likewise.
23295 ("*zero_extendqisi2_insn"): Likewise.
23296 ("*zero_extendqidi2_insn"): Likewise.
23297 ("*zero_extendhidi2_insn"): Likewise.
23298 ("*zero_extendsidi2_insn_sp64"): Likewise.
23299 ("ldfsr"): Likewise.
23300 ("prefetch_64"): Likewise.
23301 ("prefetch_32"): Likewise.
23302 ("tie_ld32"): Likewise.
23303 ("tie_ld64"): Likewise.
23304 ("*tldo_ldub_sp32"): Likewise.
23305 ("*tldo_ldub1_sp32"): Likewise.
23306 ("*tldo_ldub2_sp32"): Likewise.
23307 ("*tldo_ldub_sp64"): Likewise.
23308 ("*tldo_ldub1_sp64"): Likewise.
23309 ("*tldo_ldub2_sp64"): Likewise.
23310 ("*tldo_ldub3_sp64"): Likewise.
23311 ("*tldo_lduh_sp32"): Likewise.
23312 ("*tldo_lduh1_sp32"): Likewise.
23313 ("*tldo_lduh_sp64"): Likewise.
23314 ("*tldo_lduh1_sp64"): Likewise.
23315 ("*tldo_lduh2_sp64"): Likewise.
23316 ("*tldo_lduw_sp32"): Likewise.
23317 ("*tldo_lduw_sp64"): Likewise.
23318 ("*tldo_lduw1_sp64"): Likewise.
23319 ("*tldo_ldx_sp64"): Likewise.
23320 ("*mov<VM32:mode>_insn"): Likewise.
23321 ("*mov<VM64:mode>_insn_sp64"): Likewise.
23322 ("*mov<VM64:mode>_insn_sp32"): Likewise.
23324 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23326 * config/sparc/sparc.md ("type"): New insn type viscmp.
23327 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
23329 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
23330 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
23331 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
23332 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
23334 ("n7_vis_logical_11cycle"): Likewise.
23335 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
23336 * config/sparc/niagara2.md ("niag3_vis": Likewise.
23337 * config/sparc/niagara.md ("niag_vis"): Likewise.
23338 * config/sparc/ultra3.md ("us3_fga"): Likewise.
23339 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
23341 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23343 * config/sparc/sparc.md: New instruction type `bmask'.
23344 (bmaskdi_vis): Use the `bmask' type.
23345 (bmasksi_vis): Likewise.
23346 * config/sparc/ultra3.md (us3_array): Likewise.
23347 * config/sparc/niagara7.md (n7_array): Likewise.
23348 * config/sparc/niagara4.md (n4_array): Likewise.
23349 * config/sparc/niagara2.md (niag2_vis): Likewise.
23350 (niag3_vis): Likewise.
23351 * config/sparc/niagara.md (niag_vis): Likewise.
23353 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
23355 * ipa-comdats.c: Remove optimize check from gate.
23356 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
23357 for functions not optimized.
23358 (ipa_fn_summary_read): Skip optimize check.
23359 (ipa_fn_summary_write): Likewise.
23360 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
23362 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
23364 (can_inline_edge_p): Check flag_pcc_struct_return for match.
23365 (check_callers): Give up on caller which is not optimized.
23366 (inline_small_functions): Likewise.
23367 (ipa_inline): Do not give up when not optimizing.
23368 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
23369 away unoptimizes cdtors.
23370 (whole_program_function_and_variable_visibility): Do
23371 ipa_discover_readonly_nonaddressable_vars in LTO mode.
23372 * ipa.c (process_references): Do not check optimize.
23373 (symbol_table::remove_unreachable_nodes): Update optimize check.
23374 (set_writeonly_bit): Update optimize check.
23375 (pass_ipa_cdtor_merge::gate): Do not check optimize.
23376 (pass_ipa_single_use::gate): Remove.
23378 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23380 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
23381 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
23382 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
23383 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
23384 permute_load, permute_store, adjust_extract, adjust_splat,
23385 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
23386 replace_swap_with_copy, dump_swap_insn_table,
23387 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
23388 recombine_lvx_pattern, recombine_stvx_pattern,
23389 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
23390 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
23391 to file rs6000-p8swap.c.
23392 * config/rs6000/rs6000-p8swap.c: New file.
23393 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
23394 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
23395 and rs6000*-*-* targets.
23397 2017-07-06 David Malcolm <dmalcolm@redhat.com>
23399 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
23401 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
23403 * lto-wrapper.c (merge_and_complain): Do not merge
23404 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
23405 fsigned_zeros, ftrapping_math, fwrapv.
23406 (append_compiler_options): Do not track these options.
23407 (append_linker_options): Likewie
23409 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
23411 * cgraphunit.c (cgraph_node::finalize_function): When
23412 !flag_toplevel_reorde set no_reorder flag.
23413 (varpool_node::finalize_decl): Likewise.
23414 (symbol_table::compile): Drop no toplevel reorder path.
23416 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
23418 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
23419 edges; zero probability is not better than uninitialized.
23421 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
23423 * asan.h (asan_sanitize_allocas_p): Declare.
23424 * asan.c (asan_sanitize_allocas_p): New function.
23425 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
23426 (handle_builtin_alloca): Likewise.
23427 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
23428 if !asan_sanitize_allocas_p.
23429 * params.def (asan-instrument-allocas): Add new option.
23430 * params.h (ASAN_PROTECT_ALLOCAS): Define.
23431 * opts.c (common_handle_option): Disable allocas sanitization for
23434 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
23436 * asan.c: Include gimple-fold.h.
23437 (get_last_alloca_addr): New function.
23438 (handle_builtin_stackrestore): Likewise.
23439 (handle_builtin_alloca): Likewise.
23440 (asan_emit_allocas_unpoison): Likewise.
23441 (get_mem_refs_of_builtin_call): Add new parameter, remove const
23442 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
23443 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
23444 (instrument_builtin_call): Pass gimple iterator to
23445 get_mem_refs_of_builtin_call.
23446 (last_alloca_addr): New global.
23447 * asan.h (asan_emit_allocas_unpoison): Declare.
23448 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
23449 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
23450 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
23451 if function calls alloca.
23452 * gimple-fold.c (replace_call_with_value): Remove static keyword.
23453 * gimple-fold.h (replace_call_with_value): Declare.
23454 * internal-fn.c: Include asan.h.
23455 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
23456 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
23458 2017-07-06 David Malcolm <dmalcolm@redhat.com>
23460 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
23461 (C_SELFTEST_FLAGS): New.
23462 (CPP_SELFTEST_FLAGS): New.
23463 (SELFTEST_DEPS): New, from deps of s-selftest.
23464 (C_SELFTEST_DEPS): New, from deps of s-selftest.
23465 (CPP_SELFTEST_DEPS): New.
23466 (selftest): Add dependency on s-selftest-c++.
23467 (s-selftest): Rename to...
23468 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
23469 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
23470 than SELFTEST_FLAGS.
23471 (selftest-gdb): Rename to...
23472 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
23474 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
23475 (selftest-valgrind): Rename to...
23476 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
23478 (selftest-valgrind): Reintroduce as an alias for
23479 selftest-c-valgrind.
23480 (s-selftest-c++): New.
23481 (selftest-c++-gdb): New.
23482 (selftest-c++-valgrind): New.
23484 2017-07-06 Olivier Hainque <hainque@adacore.com>
23486 * gcc.c (process_command): When deciding if undefined variables
23487 should be ignored when processing specs, accept "gcc -v" as well.
23489 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
23491 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
23492 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
23494 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
23496 * config/arm/arm-cpus.in (armv8-r): Add new entry.
23497 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
23498 * config/arm/arm-tables.opt: Regenerate.
23499 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
23501 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
23503 2017-07-06 Carl Love <cel@us.ibm.com>
23505 * ChangeLog: Clean up from mid air collision
23507 2017-07-06 Carl Love <cel@us.ibm.com>
23509 * config/rs6000/rs6000-c.c: Add support for built-in functions
23510 vector signed int vec_subc (vector signed int, vector signed int);
23511 vector signed __int128 vec_subc (vector signed __int128,
23512 vector signed __int128);
23513 vector unsigned __int128 vec_subc (vector unsigned __int128,
23514 vector unsigned __int128);
23515 vector signed int vec_sube (vector signed int, vector signed int,
23516 vector signed int);
23517 vector unsigned int vec_sube (vector unsigned int,
23518 vector unsigned int,
23519 vector unsigned int);
23520 vector signed __int128 vec_sube (vector signed __int128,
23521 vector signed __int128,
23522 vector signed__int128);
23523 vector unsigned __int128 vec_sube (vector unsigned __int128,
23524 vector unsigned __int128,
23525 vector unsigned __int128);
23526 vector signed int vec_subec (vector signed int, vector signed int,
23527 vector signed int);
23528 vector unsigned int vec_subec (vector unsigned int,
23529 vector unsigned int,
23530 vector unsigned int);
23531 vector signed __int128 vec_subec (vector signed __int128,
23532 vector signed __int128,
23533 vector signed__int128);
23534 vector unsigned __int128 vec_subec (vector unsigned __int128,
23535 vector unsigned __int128,
23536 vector unsigned __int128);
23537 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
23538 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
23539 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
23540 BU_ALTIVEC_OVERLOAD_X definitions.
23541 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
23542 * doc/extend.texi: Update the built-in documentation file for the new
23543 built-in functions.
23545 2017-07-06 David Malcolm <dmalcolm@redhat.com>
23548 * diagnostic-show-locus.c (layout::layout): Use start and finish
23549 spelling location for the start and finish of each range.
23550 * genmatch.c (linemap_client_expand_location_to_spelling_point):
23551 Add unused aspect param.
23552 * input.c (expand_location_1): Add "aspect" param, and use it
23553 to access the correct part of the location.
23554 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
23556 (expand_location_to_spelling_point): Likewise.
23557 (linemap_client_expand_location_to_spelling_point): Add "aspect"
23558 param, and pass it to expand_location_1.
23560 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
23562 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
23563 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
23564 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
23565 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
23566 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
23567 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
23568 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
23569 _mm_maskz_getmant_ss): New intrinsics.
23570 (__builtin_ia32_getexpss128_mask): Changed to ...
23571 __builtin_ia32_getexpss128_round ... this.
23572 (__builtin_ia32_getexpsd128_mask): Changed to ...
23573 __builtin_ia32_getexpsd128_round ... this.
23574 * config/i386/i386-builtin-types.def
23575 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
23576 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
23577 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
23578 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
23579 __builtin_ia32_getmantss_mask_round): New builtins.
23580 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
23581 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
23582 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
23583 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
23584 * config/i386/sse.md
23585 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
23586 avx512f_sgetexp<mode><mask_scalar_name>
23587 <round_saeonly_scalar_name> ... this.
23588 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
23589 %0, %1, %2<round_saeonly_op3>}): Changed to ...
23590 vgetexp<ssescalarmodesuffix>
23591 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
23592 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
23593 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
23594 avx512f_vgetmant<mode><mask_scalar_name>
23595 <round_saeonly_scalar_name> ... this.
23596 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
23597 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
23598 vgetmant<ssescalarmodesuffix>
23599 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
23600 %0<mask_scalar_operand4>, %1, %2
23601 <round_saeonly_scalar_mask_op4>, %3} ... this.
23602 * config/i386/subst.md (mask_scalar_operand4,
23603 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
23604 round_saeonly_scalar_nimm_predicate): New subst attributes.
23606 2017-07-06 Julia Koval <julia.koval@intel.com>
23608 * config/i386/i386.c (ix86_erase_embedded_rounding):
23609 Remove code for old rounding pattern.
23611 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
23613 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
23615 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
23617 * doc/sourcebuild.texi (Test Directives, Variants of
23618 dg-require-support): Add documentation for dg-require-stack-check.
23620 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
23622 * config/i386/subst.md (mask_scalar, round_scalar,
23623 round_saeonly_scalar): New meta-templates.
23624 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
23625 round_scalar_mask_operand3, round_scalar_mask_op3,
23626 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
23627 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
23628 round_saeonly_scalar_constraint,
23629 round_saeonly_scalar_prefix): New subst attribute.
23630 * config/i386/sse.md
23631 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
23632 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
23633 <round_scalar_name> ... this.
23634 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
23635 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
23636 <round_scalar_name> ... this.
23637 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
23638 <sse>_vm<code><mode>3<mask_scalar_name>
23639 <round_saeonly_scalar_name> ... this.
23640 (v<plusminus_mnemonic><ssescalarmodesuffix>
23641 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
23642 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
23643 v<plusminus_mnemonic><ssescalarmodesuffix>
23644 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
23645 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
23646 (v<multdiv_mnemonic><ssescalarmodesuffix>
23647 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
23648 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
23649 v<multdiv_mnemonic><ssescalarmodesuffix>
23650 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
23651 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
23652 (v<maxmin_float><ssescalarmodesuffix>
23653 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
23654 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
23655 v<maxmin_float><ssescalarmodesuffix>
23656 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
23657 %0<mask_scalar_operand3>, %1, %<iptr>2
23658 <round_saeonly_scalar_mask_op3>} ... this.
23660 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
23662 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
23663 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
23665 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
23666 Alan Hayward <alan.hayward@arm.com>
23667 David Sherwood <david.sherwood@arm.com>
23669 * combine.c (simplify_if_then_else): Remove "enum" before
23671 * compare-elim.c (can_eliminate_compare): Likewise.
23672 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
23674 (aarch64_lookup_simd_builtin_type): Likewise.
23675 (aarch64_simd_builtin_type): Likewise.
23676 (aarch64_init_simd_builtin_types): Likewise.
23677 (aarch64_simd_expand_args): Likewise.
23678 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
23680 (aarch64_reverse_mask): Likewise.
23681 (aarch64_simd_emit_reg_reg_move): Likewise.
23682 (aarch64_gen_adjusted_ldpstp): Likewise.
23683 (aarch64_ccmp_mode_to_code): Likewise.
23684 (aarch64_operands_ok_for_ldpstp): Likewise.
23685 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
23686 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
23688 (aarch64_min_divisions_for_recip_mul): Likewise.
23689 (aarch64_reassociation_width): Likewise.
23690 (aarch64_get_condition_code_1): Likewise.
23691 (aarch64_simd_emit_reg_reg_move): Likewise.
23692 (aarch64_simd_attr_length_rglist): Likewise.
23693 (aarch64_reverse_mask): Likewise.
23694 (aarch64_operands_ok_for_ldpstp): Likewise.
23695 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
23696 (aarch64_gen_adjusted_ldpstp): Likewise.
23697 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
23699 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
23700 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
23701 (arm_lookup_simd_builtin_type): Likewise.
23702 (arm_simd_builtin_type): Likewise.
23703 (arm_init_simd_builtin_types): Likewise.
23704 (arm_expand_builtin_args): Likewise.
23705 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
23706 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
23707 (ft32_setup_incoming_varargs): Likewise.
23708 (ft32_function_arg): Likewise.
23709 (ft32_function_arg_advance): Likewise.
23710 (ft32_pass_by_reference): Likewise.
23711 (ft32_arg_partial_bytes): Likewise.
23712 (ft32_valid_pointer_mode): Likewise.
23713 (ft32_addr_space_pointer_mode): Likewise.
23714 (ft32_addr_space_legitimate_address_p): Likewise.
23715 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
23717 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
23718 (ix86_emit_outlined_ms2sysv_restore): Likewise.
23719 (iamcu_alignment): Likewise.
23720 (canonicalize_vector_int_perm): Likewise.
23721 (ix86_noce_conversion_profitable_p): Likewise.
23722 (ix86_mpx_bound_mode): Likewise.
23723 (ix86_operands_ok_for_move_multiple): Likewise.
23724 * config/microblaze/microblaze-protos.h
23725 (microblaze_expand_conditional_branch_reg): Likewise.
23726 * config/microblaze/microblaze.c
23727 (microblaze_expand_conditional_branch_reg): Likewise.
23728 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
23730 (rs6000_reassociation_width): Likewise.
23731 (rs6000_invalid_binary_op): Likewise.
23732 (fusion_p9_p): Likewise.
23733 (emit_fusion_p9_load): Likewise.
23734 (emit_fusion_p9_store): Likewise.
23735 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
23737 (riscv_hard_regno_mode_ok_p): Likewise.
23738 (riscv_address_insns): Likewise.
23739 (riscv_split_symbol): Likewise.
23740 (riscv_legitimize_move): Likewise.
23741 (riscv_function_value): Likewise.
23742 (riscv_hard_regno_nregs): Likewise.
23743 (riscv_expand_builtin): Likewise.
23744 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
23745 (riscv_build_integer): Likewise.
23746 (riscv_split_integer): Likewise.
23747 (riscv_legitimate_constant_p): Likewise.
23748 (riscv_cannot_force_const_mem): Likewise.
23749 (riscv_regno_mode_ok_for_base_p): Likewise.
23750 (riscv_valid_base_register_p): Likewise.
23751 (riscv_valid_offset_p): Likewise.
23752 (riscv_valid_lo_sum_p): Likewise.
23753 (riscv_classify_address): Likewise.
23754 (riscv_legitimate_address_p): Likewise.
23755 (riscv_address_insns): Likewise.
23756 (riscv_load_store_insns): Likewise.
23757 (riscv_force_binary): Likewise.
23758 (riscv_split_symbol): Likewise.
23759 (riscv_force_address): Likewise.
23760 (riscv_legitimize_address): Likewise.
23761 (riscv_move_integer): Likewise.
23762 (riscv_legitimize_const_move): Likewise.
23763 (riscv_legitimize_move): Likewise.
23764 (riscv_address_cost): Likewise.
23765 (riscv_subword): Likewise.
23766 (riscv_output_move): Likewise.
23767 (riscv_canonicalize_int_order_test): Likewise.
23768 (riscv_emit_int_order_test): Likewise.
23769 (riscv_function_arg_boundary): Likewise.
23770 (riscv_pass_mode_in_fpr_p): Likewise.
23771 (riscv_pass_fpr_single): Likewise.
23772 (riscv_pass_fpr_pair): Likewise.
23773 (riscv_get_arg_info): Likewise.
23774 (riscv_function_arg): Likewise.
23775 (riscv_function_arg_advance): Likewise.
23776 (riscv_arg_partial_bytes): Likewise.
23777 (riscv_function_value): Likewise.
23778 (riscv_pass_by_reference): Likewise.
23779 (riscv_setup_incoming_varargs): Likewise.
23780 (riscv_print_operand): Likewise.
23781 (riscv_elf_select_rtx_section): Likewise.
23782 (riscv_save_restore_reg): Likewise.
23783 (riscv_for_each_saved_reg): Likewise.
23784 (riscv_register_move_cost): Likewise.
23785 (riscv_hard_regno_mode_ok_p): Likewise.
23786 (riscv_hard_regno_nregs): Likewise.
23787 (riscv_class_max_nregs): Likewise.
23788 (riscv_memory_move_cost): Likewise.
23789 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
23790 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
23791 (rl78_addr_space_address_mode): Likewise.
23792 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
23794 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
23795 (rs6000_reassociation_width): Likewise.
23796 (rs6000_invalid_binary_op): Likewise.
23797 (fusion_p9_p): Likewise.
23798 (emit_fusion_p9_load): Likewise.
23799 (emit_fusion_p9_store): Likewise.
23800 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
23801 (ok_for_simple_move_operands): Likewise.
23802 (ok_for_simple_move_strict_operands): Likewise.
23803 (ok_for_simple_arith_logic_operands): Likewise.
23804 (visium_legitimize_reload_address): Likewise.
23805 (visium_select_cc_mode): Likewise.
23806 (output_cbranch): Likewise.
23807 (visium_split_double_move): Likewise.
23808 (visium_expand_copysign): Likewise.
23809 (visium_expand_int_cstore): Likewise.
23810 (visium_expand_fp_cstore): Likewise.
23811 * config/visium/visium.c (visium_pass_by_reference): Likewise.
23812 (visium_function_arg): Likewise.
23813 (visium_function_arg_advance): Likewise.
23814 (visium_libcall_value): Likewise.
23815 (visium_setup_incoming_varargs): Likewise.
23816 (visium_legitimate_constant_p): Likewise.
23817 (visium_legitimate_address_p): Likewise.
23818 (visium_legitimize_address): Likewise.
23819 (visium_secondary_reload): Likewise.
23820 (visium_register_move_cost): Likewise.
23821 (visium_memory_move_cost): Likewise.
23822 (prepare_move_operands): Likewise.
23823 (ok_for_simple_move_operands): Likewise.
23824 (ok_for_simple_move_strict_operands): Likewise.
23825 (ok_for_simple_arith_logic_operands): Likewise.
23826 (visium_function_value_1): Likewise.
23827 (rtx_ok_for_offset_p): Likewise.
23828 (visium_legitimize_reload_address): Likewise.
23829 (visium_split_double_move): Likewise.
23830 (visium_expand_copysign): Likewise.
23831 (visium_expand_int_cstore): Likewise.
23832 (visium_expand_fp_cstore): Likewise.
23833 (visium_split_cstore): Likewise.
23834 (visium_select_cc_mode): Likewise.
23835 (visium_split_cbranch): Likewise.
23836 (output_cbranch): Likewise.
23837 (visium_print_operand_address): Likewise.
23838 * expmed.c (flip_storage_order): Likewise.
23839 * expmed.h (emit_cstore): Likewise.
23840 (flip_storage_order): Likewise.
23841 * genrecog.c (validate_pattern): Likewise.
23842 * hsa-gen.c (gen_hsa_addr): Likewise.
23843 * internal-fn.c (expand_arith_overflow): Likewise.
23844 * ira-color.c (allocno_copy_cost_saving): Likewise.
23845 * lra-assigns.c (find_hard_regno_for_1): Likewise.
23846 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
23847 (process_invariant_for_inheritance): Likewise.
23848 * lra-eliminations.c (move_plus_up): Likewise.
23849 * omp-low.c (lower_oacc_reductions): Likewise.
23850 * simplify-rtx.c (simplify_subreg): Likewise.
23851 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
23852 (TARGET_CHKP_BOUND_MODE): Likewise..
23853 * targhooks.c (default_chkp_bound_mode): Likewise.
23854 (default_setup_incoming_vararg_bounds): Likewise.
23855 * targhooks.h (default_chkp_bound_mode): Likewise.
23856 (default_setup_incoming_vararg_bounds): Likewise.
23857 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
23858 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
23859 (have_whole_vector_shift): Likewise.
23860 * tree-vect-stmts.c (vectorizable_load): Likewise.
23861 * doc/tm.texi: Regenerate.
23863 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
23865 Graceful degrade if Binutils PR21472 is not available.
23868 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
23869 .rodata in flash test fails.
23870 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
23871 * confgure: Regenerate.
23872 * config.in: Regenerate.
23873 * config/avr/avr.c (avr_asm_named_section)
23874 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
23875 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
23876 (avr_asm_init_sections): Same.
23878 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
23880 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
23881 (fma<VH:mode>4_intrinsic): Likewise.
23882 (*fmsub<VCVTF:mode>4): Likewise.
23883 (*fmsub<VH:mode>4_intrinsic): Likewise.
23885 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
23888 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
23889 Don't depend on "optimize > 0".
23890 (out_movhi_r_mr, out_movqi_mr_r): Same.
23891 (out_movhi_mr_r, out_movqi_r_mr): Same.
23892 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
23893 io_address_operand on "optimize > 0".
23895 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23897 * tree-loop-distribution.c: Add general explanantion on the pass.
23898 (generate_loops_for_partition): Mark distributed loop.
23899 (pg_add_dependence_edges): New parameter. Handle alias data
23900 dependence specially and record it in the parameter if asked.
23901 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
23902 (init_partition_graph_vertices, add_partition_graph_edge): New.
23903 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
23904 (free_partition_graph_vdata, build_partition_graph): New.
23905 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
23906 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
23907 (data_ref_segment_size, latch_dominated_by_data_ref): New.
23908 (compute_alias_check_pairs, version_loop_by_alias_check): New.
23909 (version_for_distribution_p, finalize_partitions): New.
23910 (distribute_loop): Handle alias data dependence specially. Factor
23911 out loop fusion code as functions and call these functions.
23913 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23915 * tree-loop-distribution.c (classify_partition): New parameter and
23916 better handle reduction statement.
23917 (rdg_build_partitions): Revise comment.
23918 (distribute_loop): Compute statements in all partitions and pass it
23919 to classify_partition.
23921 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23923 * tree-loop-distribution.c (enum partition_type): New.
23924 (struct partition): New field type.
23925 (partition_merge_into): Add parameter. Update partition type.
23926 (data_dep_in_cycle_p, update_type_for_merge): New functions.
23927 (build_rdg_partition_for_vertex): Compute partition type.
23928 (rdg_build_partitions): Dump partition type.
23929 (distribute_loop): Update calls to partition_merge_into.
23931 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23933 * tree-loop-distribution.c (struct ddr_hasher): New.
23934 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
23936 (classify_partition): Call get_data_dependence.
23937 (pg_add_dependence_edges): Ditto.
23938 (distribute_loop): Release data dependence hash table.
23940 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23942 * tree-loop-distribution.c (ref_base_address): Delete.
23943 (similar_memory_accesses): Rename ...
23944 (share_memory_accesses): ... to this. Check if partitions access
23945 the same memory reference.
23946 (distribute_loop): Call share_memory_accesses.
23948 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23950 * tree-loop-distribution.c (struct partition): New field recording
23951 its data reference.
23952 (partition_alloc, partition_free): Init and release data refs.
23953 (partition_merge_into): Merge data refs.
23954 (build_rdg_partition_for_vertex): Collect data refs for partition.
23955 (pg_add_dependence_edges): Change parameters from vector to bitmap.
23957 (distribute_loop): Remve data refs from vertice data of partition
23960 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23962 * tree-loop-distribution.c (params.h): Include header file.
23963 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
23964 (datarefs_vec): New global var.
23965 (create_rdg_vertices): Use datarefs_vec directly.
23966 (free_rdg): Don't free data references.
23967 (build_rdg): Update use. Don't free data references.
23968 (distribute_loop): Compute global variable for data references.
23969 Bail out if there are too many data references.
23971 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23973 * tree-loop-distribution.c (loop_nest): New global var.
23974 (build_rdg): Use loop directly, rather than loop nest.
23975 (pg_add_dependence_edges): Remove loop nest parameter. Use global
23977 (distribute_loop): Compute global variable loop nest. Update use.
23979 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23981 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
23982 (partition_merge_into): New parameter. Dump reason for fusion.
23983 (distribute_loop): Update use of partition_merge_into.
23985 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23987 * tree-loop-distribution.c (bb_top_order_index): New.
23988 (bb_top_order_index_size, bb_top_order_cmp): New.
23989 (stmts_from_loop): Use topological order.
23990 (pass_loop_distribution::execute): Compute and release topological
23991 order for basic blocks.
23993 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23995 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
23998 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24000 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
24001 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
24002 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
24003 * internal-fn.def (LOOP_DIST_ALIAS): New.
24004 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
24005 (fold_loop_internal_call): ... this.
24006 (vect_loop_dist_alias_call): New function.
24007 (set_uid_loop_bbs): Call fold_loop_internal_call.
24008 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
24011 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
24014 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
24015 Require dead FLAGS_REG at the beginning of a peephole.
24017 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
24020 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
24021 arguments in the call to __builtin_ia32_sbb_u32.
24022 (_subborrow_u64): Swap _X and _Y arguments in the call to
24023 __builtin_ia32_sbb_u64.
24025 2017-07-04 Jakub Jelinek <jakub@redhat.com>
24028 * tree-vrp.c (compare_assert_loc): Turn into a function template
24029 with stable template parameter. Only test if a->e is NULL,
24030 !a->e == !b->e has been verified already. Use e == NULL or
24031 e != NULL instead of e or ! e tests. If stable is true, don't use
24032 iterative_hash_expr, on the other side allow a or b or both NULL
24033 and sort the NULLs last.
24034 (process_assert_insertions): Sort using compare_assert_loc<false>
24035 instead of compare_assert_loc, later sort using
24036 compare_assert_loc<true> before calling process_assert_insertions_for
24037 in a loop. Use break instead of continue once seen NULL pointer.
24039 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
24041 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
24042 Cortex-R7 and Cortex-R8 processors.
24044 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
24046 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
24047 uninitialized while src is not.
24049 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
24051 * common/config/arm/arm-common.c: Adjust include path for
24053 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
24054 (arm-cpu.h): Create in build directory. Adjust dependency rules.
24055 (arm-cpu-data.h): Likewise.
24056 (arm-cpu-cdata.h): Likewise.
24057 * config/arm/arm-cpu.h: Delete.
24058 * config/arm/arm-cpu-cdata.h: Delete.
24059 * config/arm/arm-cpu-data.h: Delete.
24061 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
24063 * config/arm/arm-cpus.in (cortex-a55): New.
24064 (cortex-a75): Likewise.
24065 (cortex-a75.cortex-a55): Likewise.
24066 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
24068 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
24069 * config/arm/arm-cpu-cdata.h: Regenerate.
24070 * config/arm/arm-cpu-data.h: Regenerate.
24071 * config/arm/arm-cpu.h: Regenerate.
24072 * config/arm/arm-tables.opt: Regenerate.
24073 * config/arm/arm-tune.md: Regenerate.
24075 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
24077 * haifa-sched.c (sched_create_recovery_edges): Update profile.
24079 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
24081 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
24084 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
24086 PR tree-optimization/81292
24087 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
24088 full_string_p, also call adjust_related_strinfos if the adjustment
24089 is simple, otherwise invalidate related strinfos.
24091 2017-07-04 Martin Liska <mliska@suse.cz>
24094 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
24095 newly created variable as DECL_IGNORED_P.
24097 2017-07-04 Martin Liska <mliska@suse.cz>
24100 * ipa-inline.c (inline_small_functions):
24101 Use xstrdup_for_dump.
24103 2017-07-04 Tom de Vries <tom@codesourcery.com>
24105 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
24107 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
24110 * config/darwin.c (darwin_function_switched_text_sections):
24113 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
24115 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
24117 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
24119 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
24121 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24123 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
24124 min_profitable_iters, and th as inclusive lower bounds.
24125 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
24126 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
24127 for min_profitable_iters and min_profitable_estimate.
24128 (vect_transform_loop): Treat th as an inclusive lower bound.
24129 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
24131 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
24134 * config/darwin.c (darwin_function_switched_text_sections):
24135 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
24136 in two pieces, and suppress the use of buf.
24138 2017-07-03 Nathan Sidwell <nathan@acm.org>
24140 * hash-table.h (hash_table_mod1): Fix indentation.
24142 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24144 PR middle-end/81290
24145 * predict.c (force_edge_cold): Be more careful about propagation
24147 * profile-count.h (profile_probability::guessed,
24148 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
24150 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
24152 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
24154 * doc/invoke.texi (rcpc architecture extension): Document it.
24156 2017-07-03 Richard Biener <rguenther@suse.de>
24158 PR tree-optimization/60510
24159 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
24160 the scalar reduction PHI and use it.
24161 (vectorizable_reduction): Properly guard the single_defuse_cycle
24162 path for non-SLP reduction chains where we cannot use it.
24163 Rework reduc_def/index and vector type deduction. Rework
24164 vector operand gathering during reduction op code-gen.
24165 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
24166 chains dissolve the chain and leave it to non-SLP reduction
24169 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24171 * tree-data-ref.h (dr_alignment): Declare.
24172 * tree-data-ref.c (dr_alignment): New function.
24173 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
24174 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
24176 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
24178 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24180 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
24181 and base_misalignment fields.
24182 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
24183 * tree-data-ref.c: Include builtins.h.
24184 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
24185 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
24186 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
24187 * tree-vect-data-refs.c: Include tree-cfg.h.
24188 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
24189 fields instead of calculating an alignment here.
24190 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
24191 innermost_loop_behavior fields.
24193 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24195 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
24197 (DR_STEP_ALIGNMENT): New macro.
24198 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
24199 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
24200 (create_data_ref): Print it.
24201 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
24202 to tell whether the step preserves vector (mis)alignment.
24203 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
24204 Move the check for an integer step and generalise to all INTEGER_CST.
24205 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
24206 Print the outer step alignment.
24208 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24210 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
24211 with offset_alignment.
24212 (DR_ALIGNED_TO): Delete.
24213 (DR_OFFSET_ALIGNMENT): New macro.
24214 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
24215 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
24216 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
24217 (create_data_ref): Likewise.
24218 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
24219 (vect_analyze_data_refs): Likewise.
24220 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
24221 creating dummy innermost behavior.
24223 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24225 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
24226 with a "innermost_loop_behavior *" and refeence tree.
24227 * tree-data-ref.c (dr_analyze_innermost): Likewise.
24228 (create_data_ref): Update call accordingly.
24229 * tree-predcom.c (find_looparound_phi): Likewise.
24231 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24233 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
24234 fields with dr_wrt_vec_loop.
24235 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
24236 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
24237 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
24238 (vect_dr_behavior): New function.
24239 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
24240 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
24241 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
24242 track whether the step preserves the misalignment.
24243 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
24244 Use vect_dr_behavior.
24245 (vect_setup_realignment): Update call accordingly.
24246 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
24247 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
24248 call to vect_create_addr_base_for_vector_ref.
24249 (vect_create_cond_for_align_checks): Likewise.
24250 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
24251 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
24252 (vect_recog_mask_conversion_pattern): Likewise.
24253 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
24254 (new_stmt_vec_info): Remove redundant zeroing.
24256 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
24258 * common/config/arm/arm-common.c (arm_be8_option): New function.
24259 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
24260 (ISA_ARMv6): Add isa_bit_be8.
24261 * config/arm/arm.h (arm_be8_option): Add prototype.
24262 (BE8_SPEC_FUNCTION): New define.
24263 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
24264 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
24265 (mlittle-endian): Similarly.
24266 (mbe8, mbe32): New options.
24267 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
24268 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
24270 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24272 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
24274 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24276 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
24277 (cleanup_tree_cfg_bb): Use it.
24278 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
24280 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
24282 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24285 * loop-doloop.c (add_test): Update profile.
24287 2017-07-03 Martin Liska <mliska@suse.cz>
24290 * sanopt.c (rewrite_usage_of_param): New function.
24291 (sanitize_rewrite_addressable_params): Likewise.
24292 (pass_sanopt::execute): Call rewrite_usage_of_param.
24294 2017-07-03 Richard Biener <rguenther@suse.de>
24296 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
24297 back to using VIEW_CONVERT_EXPR.
24299 2017-07-03 Martin Liska <mliska@suse.cz>
24302 * doc/extend.texi: Document when a resolver function is
24303 generated for target_clones.
24305 2017-07-03 Martin Liska <mliska@suse.cz>
24307 * asan.c (asan_emit_stack_protection): Unpoison just red zones
24308 and shadow memory of auto variables which are subject of
24309 use-after-scope sanitization.
24310 (asan_expand_mark_ifn): Add do set only when is_poison.
24312 2016-07-03 Richard Biener <rguenther@suse.de>
24314 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
24316 (vect_force_simple_reduction): Record reduction def -> phi mapping.
24317 (vectorizable_reduction): Perform reduction PHI creation when
24318 visiting a reduction PHI and adjust and simplify code generation
24319 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
24320 (vect_transform_loop): Visit reduction PHIs.
24321 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
24322 defs into the SLP tree.
24323 (vect_build_slp_tree): Reduction defs terminate the recursion.
24324 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
24326 (vect_get_vec_defs_for_stmt_copy): Export.
24327 (vect_get_vec_defs): Likewise.
24328 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
24330 (vect_get_vec_defs_for_stmt_copy): Declare.
24331 (vect_get_vec_defs): Likewise.
24333 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24335 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
24336 parameter with a "loop" parameter and use it instead of the
24337 loop containing DR_STMT. Don't check simple_iv when doing
24338 BB analysis. Describe the two analysis modes in the comment.
24340 2017-07-03 Tom de Vries <tom@codesourcery.com>
24342 PR tree-optimization/69468
24343 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
24344 (find_same_succ_bb): Handle ignore_edge_flags.
24346 2017-07-03 Tom de Vries <tom@codesourcery.com>
24348 PR tree-optimization/81192
24349 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
24351 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
24353 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
24355 2017-07-03 Tom de Vries <tom@codesourcery.com>
24357 PR tree-optimization/81192
24358 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
24359 BB_SAME_SUCC (bb) == NULL.
24361 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24363 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
24366 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24368 * dumpfile.c: Include profile-count.h
24369 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
24371 (insert_cond_bb): Update profile.
24372 * tree-cfg.h (insert_cond_bb): Update prototype.
24373 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
24374 * tree-dump.c: Do not include tree-cfg.
24376 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24378 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
24380 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24382 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
24385 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24387 * tree-complex.c (expand_complex_div_wide): update profile.
24389 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
24390 Alan Hayward <alan.hayward@arm.com>
24391 David Sherwood <david.sherwood@arm.com>
24393 * Makefile.in (MACHMODE_H): Remove insn-modes.h
24394 (CORETYPES_H): New define.
24395 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
24396 (insn-modes-inline.h, s-modes-inline-h): New rules.
24397 (generated_files): Add insn-modes-inline.h.
24398 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
24399 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
24400 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
24401 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
24402 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
24403 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
24404 (build/gencodes.o, build/genconditions.o): Likewise.
24405 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
24406 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
24407 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
24408 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
24409 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
24410 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
24411 * coretypes.h: Include everything up to real.h for generators.
24412 Include insn-modes.h first. Include wide-int-print.h after
24413 wide-int.h. Include insn-modes-inline.h and then machmode.h.
24414 * machmode.h: Don't include insn-modes.h here.
24415 * function-tests.c: Remove includes of signop.h, machmode.h,
24416 double-int.h and wide-int.h.
24418 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
24420 * optc-save-gen.awk: Likewise.
24421 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
24422 * godump.c: Remove include of wide-int-print.h.
24423 * pretty-print.h: Likewise.
24424 * wide-int-print.cc: Likewise.
24425 * wide-int.cc: Likewise.
24426 * hash-map-tests.c: Remove include of signop.h.
24427 * hash-set-tests.c: Likewise.
24428 * rtl-tests.c: Likewise.
24429 * mkconfig.sh: Remove include of machmode.h.
24430 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
24432 (emit_insn_modes_inline_h): ...this new function. Emit the code
24433 into an insn-modes-inline.h header file, adding appropriate
24434 include guards and end comments.
24435 (emit_insn_modes_c_header): Remove include of machmode.h.
24436 (emit_min_insn_modes_c_header): Include coretypes.h rather than
24438 (main): Handle -i flag and call emit_insn_modes_inline_h when
24441 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
24443 * tree-ssa-strlen.c (strinfo): Rename the length field to
24444 nonzero_chars. Add a full_string_p field.
24445 (compare_nonzero_chars, zero_length_string_p): New functions.
24446 (get_addr_stridx): Add an offset_out parameter.
24447 Use compare_nonzero_chars.
24448 (get_stridx): Update accordingly. Use compare_nonzero_chars.
24449 (new_strinfo): Update after above changes to strinfo.
24450 (set_endptr_and_length): Set full_string_p.
24451 (get_string_length): Update after above changes to strinfo.
24452 (unshare_strinfo): Update call to new_strinfo.
24453 (maybe_invalidate): Likewise.
24454 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
24455 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
24456 as a uhwi instead of an shwi. Update after above changes to
24457 strinfo and new_strinfo.
24458 (zero_length_string): Assert that chainsi contains full strings.
24459 Use zero_length_string_p. Update call to new_strinfo.
24460 (adjust_related_strinfos): Update after above changes to strinfo.
24461 Copy full_string_p from origsi.
24462 (adjust_last_stmt): Use zero_length_string_p.
24463 (handle_builtin_strlen): Update after above changes to strinfo and
24464 new_strinfo. Install the lhs as the string length if the previous
24465 entry didn't describe a full string.
24466 (handle_builtin_strchr): Update after above changes to strinfo
24468 (handle_builtin_strcpy): Likewise.
24469 (handle_builtin_strcat): Likewise.
24470 (handle_builtin_malloc): Likewise.
24471 (handle_pointer_plus): Likewise.
24472 (handle_builtin_memcpy): Likewise. Track nonzero characters
24473 that aren't necessarily followed by a nul terminator.
24474 (handle_char_store): Likewise.
24476 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
24478 PR tree-optimization/80769
24479 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
24480 for malloc and calloc. Document the new invariant that all related
24481 strinfos have delayed lengths or none do.
24482 (verify_related_strinfos): Move earlier in file.
24483 (set_endptr_and_length): New function, split out from...
24484 (get_string_length): ...here. Also set the lengths of related
24486 (zero_length_string): Assert that chainsi has known (rather than
24488 (adjust_related_strinfos): Likewise.
24490 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
24492 PR tree-optimization/81136
24493 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
24494 assert that two references with the same misalignment have the same
24495 compile-time misalignment if those compile-time misalignments
24498 2017-07-01 Andi Kleen <ak@linux.intel.com>
24500 * print-tree.c (print_node): Print all attributes.
24502 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
24504 * cfg.c (scale_bbs_frequencies): New function.
24505 * cfg.h (scale_bbs_frequencies): Declare it.
24506 * cfgloopanal.c (single_likely_exit): Cleanup.
24507 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
24509 (scale_loop_profile): Likewise.
24510 (loop_version): Likewise.
24511 (create_empty_loop_on_edge): Update.
24512 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
24513 scale_loop_frequencies, scale_loop_profile, loopify,
24514 loop_version): Update prototypes.
24515 * modulo-sched.c (sms_schedule): Update.
24516 * predict.c (unlikely_executed_edge_p): Also check probability.
24517 (probably_never_executed_edge_p): Fix typo.
24518 * tree-if-conv.c (version_loop_for_if_conversion): Update.
24519 * tree-parloops.c (gen_parallel_loop): Update.
24520 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
24521 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
24522 * tree-ssa-loop-split.c (split_loop): Update.
24523 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
24524 * tree-vect-loop-manip.c (vect_do_peeling): Update.
24525 (vect_loop_versioning): Update.
24526 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
24528 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
24530 * trans-mem.c (split_bb_make_tm_edge): Update profile.
24532 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
24534 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
24535 to keep profile consistent.
24537 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
24539 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
24540 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
24541 * profile-count.h (max_safe_multiplier): Make unsigned.
24542 (profile_count::guessed_zero): New.
24544 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
24546 * bb-reorder.c (fix_up_crossing_landing_pad,
24547 fix_crossing_conditional_branches): Use make_single_succ_edge
24548 to keep profile consistent.
24550 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
24552 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
24555 2017-07-01 Jakub Jelinek <jakub@redhat.com>
24558 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
24559 the right scopes, make sure cond_jump isn't preserved between multiple
24560 iterations. Search for fallthru edge whenever there are 3+ edges and
24561 use find_fallthru_edge for it.
24563 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
24565 Patch by Alexander Monakov <amonakov@ispras.ru>
24566 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
24567 probabilities consistently.
24569 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
24571 * pa.c (pa_expand_compare_and_swap_loop): Update call of
24572 emit_cmp_and_jump_insns.
24574 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
24577 * tree-inline.c (expand_call_inline): Combine profile statuses.
24579 2017-06-30 Andrew Pinski <apinski@cavium.com>
24581 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
24582 fold_stmt returned true.
24584 2017-06-30 Nathan Sidwell <nathan@acm.org>
24586 * ggc.h (empty_string): Delete.
24587 * cfgexpand.c (expand_asm_stmt): Use plain "".
24588 * optabs.c (expand_asm_memory_barrier): Likewise.
24589 * stringpool.c (empty_string): Delete.
24590 (digit_vector, digit_string): Delete.
24591 (ggc_alloc_string): Use plain "", don't optimize single digit
24592 strings. Use ggc_alloc_atomic.
24594 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
24596 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
24597 comparison set and one other set, use the cost of the non-comparison
24600 2017-06-30 Nathan Sidwell <nathan@acm.org>
24602 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
24605 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
24607 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
24608 loops. Remove now unneeded calls to gimple_switch_set_label() that
24609 just set removed labels to NULL_TREE.
24611 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
24613 * tree-ssanames.c (set_range_info_raw): Abstract from ...
24614 (set_range_info): ...here. Only call set_range_info_raw if domain
24616 (set_nonzero_bits): Call set_range_info_raw.
24617 * tree-ssanames.h (set_range_info_raw): New.
24619 2017-06-30 Jakub Jelinek <jakub@redhat.com>
24622 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
24623 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
24624 of nonimmediate_operand and <store_mask_constraint> instead of m
24625 for the input operand. For V8FI iterator, always split if input
24626 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
24627 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
24628 <store_mask_predicate> instead of register_operand and
24629 <store_mask_constraint> instead of v for the input operand. Make
24630 sure both operands aren't MEMs for if not <mask_applied>.
24632 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
24634 * lto-wrapper.c (copy_file) Close both file descriptors before
24637 2017-06-30 Martin Liska <mliska@suse.cz>
24640 * multiple_target.c (create_dispatcher_calls): Make ifunc
24641 also for function that don't have calls or are not referenced.
24643 2017-06-30 Richard Biener <rguenther@suse.de>
24645 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
24646 analyze the first scalar stmt. Move vector type computation
24647 for the BB case here from ...
24648 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
24649 live operation processing in the SLP case properly.
24651 2017-06-30 Richard Biener <rguenther@suse.de>
24653 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
24655 2017-06-30 Martin Liska <mliska@suse.cz>
24658 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
24659 before BUILT_IN_UNWIND_RESUME when ASAN is used.
24661 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
24663 * doc/invoke.texi (AArch64): Add missing options and remove redundant
24666 2017-06-30 Richard Biener <rguenther@suse.de>
24668 PR tree-optimization/81249
24669 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
24670 condition reduction result to original scalar type.
24672 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24674 * profile-count.h (enum profile_quality): Fix typos and whitespace
24677 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24679 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
24680 type for branch probabilities.
24682 2017-06-29 Julian Brown <julian@codesourcery.com>
24683 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24685 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
24686 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
24687 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
24688 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
24690 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24692 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
24693 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
24694 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
24695 CC usage from generic code to here.
24696 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
24697 CC usage into the target macros.
24699 2017-06-29 Maya Rashish <coypu@sdf.org>
24701 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
24704 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
24706 * arm/arm-builtins.c: Include profile-count.h
24707 * except.c (sjlj_emit_function_enter): Use
24708 profile_probability::unlikely.
24710 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24712 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
24713 and tocrel_offset be pointer args rather than implicitly using
24715 (legitimate_constant_pool_address_p, rs6000_emit_move,
24716 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
24717 tocrel_offset and use in toc_relative_expr_p call.
24718 (print_operand, print_operand_address): Use static tocrel_base_oac
24719 and tocrel_offset_oac.
24720 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
24723 2017-06-29 Maya Rashish <coypu@sdf.org>
24725 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
24727 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
24729 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
24730 objects, take into account only the alignment of 'op0' and 'mode1' if
24733 2017-06-29 Steve Ellcey <sellcey@cavium.com>
24735 * ccmp.c (ccmp_tree_comparison_p): New function.
24736 (ccmp_candidate_p): Update to use above function.
24737 (get_compare_parts): New function.
24738 (expand_ccmp_next): Update to use new functions.
24739 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
24741 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
24742 take mode as argument.
24743 * ccmp.h (expand_ccmp_expr): Add mode as argument.
24744 * expr.c (expand_expr_real_1): Pass mode as argument.
24746 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
24748 * combine.c (combine_instructions): Print insns to dump_file, together
24751 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
24753 * asan.c (asan_emit_stack_protection): Update.
24754 (create_cond_insert_point): Update.
24755 * auto-profile.c (afdo_propagate_circuit): Update.
24756 * basic-block.h (struct edge_def): Turn probability to
24757 profile_probability.
24758 (EDGE_FREQUENCY): Update.
24759 * bb-reorder.c (find_traces_1_round): Update.
24760 (better_edge_p): Update.
24761 (sanitize_hot_paths): Update.
24762 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
24763 (make_single_succ_edge): Update.
24764 (check_bb_profile): Update.
24765 (dump_edge_info): Update.
24766 (update_bb_profile_for_threading): Update.
24767 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
24768 probabilitycount to 0.
24769 * cfgbuild.c (compute_outgoing_frequencies): Update.
24770 * cfgcleanup.c (try_forward_edges): Update.
24771 (outgoing_edges_match): Update.
24772 (try_crossjump_to_edge): Update.
24773 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
24774 (expand_gimple_tailcall): Update.
24775 (construct_init_block): Use make_single_succ_edge.
24776 (construct_exit_block): Use make_single_succ_edge.
24777 * cfghooks.c (verify_flow_info): Update.
24778 (redirect_edge_succ_nodup): Update.
24779 (split_edge): Update.
24780 (account_profile_record): Update.
24781 * cfgloopanal.c (single_likely_exit): Update.
24782 * cfgloopmanip.c (scale_loop_profile): Update.
24783 (set_zero_probability): Remove.
24784 (duplicate_loop_to_header_edge): Update.
24785 * cfgloopmanip.h (loop_version): Update prototype.
24786 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
24787 (force_nonfallthru_and_redirect): Update.
24788 (update_br_prob_note): Update.
24789 (rtl_verify_edges): Update.
24790 (purge_dead_edges): Update.
24791 (rtl_lv_add_condition_to_bb): Update.
24792 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
24793 * cgraphunit.c (init_lowered_empty_function): Update.
24794 (cgraph_node::expand_thunk): Update.
24795 * cilk-common.c: Include profile-count.h
24796 * dojump.c (inv): Remove.
24797 (jumpifnot): Update.
24798 (jumpifnot_1): Update.
24799 (do_jump_1): Update.
24801 (do_jump_by_parts_greater_rtx): Update.
24802 (do_compare_rtx_and_jump): Update.
24803 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
24804 do_jump_1. do_compare_rtx_and_jump): Update prototype.
24805 * dwarf2cfi.c: Include profile-count.h
24806 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
24807 (sjlj_emit_dispatch_table): Likewise.
24808 * explow.c: Include profile-count.h
24809 * expmed.c (emit_store_flag_force): Update.
24810 (do_cmp_and_jump): Update.
24811 * expr.c (compare_by_pieces_d::generate): Update.
24812 (compare_by_pieces_d::finish_mode): Update.
24813 (emit_block_move_via_loop): Update.
24814 (store_expr_with_bounds): Update.
24815 (store_constructor): Update.
24816 (expand_expr_real_2): Update.
24817 (expand_expr_real_1): Update.
24818 * expr.h (try_casesi, try_tablejump): Update prototypes.
24819 * gimple-pretty-print.c (dump_probability): Update.
24820 (dump_profile): New.
24821 (dump_gimple_label): Update.
24822 (dump_gimple_bb_header): Update.
24823 * graph.c (draw_cfg_node_succ_edges): Update.
24824 * hsa-gen.c (convert_switch_statements): Update.
24825 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
24826 (find_if_case_1): Update.
24827 (find_if_case_2): Update.
24828 * internal-fn.c (expand_arith_overflow_result_store): Update.
24829 (expand_addsub_overflow): Update.
24830 (expand_neg_overflow): Update.
24831 (expand_mul_overflow): Update.
24832 (expand_vector_ubsan_overflow): Update.
24833 * ipa-cp.c (good_cloning_opportunity_p): Update.
24834 * ipa-split.c (split_function): Use make_single_succ_edge.
24835 * ipa-utils.c (ipa_merge_profiles): Update.
24836 * loop-doloop.c (add_test): Update.
24837 (doloop_modify): Update.
24838 * loop-unroll.c (compare_and_jump_seq): Update.
24839 (unroll_loop_runtime_iterations): Update.
24840 * lra-constraints.c (lra_inheritance): Update.
24841 * lto-streamer-in.c (input_cfg): Update.
24842 * lto-streamer-out.c (output_cfg): Update.
24843 * mcf.c (adjust_cfg_counts): Update.
24844 * modulo-sched.c (sms_schedule): Update.
24845 * omp-expand.c (expand_omp_for_init_counts): Update.
24846 (extract_omp_for_update_vars): Update.
24847 (expand_omp_ordered_sink): Update.
24848 (expand_omp_for_ordered_loops): Update.
24849 (expand_omp_for_generic): Update.
24850 (expand_omp_for_static_nochunk): Update.
24851 (expand_omp_for_static_chunk): Update.
24852 (expand_cilk_for): Update.
24853 (expand_omp_simd): Update.
24854 (expand_omp_taskloop_for_outer): Update.
24855 (expand_omp_taskloop_for_inner): Update.
24856 * omp-simd-clone.c (simd_clone_adjust): Update.
24857 * optabs.c (expand_doubleword_shift): Update.
24858 (expand_abs): Update.
24859 (emit_cmp_and_jump_insn_1): Update.
24860 (expand_compare_and_swap_loop): Update.
24861 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
24862 * predict.c (predictable_edge_p): Update.
24863 (edge_probability_reliable_p): Update.
24864 (set_even_probabilities): Update.
24865 (combine_predictions_for_insn): Update.
24866 (combine_predictions_for_bb): Update.
24867 (propagate_freq): Update.
24868 (estimate_bb_frequencies): Update.
24869 (force_edge_cold): Update.
24870 * profile-count.c (profile_count::dump): Add missing space into dump.
24871 (profile_count::debug): Add newline.
24872 (profile_count::differs_from_p): Explicitly convert to unsigned.
24873 (profile_count::stream_in): Update.
24874 (profile_probability::dump): New member function.
24875 (profile_probability::debug): New member function.
24876 (profile_probability::differs_from_p): New member function.
24877 (profile_probability::differs_lot_from_p): New member function.
24878 (profile_probability::stream_in): New member function.
24879 (profile_probability::stream_out): New member function.
24880 * profile-count.h (profile_count_quality): Rename to ...
24881 (profile_quality): ... this one.
24882 (profile_probability): New.
24883 (profile_count): Update.
24884 * profile.c (compute_branch_probabilities): Update.
24885 * recog.c (peep2_attempt): Update.
24886 * sched-ebb.c (schedule_ebbs): Update.
24887 * sched-rgn.c (find_single_block_region): Update.
24888 (compute_dom_prob_ps): Update.
24889 (schedule_region): Update.
24890 * sel-sched-ir.c (compute_succs_info): Update.
24891 * stmt.c (struct case_node): Update.
24892 (do_jump_if_equal): Update.
24893 (get_outgoing_edge_probs): Update.
24894 (conditional_probability): Update.
24895 (emit_case_dispatch_table): Update.
24896 (expand_case): Update.
24897 (expand_sjlj_dispatch_table): Update.
24898 (emit_case_nodes): Update.
24899 * targhooks.c: Update.
24900 * tracer.c (better_p): Update.
24901 (find_best_successor): Update.
24902 * trans-mem.c (expand_transaction): Update.
24903 * tree-call-cdce.c: Update.
24904 * tree-cfg.c (gimple_split_edge): Upate.
24905 (move_sese_region_to_fn): Upate.
24906 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
24907 * tree-eh.c (lower_resx): Upate.
24908 (cleanup_empty_eh_move_lp): Upate.
24909 * tree-if-conv.c (version_loop_for_if_conversion): Update.
24910 * tree-inline.c (copy_edges_for_bb): Update.
24911 (copy_cfg_body): Update.
24912 * tree-parloops.c (gen_parallel_loop): Update.
24913 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
24914 (gimple_gen_time_profiler): Update.
24915 * tree-ssa-dce.c (remove_dead_stmt): Update.
24916 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
24917 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
24918 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
24919 (unloop_loops): Update.
24920 (try_peel_loop): Update.
24921 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
24922 * tree-ssa-loop-split.c (connect_loops): Update.
24923 (split_loop): Update.
24924 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
24925 (hoist_guard): Update.
24926 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
24927 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
24928 (value_replacement): Update.
24929 * tree-ssa-reassoc.c (branch_fixup): Update.
24930 * tree-ssa-tail-merge.c (replace_block_by): Update.
24931 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
24932 (create_edge_and_update_destination_phis): Update.
24933 (compute_path_counts): Update.
24934 (recompute_probabilities): Update.
24935 (update_joiner_offpath_counts): Update.
24936 (freqs_to_counts_path): Update.
24937 (duplicate_thread_path): Update.
24938 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
24939 (struct switch_conv_info): Update.
24940 (gen_inbound_check): Update.
24941 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
24942 (vect_do_peeling): Update.
24943 (vect_loop_versioning): Update.
24944 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
24945 (optimize_mask_stores): Update.
24946 * ubsan.c (ubsan_expand_null_ifn): Update.
24947 * value-prof.c (gimple_divmod_fixed_value): Update.
24948 (gimple_divmod_fixed_value_transform): Update.
24949 (gimple_mod_pow2): Update.
24950 (gimple_mod_pow2_value_transform): Update.
24951 (gimple_mod_subtract): Update.
24952 (gimple_mod_subtract_transform): Update.
24953 (gimple_ic): Update.
24954 (gimple_stringop_fixed_value): Update.
24955 (gimple_stringops_transform): Update.
24956 * value-prof.h: Update.
24958 2017-06-29 Carl Love <cel@us.ibm.com>
24960 * config/rs6000/rs6000-c.c: Add support for built-in functions
24961 vector signed int vec_signed (vector float);
24962 vector signed long long vec_signed (vector double);
24963 vector signed int vec_signed2 (vector double, vector double);
24964 vector signed int vec_signede (vector double);
24965 vector signed int vec_signedo (vector double);
24966 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
24967 instruction generator.
24968 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
24969 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
24970 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
24972 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
24973 vunsignede_v2df): Add define_expands.
24974 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
24975 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
24976 VEC_UNSIGNEDO): Add definitions.
24977 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
24978 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
24979 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
24980 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
24981 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
24982 * config/rs6000/altivec.h (vec_signed, vec_signed2,
24983 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
24984 vec_unsignede, vec_unsignedo): Add builtin defines.
24985 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
24987 * doc/extend.texi: Update the built-in documentation file for the
24988 new built-in functions.
24990 2017-06-29 Richard Biener <rguenther@suse.de>
24992 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
24993 reduction chains to LOOP_VINFO_REDUCTIONS.
24994 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
24995 SLP reductions after processing reduction chains.
24997 2017-06-29 Nathan Sidwell <nathan@acm.org>
24999 * builtins.c (fold_builtin_FUNCTION): Use
25000 lang_hooks.decl_printable_name.
25002 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
25004 PR middle-end/81194
25005 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
25006 with only one label.
25007 * stmt.c (expand_case): Assert NCASES is greater than one.
25009 2017-06-29 Richard Biener <rguenther@suse.de>
25011 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
25013 (group_case_labels): Likewise.
25014 (find_taken_edge): Push sanity checking on val to workers...
25015 (find_taken_edge_cond_expr): ... here
25016 (find_taken_edge_switch_expr): ... and here, handle cases
25017 with just a default label.
25018 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
25019 (group_case_labels): Likewise.
25020 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
25021 group_case_labels does anything cleanup the CFG again.
25023 2017-06-29 Bin Cheng <bin.cheng@arm.com>
25025 PR tree-optimization/81196
25026 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
25027 exit condition comparing two IVs.
25029 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
25031 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
25032 profile to the dummy entry at the end of the list of architectures.
25033 * config/arm/arm-cpu-cdata.h: Regenerated.
25035 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25036 Michael Collison <michael.collison@arm.com>
25039 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
25041 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
25042 (*aarch64_reg_<mode>3_minus_mask): New pattern.
25043 (*aarch64_<optab>_reg_di3_mask2): New pattern.
25044 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
25045 of shift when the shift amount is masked with constant equal to
25046 the size of the mode.
25047 * config/aarch64/predicates.md (subreg_lowpart_operator): New
25050 2017-06-29 Martin Liska <mliska@suse.cz>
25052 * config/i386/i386.opt: Change range from [1,5] to [0,5].
25054 2017-06-29 Yury Gribov <tetra2005@gmail.com>
25057 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
25060 (edge_growth_cache_entry::edge_growth_cache_entry): New
25062 (reset_edge_growth_cache): Update to use constructor.
25064 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25066 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
25067 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
25068 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
25070 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
25072 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
25073 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
25075 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
25077 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
25078 (*-linux-uclibc*): Add t-uclibc tmake_file.
25079 * config/t-musl: New.
25080 * config/t-uclibc: New.
25082 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
25084 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
25086 (gen_comm_data): Emit architectural setting of arch_prof.
25087 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
25089 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
25090 (armv8-m.base, armv8-m.main): Likewise.
25091 * arm-protos.h (arm_build_target): Add profile field.
25092 (arch_option): Likewise.
25093 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
25095 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
25096 arm_active_target.profile.
25098 2017-06-28 Richard Biener <rguenther@suse.de>
25100 PR middle-end/81227
25101 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
25102 TYPE_OVERFLOW_WRAPS.
25103 * match.pd (negate_expr_p): Likewise.
25104 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
25105 fold_build2, not fold_binary.
25107 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
25109 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
25110 Convert memory address to Pmode.
25111 (aarch64_print_operand): Assert MEM operands are always Pmode.
25113 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
25116 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
25117 Remove redundant if.
25118 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
25119 * config/arm/aarch-common-protos.h
25120 (aarch_forward_to_shift_is_not_shifted_re): Remove.
25121 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
25123 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
25126 * multiple_target.c (create_dispatcher_calls): Set the default
25127 clone to be static, not public.
25129 2017-06-28 Richard Biener <rguenther@suse.de>
25131 * tree-vect-loop.c (vectorizable_reduction): Move special
25132 cond reduction IV var creation ...
25133 (vect_create_epilog_for_reduction): ... here. Remove induction_index
25134 parameter. Use STMT_VINFO_VECTYPE.
25135 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
25138 2017-06-28 Martin Liska <mliska@suse.cz>
25141 * ipa-visibility.c (non_local_p): Handle visibility.
25143 2017-06-28 Martin Liska <mliska@suse.cz>
25146 * common.opt: Add IntegerRange to various options.
25147 * opt-functions.awk (integer_range_info): New function.
25148 * optc-gen.awk: Add integer_range_info to cl_options struct.
25149 * opts-common.c (decode_cmdline_option): Handle
25150 CL_ERR_INT_RANGE_ARG.
25151 (cmdline_handle_error): Likewise.
25152 * opts.c (print_filtered_help): Show valid interval in
25153 when --help is provided.
25154 * opts.h (struct cl_option): Add range_min and range_max fields.
25155 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
25157 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
25159 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
25160 (x * C EQ/NE y * C): New transformation.
25162 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
25164 * genmultilib (combination_space): Accept '+' in option names.
25166 2017-06-28 Martin Liska <mliska@suse.cz>
25169 * asan.c (instrument_derefs): Bail out inner references
25170 that are hard register variables.
25172 2017-06-28 Jakub Jelinek <jakub@redhat.com>
25175 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
25176 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
25178 2017-06-28 Richard Biener <rguenther@suse.de>
25180 * tree-vectorizer.h (vect_get_vec_defs): Remove.
25181 (vect_get_slp_defs): Adjust.
25182 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
25184 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
25186 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
25187 get_initial_defs_for_reduction instead of vect_get_vec_defs.
25188 (vectorizable_reduction): Adjust.
25189 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
25191 (vect_get_slp_defs): Likewise.
25192 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
25193 (vectorizable_bswap): Adjust.
25194 (vectorizable_call): Likewise.
25195 (vectorizable_conversion): Likewise.
25196 (vectorizable_assignment): Likewise.
25197 (vectorizable_shift): Likewise.
25198 (vectorizable_operation): Likewise.
25199 (vectorizable_store): Likewise.
25200 (vectorizable_condition): Likewise.
25201 (vectorizable_comparison): Likewise.
25203 2017-06-28 Michael Collison <michael.collison@arm.com>
25206 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
25208 (arm_gen_movmemqi): Removed unused variable 'i'.
25209 Convert 'for' loop into 'while' loop.
25210 (arm_expand_prologue): Remove last unnecessary set of insn.
25211 (thumb_pop): Remove unused variable 'pushed_words'.
25212 (thumb_exit): Remove last unnecessary set of regs_to_pop.
25214 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25216 * config/s390/predicates.md: Use s390_rel_address_ok_p.
25217 * config/s390/s390-protos.h: Add prototype of
25218 s390_rel_address_ok_p.
25219 * config/s390/s390.c (s390_got_symbol): New function.
25220 (s390_rel_address_ok_p): New function.
25221 (legitimize_pic_address): Use s390_rel_address_ok_p.
25222 (s390_load_got): Use s390_got_symbol.
25223 (s390_option_override): Issue error if
25224 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
25225 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
25227 * config/s390/s390.opt: New option mpic-data-is-text-relative.
25229 2017-06-27 Andrew Pinski <apinski@cavium.com>
25231 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
25232 (X * copysign (1.0, X)): New pattern.
25233 (X * copysign (1.0, -X)): New pattern.
25234 (copysign (-1.0, CST)): New pattern.
25236 2017-06-27 Joseph Myers <joseph@codesourcery.com>
25238 * genmultilib (combination_space): Remove variable.
25239 Validate reuse rules against regular expression for any sequence
25240 of multilib options in any order.
25242 2017-06-27 Michael Collison <michael.collison@arm.com>
25244 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
25245 call aarch64_split_simd_combine.
25246 * (aarch64_combine_internal<mode>): Delete pattern.
25247 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
25248 Allow register and subreg operands.
25250 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25252 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
25253 specific need, just fallback on defaults.
25254 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
25256 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25257 Olivier Hainque <hainque@adacore.com>
25259 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
25261 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
25262 targets. Pick a default if no particular attempt applied.
25263 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
25266 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25268 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
25269 (x86_64-wrs-vxworks7): Likewise.
25271 2017-06-27 Marek Polacek <polacek@redhat.com>
25274 * ubsan.c (instrument_null): Check get_base_address's result for null.
25276 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
25278 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
25280 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
25282 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
25283 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
25284 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
25285 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
25286 New function types.
25287 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
25288 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
25289 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
25290 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
25291 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
25292 BUILT_IN_FEUPDATEENV): New builtins.
25293 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
25294 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
25295 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
25296 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
25298 (builtin_structptr_types): Adjust size.
25299 * tree.c (builtin_structptr_types): Add four entries.
25301 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25302 Olivier Hainque <hainque@adacore.com>
25304 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
25305 (TLS_SYM): New local macro, forcing reference to __tls__ on
25306 link command lines for VxWorks 7 RTPs, triggering initialization
25308 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
25309 OS features TLS support, true for RTPs on VxWorks 7.
25310 * config/vxworks.c (vxworks_override_options): Setup emutls
25313 2017-06-27 Jakub Jelinek <jakub@redhat.com>
25315 * predict.c (test_prediction_value_range): Use -1U instead of -1
25316 to avoid narrowing conversion warning.
25317 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
25318 to avoid narrowing conversion warning.
25319 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
25321 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
25323 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25325 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
25326 64bit configurations.
25327 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
25328 (SIZE_TYPE): Likewise.
25329 * config/vxworks.c (vxworks_emutls_var_fields): Use
25330 long_unsigned_type_node instead of unsigned_type_node as the offset
25331 field type, which is "pointer" mode in emutls.c.
25333 2017-06-27 Jakub Jelinek <jakub@redhat.com>
25336 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
25338 PR middle-end/81207
25339 * gimple-fold.c (replace_call_with_call_and_fold): Handle
25340 gimple_vuse copying separately from gimple_vdef copying.
25342 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25344 * value-prof.c (free_hist): Remove call to memset and the enclosing if
25347 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
25348 Olivier Hainque <hainque@adacore.com>
25350 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
25351 for all vxworks7 targets.
25352 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
25353 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
25354 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
25355 variations for VX6/VX7 and 32/64bits later on in ...
25356 (VXWORKS_LIB_SPEC): Leverage new macros.
25357 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
25358 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
25360 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
25362 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
25363 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
25365 2017-06-26 Carl Love <cel@us.ibm.com>
25367 * config/rs6000/rs6000-c.c: Add support for built-in functions
25368 vector bool char vec_reve (vector bool char);
25369 vector signed char vec_reve (vector signed char);
25370 vector unsigned char vec_reve (vector unsigned char);
25371 vector bool int vec_reve (vector bool int);
25372 vector signed int vec_reve (vector signed int);
25373 vector unsigned int vec_reve (vector unsigned int);
25374 vector bool long long vec_reve (vector bool long long);
25375 vector signed long long vec_reve (vector signed long long);
25376 vector unsigned long long vec_reve (vector unsigned long long);
25377 vector bool short vec_reve (vector bool short);
25378 vector signed short vec_reve (vector signed short);
25379 vector double vec_reve (vector double);
25380 vector float vec_reve (vector float);
25381 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
25382 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
25383 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
25384 (altivec_vreve): New pattern.
25385 * config/rs6000/altivec.h (vec_reve): New define.
25386 * doc/extend.texi (vec_rev): Update the built-in documentation file
25387 for the new built-in functions.
25389 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25391 PR tree-optimization/71815
25392 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
25394 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
25396 (slsr_process_phi): Likewise.
25397 (replace_uncond_cands_and_profitable_phis): Don't replace a
25398 multiply candidate with a stride of 1 (copy or cast).
25399 (phi_incr_cost): Call uses_consumed_by_stmt rather than
25401 (lowest_cost_path): Likewise.
25402 (total_savings): Likewise.
25404 2017-06-26 Richard Biener <rguenther@suse.de>
25407 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
25408 Use def_builtin_pure for all gather builtins.
25410 2017-06-26 Richard Biener <rguenther@suse.de>
25412 PR tree-optimization/81203
25413 * tree-tailcall.c (find_tail_calls): Do not move stmts into
25414 non-dominating BBs.
25416 2017-06-26 Marek Polacek <polacek@redhat.com>
25419 * doc/invoke.texi: Document -Wmultistatement-macros.
25421 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
25423 * doc/sourcebuild.texi (ARM-specific attributes): Document new
25424 arm_neon_ok_no_float_abi effective target.
25426 2017-06-26 Richard Biener <rguenther@suse.de>
25428 PR tree-optimization/80928
25429 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
25430 (copy_bbs): Set BB_DUPLICATED flag early.
25431 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
25433 (execute_on_shrinking_pred): Likewise.
25434 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
25435 BB_DUPLICATED blocks.
25436 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
25437 iterate over all PHIs considering removal of *gsi.
25439 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
25441 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
25444 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25446 * config/rs6000/rs6000-string.c: (expand_block_clear,
25447 do_load_for_compare, select_block_compare_mode,
25448 compute_current_alignment, expand_block_compare,
25449 expand_strncmp_align_check, expand_strn_compare,
25450 expand_block_move, rs6000_output_load_multiple)
25451 Move functions related to string/block move/compare
25452 to a separate file.
25453 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
25454 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
25455 for this function which is now used in two files.
25456 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
25457 * config.gcc: Add rs6000-string.o to extra_objs for
25458 targets powerpc*-*-* and rs6000*-*-*.
25460 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
25463 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
25464 32-bit, since indexed is not valid for DImode.
25465 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
25466 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
25467 (define_peephole2 for Altivec d-form load): Add 32-bit support.
25468 (define_peephole2 for Altivec d-form store): Likewise.
25471 * multiple_target.c (create_dispatcher_calls): Only create the
25472 dispatcher call if the function is the default clone of a
25473 versioned function.
25475 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
25477 PR middle-end/80902
25478 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
25479 a call, force the call to not be a tail call.
25481 2017-06-23 Jeff Law <law@redhat.com>
25483 * doc/contrib.texi: Add entry for Steven Pemberton's work on
25486 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
25488 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
25489 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
25490 handling for early expansion of vector shifts (sl,sr,sra,rl).
25491 (builtin_function_type): Add vector shift right instructions
25492 to the unsigned argument list.
25494 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
25496 rtl-optimizatoin/79286
25497 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
25498 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
25499 trap. PIC register plus a const unspec without offset can never trap.
25501 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
25503 * tree.h (builtin_structptr_type): New type.
25504 (builtin_structptr_types): Declare new array.
25505 * tree.c (builtin_structptr_types): New array.
25506 (free_lang_data, build_common_tree_nodes): Use it.
25508 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
25511 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
25514 2017-06-22 Matt Turner <mattst88@gmail.com>
25516 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
25517 Lake models to skylake case. Assume skylake for unknown
25518 models with clflushopt.
25520 2017-06-22 Jeff Law <law@redhat.com>
25522 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
25523 frame sizes that do not satisfy aarch64_uimm12_shift.
25525 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
25527 * profile-count.h (apply_probability,
25528 apply_scale, probability_in): Fix checks for zero.
25530 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
25532 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
25533 * doc/cppdiropts.texi (-I @var{dir}): Document it.
25535 2016-06-22 Richard Biener <rguenther@suse.de>
25537 * tree-vect-loop.c (vect_model_reduction_cost): Handle
25538 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
25539 REDUC_MAX_EXPR support.
25540 (vectorizable_reduction): Likewise.
25541 (vect_create_epilog_for_reduction): Likewise.
25543 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
25545 * match.pd (A / (1 << B) -> A >> B): New.
25546 * generic-match-head.c: Include optabs-tree.h.
25547 * gimple-match-head.c: Likewise.
25548 * optabs-tree.h (target_supports_op_p): New.
25549 * optabs-tree.c (target_supports_op_p): New.
25551 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
25553 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
25554 $gcc_cv_ld --help output.
25555 (gcc_cv_ld_demangle): Likewise.
25556 (gcc_cv_ld_eh_frame_hdr): Likewise.
25557 (gcc_cv_ld_pie): Likewise.
25558 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
25559 (gcc_cv_ld_buildid): Likewise.
25560 (gcc_cv_ld_sysroot): Likewise.
25561 (ld_bndplt_support): Likewise.
25562 (ld_pushpopstate_support): Likewise.
25563 * configure: Regenerate.
25564 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
25566 2017-06-21 Jakub Jelinek <jakub@redhat.com>
25569 * config/i386/sse.md (round<mode>2): Renumber match_dup and
25570 operands indexes to avoid gap between operands and match_dups.
25572 2017-06-21 Andrew Pinski <apinski@cavium.com>
25574 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
25575 Increment Arith_shift and Arith_shift_reg by 1.
25576 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
25578 * config/aarch64/aarch64.c (thunderx_tunings): Enable
25579 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
25580 (aarch64_strip_extend): Add new argument and test for it.
25581 (aarch64_cheap_mult_shift_p): New function.
25582 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
25583 add a cost if it is true.
25584 Update calls to aarch64_strip_extend.
25585 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
25587 2017-06-21 Andrew Pinski <apinski@cavium.com>
25589 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
25591 (thunderxt88): Likewise.
25592 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
25593 (thunderx_prefetch_tune): New variable.
25594 (thunderx2t99_prefetch_tune): Update for the correct values.
25595 (thunderxt88_tunings): New variable.
25596 (thunderx_tunings): Use thunderx_prefetch_tune instead of
25597 generic_prefetch_tune.
25598 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
25600 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25602 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
25603 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
25604 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
25605 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
25606 (aarch64_atomic_cas<mode>, GPI): Likewise.
25608 2017-06-21 Martin Liska <mliska@suse.cz>
25610 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
25611 statements on cold and hot labels.
25612 * predict.c (tree_estimate_probability_bb): Remove the
25613 prediction from this place.
25615 2017-06-21 Martin Liska <mliska@suse.cz>
25617 PR tree-optimization/79489
25618 * gimplify.c (maybe_add_early_return_predict_stmt): New
25620 (gimplify_return_expr): Call the function.
25621 * predict.c (tree_estimate_probability_bb): Remove handling
25623 * predict.def: Update comment about early return predictor.
25624 * gimple-predict.h (is_gimple_predict): New function.
25625 * predict.def: Change default value of early return to 66.
25626 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
25628 * passes.def: Put pass_strip_predict_hints to the beginning of
25631 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
25633 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
25634 FUNCTION_DECL declarations.
25635 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
25637 (dwaf2out_decl): Likewise.
25638 * godump.c (go_early_global_decl): Skip call to the real debug hook
25639 for FUNCTION_DECL declarations.
25640 * passes.c (rest_of_decl_compilation): Skip call to the
25641 early_global_decl debug hook for FUNCTION_DECL declarations, unless
25642 -fdump-go-spec is passed.
25644 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
25646 * config/i386/i386.c (struct builtin_isa): New field pure_p.
25647 Reorder for compactness.
25648 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
25649 (def_builtin_pure, def_builtin_pure2): New functions.
25650 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
25652 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
25654 * match.pd (nop_convert): New predicate.
25655 ((A +- CST1) +- CST2): Allow some NOP conversions.
25657 2017-06-21 Jakub Jelinek <jakub@redhat.com>
25660 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
25661 with ctors/dtors if GOVD_SHARED is set.
25663 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
25665 * config/aarch64/aarch64.md (movti_aarch64):
25666 Emit mov rather than orr.
25667 (movtf_aarch64): Likewise.
25668 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
25669 Emit mov rather than orr.
25671 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
25673 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
25674 Swap alternatives, make integer dup more expensive.
25676 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
25678 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
25679 Return true for non-tls symbols.
25681 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
25683 * config/aarch64/aarch64-cores.def (cortex-a55): New.
25684 (cortex-a75): Likewise.
25685 (cortex-a75.cortex-a55): Likewise.
25686 * config/aarch64/aarch64-tune.md: Regenerate.
25687 * doc/invoke.texi (-mtune): Document new values for -mtune.
25689 2017-06-21 Tom de Vries <tom@codesourcery.com>
25691 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
25692 stack_size feature.
25693 (Effective-Target Keywords, Other attributes): Suggest using
25694 dg-add-options stack_size feature to get stack limit in stack_size
25695 effective target documentation.
25697 2017-06-21 Julian Brown <julian@codesourcery.com>
25698 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25700 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
25701 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
25702 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
25704 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
25705 attribute type list for neon_multiply.
25706 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
25707 attribute type list for neon_multiply.
25708 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
25709 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
25710 attribute type list for neon_multiply.
25711 * config/arm/types.md (crypto_pmull): Add.
25712 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
25713 attribute type list.
25715 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
25717 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
25720 2017-06-20 Jakub Jelinek <jakub@redhat.com>
25722 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
25723 to make sure not to dereference a NULL cost_classes_ptr pointer.
25725 2017-06-20 Carl Love <cel@us.ibm.com>
25727 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
25728 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
25729 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
25730 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
25731 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
25732 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
25733 VMULOSW): New enum "unspec" values.
25734 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
25735 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
25736 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
25737 altivec_vmulosw): New patterns.
25738 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
25739 VMULOSW): Add definitions.
25741 2017-06-20 Julia Koval <julia.koval@intel.com>
25743 * config/i386/i386.c: Fix rounding expand for new pattern.
25744 * config/i386/subst.md: Fix pattern (parallel -> unspec).
25746 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
25748 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
25749 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
25751 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
25753 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
25756 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
25758 * config/aarch64/aarch64-cores.def: Rearrange to sort by
25759 architecture, then by implementer ID.
25760 * config/aarch64/aarch64-tune.md: Regenerate.
25762 2017-06-20 Richard Biener <rguenther@suse.de>
25764 PR middle-end/81097
25765 * fold-const.c (split_tree): Fold to type before negating.
25767 2017-06-20 David Malcolm <dmalcolm@redhat.com>
25769 * diagnostic-show-locus.c
25770 (selftest::test_fixit_deletion_affecting_newline): New function.
25771 (selftest::diagnostic_show_locus_c_tests): Call it.
25773 2017-06-20 Andreas Schwab <schwab@suse.de>
25776 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
25779 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25781 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
25782 __ARM_FEATURE_COPROC according to support.
25784 2017-06-20 Jakub Jelinek <jakub@redhat.com>
25786 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
25787 Rewritten to avoid overflow for > 32-bit pointers.
25790 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
25791 by removing enum keyword.
25792 (ubsan_type_descriptor): Likewise. Formatting fix.
25795 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
25796 splitter): Require TARGET_SSE2 in the condition.
25798 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
25801 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
25802 for doing vector set of SFmode on ISA 3.0.
25803 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
25804 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
25806 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
25807 SFmode value into a V4SF variable that was extracted from another
25808 V4SF variable without converting the element to double precision
25809 and back to single precision vector format.
25810 (vsx_insert_extract_v4sf_p9_2): Likewise.
25812 2017-06-19 Jakub Jelinek <jakub@redhat.com>
25814 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
25815 in UWHI to avoid undefined overflow.
25818 * ubsan.h (enum ubsan_encode_value_phase): New.
25819 (ubsan_encode_value): Change second argument to
25820 enum ubsan_encode_value_phase with default value of
25821 UBSAN_ENCODE_VALUE_GENERIC.
25822 * ubsan.c (ubsan_encode_value): Change second argument to
25823 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
25824 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
25825 create_tmp_var_raw instead of create_tmp_var and use a
25827 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
25828 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
25829 ubsan_encode_value callers.
25832 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
25833 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
25834 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
25836 2017-06-19 Richard Biener <rguenther@suse.de>
25838 PR middle-end/81118
25839 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
25840 estimates if we changed anything.
25842 2017-06-19 Richard Biener <rguenther@suse.de>
25844 PR tree-optimization/80887
25845 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
25846 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
25847 simplified lookups, then reset mprts_hook.
25848 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
25850 (try_to_simplify): Likewise.
25852 2017-06-19 Martin Liska <mliska@suse.cz>
25855 * gimplify.c (gimplify_switch_expr):
25856 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
25858 2017-06-19 Martin Liska <mliska@suse.cz>
25860 * doc/install.texi: Document that PGO runs in 4 stages.
25862 2017-06-19 Martin Liska <mliska@suse.cz>
25865 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
25866 to dispatcher function name.
25867 * multiple_target.c (replace_function_decl): New function.
25868 (create_dispatcher_calls): Redirect both edges and references.
25870 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
25872 * profile-count.c (profile_count::dump): Dump quality.
25873 (profile_count::differs_from_p): Update for unsigned val.
25874 * profile-count.h (profile_count_quality): New enum.
25875 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
25877 2017-06-19 Richard Biener <rguenther@suse.de>
25879 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
25880 struct function as arg.
25881 (estimate_numbers_of_iterations): Export overload with loop arg.
25882 (free_numbers_of_iterations_estimates_loop): Use an overload of
25883 free_numbers_of_iterations_estimates instead.
25884 * tree-cfg.c (remove_bb): Adjust.
25885 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
25886 * tree-parloops.c (gen_parallel_loop): Likewise.
25887 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
25889 (tree_unroll_loops_completely): Likewise.
25890 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
25891 Use an overload instead and export.
25892 (estimated_loop_iterations): Adjust.
25893 (max_loop_iterations): Likewise.
25894 (likely_max_loop_iterations): Likewise.
25895 (estimate_numbers_of_iterations): Take struct function as arg
25897 (loop_exits_before_overflow): Adjust.
25898 (free_numbers_of_iterations_estimates_loop): Use an overload.
25899 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
25900 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
25902 2017-06-19 Richard Biener <rguenther@suse.de>
25905 * ipa-prop.c (find_constructor_constant_at_offset): Handle
25906 RANGE_EXPR conservatively.
25908 2017-06-16 Carl Love <cel@us.ibm.com>
25910 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
25911 definitions for vec_float, vec_float2, vec_floato,
25912 vec_floate built-ins.
25913 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
25914 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
25916 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
25917 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
25918 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
25919 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
25920 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
25921 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
25922 vec_floato): Add builtin defines.
25923 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
25924 Update the built-in documentation file for the new built-in
25927 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25929 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
25930 (mthumb): Mark as the negative of -marm.
25932 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25934 * doc/invoke.texi (ARM Options, -mcpu): Document supported
25936 (ARM Options, -mtune): Document that this accepts the same
25937 extension options as -mcpu.
25938 (ARM Options, -mfpu): Document addition of -mfpu=auto.
25940 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25942 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
25943 permitted extensions.
25945 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25947 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
25948 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
25949 (armv8-m.main): Add option +nodsp.
25950 * config/arm/arm-cpu-cdata.h: Regenerated.
25952 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25954 * config/arm/t-fuchsia: New file.
25955 * config.gcc (arm*-*-fuchsia*): Use it.
25957 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25959 * config/arm/t-symbian: Rewrite for new option infrastructure.
25961 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25963 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
25964 (MULTILIB_REQUIRED): Likewise.
25966 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25968 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
25969 (MULTILIB_RESUE): Likewise.
25970 (MULTILIB_MATCHES): Likewise.
25971 (MULTLIB_REQUIRED): Likewise.
25973 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25975 * config/arm/t-rtems: Rewrite for new option framework.
25977 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25979 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
25980 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
25981 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
25982 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
25983 * config/arm/t-multilib: ... here.
25984 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
25985 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
25986 armv7-a and armv8*-a when A-profile libraries have not been built.
25987 * config/arm/t-rmprofile: Rewrite.
25989 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25991 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
25992 with a backslash. Remove the backslash after substituting unescaped
25994 * doc/fragments.texi (MULTILIB_REUSE): Document it.
25996 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25998 * config.gcc: (arm*-*-*): When building a-profile libraries, force
25999 the driver to pass through the default setting of -mfloat-abi.
26000 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
26002 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
26003 (all_feat_combs): New rule.
26004 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
26006 * config/arm/t-aprofile: Rewrite.
26008 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26010 * config/arm/arm.h (FPUTYPE_AUTO): Define.
26011 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
26012 fpu is not specified by the user/command-line.
26013 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
26014 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
26015 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
26016 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
26017 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
26018 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
26020 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26022 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
26023 * config/arm/t-arm-elf: Rewritten.
26025 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26027 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
26028 have some floating-point instructions.
26029 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
26030 (TARGET_MAYBE_HARD_FLOAT): New macro.
26031 * config/arm/arm-builtins.c (arm_init_builtins): Use
26032 TARGET_MAYBE_HARD_FLOAT.
26033 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
26035 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26037 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
26038 (configargs.h): Include it.
26039 (arm_print_hint_for_fpu_option): New function.
26040 (arm_parse_fpu_option): New function.
26041 (candidate_extension): New class.
26042 (arm_canon_for_multilib): New function.
26043 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
26044 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
26045 (ARCH_CANONICAL_SPECS): New macro.
26046 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
26048 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26050 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
26051 are set after handling multilib fragments. Set target_cpu_default2
26054 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26056 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
26058 (arm*-*-*): Set target_cpu_default2 to a quoted string.
26059 * config/arm/parsecpu.awk (check_cpu): Validate any extension
26061 (check_arch): Likewise.
26062 * config/arm/arm.c (arm_configure_build_target): Handle
26063 TARGET_CPU_DEFAULT being a string constant. Scan any feature
26064 options in the default.
26066 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26068 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
26069 when an option is an alias of another.
26070 * config/arm/parsecpu.awk (optalias): New parser token.
26071 (gen_comm_data): Mark non-alias options as such. Emit entries
26072 for extension aliases.
26073 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
26074 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
26075 (armv6kz, armv6zk, armv6t2): Likewise.
26076 (armv7): Make vfpv3-d16 an alias.
26077 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
26079 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
26080 Sort in canonical order.
26081 (armv8-a): Sort in canonical order.
26082 (armv8.1-a, armv8.2-a): Likewise.
26083 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
26085 (cortex-a9): Sort in canonical order.
26086 * config/arm/arm.c (selftests.h): Include it.
26087 (arm_test_cpu_arch_data): New function.
26088 (arm_run_self_tests): New function.
26089 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
26090 (targetm): Move declaration to the end of the file.
26091 * arm-cpu-cdata.h: Regenerated.
26093 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26095 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
26096 call to target_mode_check describing the type of option passed.
26097 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
26098 (arm_target_thumb_only): Use arm_parse_arch_option_name or
26099 arm_parse_cpu_option_name to match parameters against list of
26101 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
26102 arm_arch_core_flags data structure.
26103 * config/arm/arm-cpu_cdata.h: Regenerated.
26105 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26107 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
26109 (arm_print_hint_for_cpu_option): Likewise.
26110 (arm_print_hint_for_arch_option): Likewise.
26111 (arm_parse_cpu_option_name): Likewise.
26112 (arm_parse_arch_option_name): Likewise.
26113 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
26114 of entries in the all_fpus list.
26115 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
26116 (arm_parse_cpu_option_name): Declare.
26117 (arm_parse_arch_option_name): Declare.
26118 (arm_parse_option_features): Declare.
26119 (arm_intialize_isa): Declare.
26120 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
26122 (gen_comm_data): ... here. Make definitions non-static.
26123 * config/arm/arm-cpu-data.h: Regenerated.
26124 * config/arm/arm-cpu-cdata.h: Regenerated.
26126 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26128 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
26129 (cpu_arch_extension): New structure.
26130 (cpu_arch_option, arch_option, cpu_option): New structures.
26131 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
26132 architecture types.
26133 (gen_data): Generate new format data tables.
26134 * config/arm/arm.c (cpu_tune): New structure.
26135 (cpu_option, processors): Delete.
26136 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
26137 (arm_print_hint_for_cpu_option): ... this and ...
26138 (arm_print_hint_for_arch_option): ... this.
26139 (arm_parse_arch_cpu_name): Delete. Replace with ...
26140 (arm_parse_cpu_option_name): ... this and ...
26141 (arm_parse_arch_option_name): ... this.
26142 (arm_unrecognized_feature): Change type of target parameter to
26144 (arm_parse_arch_cpu_features): Delete. Replace with ...
26145 (arm_parse_option_features): ... this.
26146 (arm_configure_build_target): Rework to use new configuration data
26148 (arm_print_tune_info): Rework for new configuration data tables.
26149 * config/arm/arm-cpu-data.h: Regenerated.
26150 * config/arm/arm-cpu.h: Regenerated.
26152 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26154 * Makefile.in (OBJS): Move sbitmap.o from here ...
26155 (OBJS-libcommon): ... to here.
26157 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26159 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
26160 (ISA_ALL_CRYPTO): New macro.
26161 (ISA_ALL_SIMD): New macro
26162 (ISA_ALL_FP): New macro.
26163 * config/arm/arm.c (fpu_bitlist): Update initializer.
26164 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
26166 (arm9e): Add fpu. Add option for nofp
26167 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
26168 (arm926ej-s, arm1026ej-s): Likewise.
26169 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
26170 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
26171 neon-fp16, neon-vfpv4, nofp and nosimd.
26172 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
26173 (cortex-a8): Add fpu. Add option for nofp.
26174 (cortex-a9): Add fpu. Add options for nosimd and nofp.
26175 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
26176 (cortex-r4f): Add fpu.
26177 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
26178 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
26180 (cortex-r8): Likewise.
26181 (cortex-m4): Add fpu. Add option for nofp.
26182 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
26183 (cortex-a17.cortex-a7): Likewise.
26184 (cortex-a32): Add fpu. Add options for crypto and nofp.
26185 (cortex-a35, cortex-a53): Likewise.
26186 (cortex-a57): Add fpu. Add option for crypto.
26187 (cortex-a72, cortex-a73): Likewise.
26188 (exynos-m1): Likewise.
26189 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
26190 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
26191 (cortex-m33): Add fpu. Add option for nofp.
26192 * config/arm/arm-cpu-cdata.h: Regenerated
26193 * config/arm/arm-cpu-data.h: Regenerated.
26195 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26197 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
26198 (armv5te, armv5tej): Likewise.
26199 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
26200 (armv7): Add options fp and vfpv3-d16.
26201 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
26202 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
26204 (armv7ve): Likewise.
26205 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
26206 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
26207 (armv8-a): Add nocrypto option.
26208 (armv8.1-a, armv8.2-a): Likewise.
26209 (armv8-m.main): add options fp, fp.dp and nofp.
26211 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26213 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
26215 (armv8-a+crc): Delete.
26216 (armv8.1-a): Add options simd, crypto and nofp.
26217 (armv8.2-a): Add options fp16, simd, crypto and nofp.
26218 (armv8.2-a+fp16): Delete.
26219 (armv8-m.main): Add option dsp.
26220 (armv8-m.main+dsp): Delete.
26221 (cortex-a8): Add fpu. Add nofp option.
26222 (cortex-a9): Add fpu. Add nofp and nosimd options.
26223 * config/arm/parsecpu.awk (gen_data): Generate option tables and
26224 link to main cpu and architecture data structures.
26225 (gen_comm_data): Only put isa attributes from the main architecture
26227 (option): New statement for architecture and CPU entries.
26228 * arm.c (struct cpu_option): New structure.
26229 (struct processors): Add entry for options.
26230 (arm_unrecognized_feature): New function.
26231 (arm_parse_arch_cpu_name): Ignore any characters after the first
26233 (arm_parse_arch_cpu_feature): New function.
26234 (arm_configure_build_target): Separate out any CPU and architecture
26235 features and parse separately. Don't error out if -mfpu=auto is
26236 used with only an architecture string.
26237 (arm_print_asm_arch_directives): New function.
26238 (arm_file_start): Call it.
26239 * config/arm/arm-cpu-cdata.h: Regenerated.
26240 * config/arm/arm-cpu-data.h: Likewise.
26241 * config/arm/arm-tables.opt: Likewise.
26243 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26245 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
26246 assembler when it is not -mfpu=auto.
26248 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26250 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
26251 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
26252 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
26253 (ASM_CPU_SPEC): Rewrite.
26254 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
26255 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
26256 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
26257 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
26258 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
26259 copied string is NUL-terminated. Also strip any characters prefixed
26261 (arm_rewrite_selected_arch): New function.
26262 (arm_rewrite_march): New function.
26264 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26266 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
26267 (x_arm_cpu_string, x_arm_tune_string): Likewise.
26268 (march, mcpu, mtune): Convert to string-based options.
26269 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
26270 (arm_parse_arch_cpu_name): New function.
26271 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
26272 identify selected architecture or CPU.
26273 (arm_option_save): New function.
26274 (TARGET_OPTION_SAVE): Redefine.
26275 (arm_option_restore): Restore string options.
26276 (arm_option_print): Print string options.
26278 2017-06-16 Martin Sebor <msebor@redhat.com>
26280 PR tree-optimization/80933
26281 PR tree-optimization/80934
26282 * builtins.c (fold_builtin_3): Do not handle bcmp here.
26283 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
26284 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
26285 (gimple_fold_builtin): Call them.
26287 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26289 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
26290 as unlikely; update profile.
26292 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26294 * predict.c (force_edge_cold): Handle declaring edges impossible
26297 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26299 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
26301 (try_unroll_loop_completely): Fix reporting.
26303 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26305 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
26307 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
26310 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
26311 if given a non-constant argument for an intrinsic which requires a
26314 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26316 * profile.c (compare_freqs): New function.
26317 (branch_prob): Sort edge list.
26318 (find_spanning_tree): Assume that the list is priority sorted.
26320 2017-06-16 Richard Biener <rguenther@suse.de>
26322 PR tree-optimization/81090
26323 * passes.def (pass_record_bounds): Remove.
26324 * tree-pass.h (make_pass_record_bounds): Likewise.
26325 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
26326 make_pass_record_bounds): Likewise.
26327 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
26328 not free niter estimates at the beginning but at the end.
26329 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
26331 2017-06-16 Richard Biener <rguenther@suse.de>
26333 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
26334 initializer to workaround ICE in host GCC 4.8.
26336 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26338 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
26340 (clone_inlined_nodes): Update.
26342 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
26344 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
26345 prefetch settings, and enable prefetching by default at -O3.
26347 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
26349 * config/aarch64/aarch64.c (aarch64_override_options_internal):
26350 Set flag_prefetch_loop_arrays according to tuning data.
26352 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
26354 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
26355 New tune structure.
26356 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
26357 [Unrelated to main purpose of the patch] Place the pointer field last
26358 to enable type checking errors when tune structure are wrongly merged.
26359 * config/aarch64/aarch64.c (generic_prefetch_tune,)
26360 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
26361 (thunderx2t99_prefetch_tune): New tune constants.
26362 (tune_params *_tunings): Update all tunings (no functional change).
26363 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
26364 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
26365 from tunings structures.
26367 2017-06-16 Jakub Jelinek <jakub@redhat.com>
26370 * ubsan.c (instrument_null): Add T argument, use it instead
26371 of computing it based on IS_LHS.
26372 (instrument_object_size): Likewise.
26373 (pass_ubsan::execute): Adjust instrument_null and
26374 instrument_object_size callers to pass gimple_get_lhs or
26375 gimple_assign_rhs1 result to it. Use instrument_null instead of
26376 calling get_base_address and instrument_mem_ref. Handle
26377 aggregate call arguments for object-size sanitization.
26379 2017-06-16 Yury Gribov <tetra2005@gmail.com>
26381 PR tree-optimization/81089
26382 * tree-vrp.c (is_masked_range_test): Validate operands of
26385 2017-06-15 Martin Sebor <msebor@redhat.com>
26388 * dumpfile.c (dump_register): Avoid calling memset to initialize
26389 a class with a default ctor.
26390 * gcc.c (struct compiler): Remove const qualification.
26391 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
26392 * hash-table.h: Ditto.
26393 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
26395 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
26396 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
26398 * params.h (struct param_info): Make struct members non-const.
26399 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
26400 with default initialization.
26401 * vec.h (vec_copy_construct, vec_default_construct): New helper
26403 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
26404 with vec_copy_construct.
26405 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
26406 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
26407 * doc/invoke.texi (-Wclass-memaccess): Document.
26409 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26411 * emit-rtl.h (is_leaf): Update comment about local
26412 register allocator.
26414 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
26417 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
26418 for a variable to have a section before checking if the section has a
26420 Set section to.persistent if persistent attribute is set.
26421 Warn if .persistent attribute is used on an automatic variable.
26423 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
26425 PR rtl-optimization/80474
26426 * reorg.c (update_block): Do not ignore instructions in a delay slot.
26428 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
26430 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
26433 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
26435 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
26436 (casesi): Emit bounds checking as RTL.
26437 (casesi_internal_mips16_<mode>): Remove bounds checking.
26439 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
26441 * config/xtensa/xtensa.c (xtensa_option_override): Append
26442 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
26443 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
26444 xtensa_doloop_hooks): Define unconditionally.
26445 (xtensa_reorg_loops): Only call reorg_loops in the presence of
26447 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
26448 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
26449 for it in xtensa_option_override.
26450 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
26451 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
26453 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
26455 * doc/cppopts.texi: Document '-' special value to -MF.
26457 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
26459 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
26460 (cortex_a53_fconst): Likewise.
26461 (cortex_a53_fpmul): Likewise.
26462 (cortex_a53_f_load_64): Likewise.
26463 (cortex_a53_f_load_many): Likewise.
26464 (cortex_a53_advsimd_alu): Likewise.
26465 (cortex_a53_advsimd_alu_q): Likewise.
26466 (cortex_a53_advsimd_mul): Likewise.
26467 (cortex_a53_advsimd_mul_q): Likewise.
26468 (fpmac bypass): Add new bypass for fpmac-fpmac case.
26469 Add missing fmul, r2f_cvt and fconst cases.
26471 2017-06-14 Richard Biener <rguenther@suse.de>
26473 PR middle-end/81088
26474 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
26476 (fold_binary_loc): When associating do not treat pre-existing
26477 TREE_OVERFLOW on literal constants as a reason to allow
26478 TREE_OVERFLOW on associated literal constants.
26480 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
26482 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
26483 (MASK_FEATURES): New macro.
26484 * config/sparc/sparc.c (sparc_option_override): Remove the special
26485 handling of -mfpu and generalize it to all MASK_FEATURES switches.
26487 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
26489 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
26490 a division of 0 if non-call exceptions are enabled.
26492 2017-06-14 Andrew Pinski <apinski@cavium.com>
26493 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26496 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
26497 Improve vector initialization code gen for only variable case.
26499 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
26501 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
26503 2017-06-14 Richard Biener <rguenther@suse.de>
26505 PR tree-optimization/81083
26506 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
26509 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26511 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
26512 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
26513 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
26514 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
26515 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
26516 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
26518 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26520 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
26521 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
26523 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26525 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
26527 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26529 * config/rs6000/t-rtems: Don't handle SPE.
26531 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26533 * config/rs6000/t-linux: Don't handle SPE.
26535 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26537 * config/rs6000/eabispe.h: Delete file.
26539 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26541 * config/rs6000/t-spe: Delete file.
26543 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26545 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
26546 (rs6000_legitimate_offset_address_p): Return false for anything in
26547 V2SImode or V2SFmode.
26549 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26551 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
26552 except V2SF and V2SI. Rearrange the vector modes, and add comments.
26553 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
26555 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
26556 (rs6000_legitimate_offset_address_p): Ditto.
26557 (rs6000_emit_move): Ditto.
26558 (rs6000_init_builtins): Remove V4HI_type_node.
26560 2017-06-13 Martin Liska <mliska@suse.cz>
26563 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
26564 (gate_asan): Likewise.
26565 * asan.h (asan_no_sanitize_address_p): Remove the function.
26566 (sanitize_flags_p): New function.
26567 * builtins.def: Fix coding style.
26568 * common.opt: Use renamed enum value.
26569 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
26570 * doc/extend.texi: Document no_sanitize attribute.
26571 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
26572 to SANITIZE_UNDEFINED_NONDEFAULT.
26573 * gcc.c (sanitize_spec_function): Use the renamed enum value.
26574 * gimple-fold.c (optimize_atomic_compare_exchange_p):
26575 Use sanitize_flags_p.
26576 * gimplify.c (gimplify_function_tree): Likewise.
26577 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
26578 * opts.c (parse_no_sanitize_attribute): New function.
26579 (common_handle_option): Use renamed enum value.
26580 * opts.h (parse_no_sanitize_attribute): Declare.
26581 * tree.c (sanitize_flags_p): New function.
26582 * tree.h: Declared here.
26583 * tsan.c: Use sanitize_flags_p.
26584 * ubsan.c (ubsan_expand_null_ifn): Likewise.
26585 (instrument_mem_ref): Likewise.
26586 (instrument_bool_enum_load): Likewise.
26587 (do_ubsan_in_current_function): Remove the function.
26588 (pass_ubsan::execute): Use sanitize_flags_p.
26589 * ubsan.h: Remove do_ubsan_in_current_function
26590 * tree-cfg.c (print_no_sanitize_attr_value): New function.
26591 (dump_function_to_file): Use it here.
26593 2017-06-13 Martin Jambor <mjambor@suse.cz>
26595 PR tree-optimization/80803
26596 PR tree-optimization/81063
26597 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
26598 (propagate_subaccesses_across_link): Enqueue subtree whenever
26599 necessary instead of relying on the caller.
26601 2017-06-13 Martin Jambor <mjambor@suse.cz>
26603 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
26604 that have a first_link.
26605 (sort_and_splice_var_accesses): Do not check first_link before
26607 (subtree_mark_written_and_enqueue): Likewise.
26608 (propagate_all_subaccesses): Likewise and do not stop at first
26609 parent with a first_link.
26611 2017-06-13 Martin Jambor <mjambor@suse.cz>
26613 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
26616 2017-06-13 Yury Gribov <tetra2005@gmail.com>
26618 * match.pd: New pattern.
26620 2017-06-13 Yury Gribov <tetra2005@gmail.com>
26622 * tree-vrp.c (is_masked_range_test): New function.
26623 (register_edge_assert_for): Determine ranges for
26626 2017-06-13 Yury Gribov <tetra2005@gmail.com>
26628 PR tree-optimization/67328
26629 * fold-const.c (maskable_range_p): New function.
26630 (build_range_check): Generate bittests if possible.
26632 2017-06-13 Martin Liska <mliska@suse.cz>
26634 * gimple-pretty-print.c (dump_probability): Add new argument.
26635 (dump_edge_probability): Dump both probability and count.
26636 (dump_gimple_label): Likewise.
26637 (dump_gimple_bb_header): Likewise.
26639 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
26642 * config/avr/avr-devices.c: Fix indentation.
26643 * config/avr/gen-avr-mmcu-specs.c: Dito.
26645 2017-06-13 Richard Biener <rguenther@suse.de>
26647 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
26648 instead get vector type from stmt_info.
26649 (vectorizable_reduction): Adjust. Remove dead code.
26651 2017-06-13 Richard Biener <rguenther@suse.de>
26653 PR middle-end/81065
26654 * fold-const.c (extract_muldiv_1): Remove bogus distribution
26655 case of C * (x * C2 + C3).
26656 (fold_addr_of_array_ref_difference): Properly fold index difference.
26658 2017-06-12 David S. Miller <davem@davemloft.net>
26661 * config/sparc/sparc.md (return expander): Emit frame blockage if
26662 function uses alloca.
26664 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
26666 * combine.c (make_field_assignment): Check len rather than the mode
26667 precision when calling force_to_mode.
26669 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
26671 Support multilibs and devices that see flash in RAM address range.
26674 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
26675 (avr_mcu_t) <flash_pm_offset>: New field.
26676 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
26677 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
26678 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
26679 (AVR_TINY_PM_OFFSET): Remove macro.
26680 * config/avr/avr.opt (-mshort-calls): New option.
26681 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
26682 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
26683 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
26684 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
26685 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
26686 instead of avr_arch->have_jmp_call.
26687 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
26688 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
26689 avr_arch->flash_pm_offset to define.
26690 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
26691 new field flash_pm_offset. Add entry for avrxmega3.
26692 (avr_texinfo): Add entry for avrxmega3.
26693 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
26694 attiny212, attiny214,
26695 attiny412, attiny414, attiny416, attiny417,
26696 attiny814, attiny816, attiny817,
26697 attiny1614, attiny1616, attiny1617,
26698 attiny3214, attiny3216, attiny3217.
26699 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
26700 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
26701 (avr_print_operand_address) [AVR_TINY]: Same.
26702 (avr_asm_init_sections) <readonly_data_section>: Only patch
26703 callback if avr_arch->flash_pm_offset = 0.
26704 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
26705 for rodata if avr_arch->flash_pm_offset != 0.
26706 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
26707 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
26708 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
26709 (m_options): Append opt_rcall.
26710 (m_dirnames): Append dir_rcall.
26711 * config/avr/t-multilib: Regenerate.
26713 * configure.ac [target=avr]: Check whether avrxmega3 default
26714 linker description file works as needed.
26715 * configure: Regenerate.
26716 * doc/avr-mmcu.texi: Regenerate.
26717 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
26718 <__AVR_ARCH__>: Document avrxmega3 and 103.
26719 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
26720 <__AVR_SHORT_CALLS__>: Document it.
26721 <__AVR_PM_BASE_ADDRESS__>: Document it.
26722 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
26723 (AVR Variable Attributes) <progmem>: Document this is
26724 not needed for avrxmega3.
26725 (AVR Named Address Spaces) <__flash>: Dito.
26727 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
26729 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
26731 2017-06-12 Doug Rupp <rupp@adacore.com>
26733 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
26734 Append vxworks-stdint.h to the tm_file list.
26735 * config/vxworks-stdint.h: New file.
26737 2017-06-12 Martin Liska <mliska@suse.cz>
26739 PR tree-optimization/81041
26740 * tree-profile.c (gimple_gen_ic_func_profiler):
26741 Create an extra BB in profile-generate
26742 (gimple_gen_time_profiler): Likewise.
26744 2017-06-12 Jakub Jelinek <jakub@redhat.com>
26746 PR tree-optimization/81003
26747 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
26748 (update_range_test): Use it instead of force_gimple_operand_gsi.
26750 2017-06-12 Richard Biener <rguenther@suse.de>
26752 PR tree-optimization/81053
26753 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
26754 with backedge value not defined in loop. Simplify def stmt
26757 2017-06-11 Tom de Vries <tom@codesourcery.com>
26760 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
26762 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
26763 nvptx_cannot_force_const_mem.
26765 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
26767 * opts.c (finish_options): Move test for flag_split_stack after
26768 it has been initialized.
26770 2017-06-11 Jason Merrill <jason@redhat.com>
26772 * tree.h (id_equal): New.
26773 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
26774 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
26775 instead of strcmp of IDENTIFIER_POINTER.
26777 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
26779 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
26780 (mark_all_inlined_calls_cdtor): Fix formating.
26781 (inline_transform): Rescale profile before inlining.
26783 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
26785 * cgraph.h (cgraph_edge::clone): Update prototype.
26786 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
26787 (cgraph_node::create_clone): Update.
26788 (cgraph_node::create_version_clone): Update.
26789 * tree-inline.c (copy_bb): Update.
26790 (expand_call_inline): Update.
26792 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
26794 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
26795 factored out from ...
26796 (rs6000_emit_prologue): ... here.
26798 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
26800 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
26801 factored out from ...
26802 (rs6000_emit_prologue): ... here.
26804 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
26806 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
26808 (handle_missing_profiles): Fix computation of tp_first_run.
26809 (counts_to_freqs): Do not touch freqs when count is 0.
26811 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
26813 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
26816 2017-06-10 Tom de Vries <tom@codesourcery.com>
26818 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
26819 attributes): Document signal effective target.
26821 2017-06-10 Tom de Vries <tom@codesourcery.com>
26823 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
26824 Document effective target stack_size.
26826 2017-06-09 David Malcolm <dmalcolm@redhat.com>
26828 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
26829 to the edit_context if they can be auto-applied.
26831 2017-06-9 Ian Lance Taylor <iant@golang.org>
26833 * opts.c (finish_options): If -fsplit-stack, disable implicit
26834 -forder-blocks-and-partition.
26835 * doc/invoke.texi (Optimize Options): Document that when using
26836 -fsplit-stack -forder-blocks-and-partition is not implicitly
26839 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
26841 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
26842 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
26843 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
26844 * builtins.def (abort, trap, unreachable): Declare cold.
26845 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
26846 * tree-core.h (ECF_COLD): New.
26847 * tree.c (set_call_expr_flags): Handle ECF_COLD.
26848 (build_common_builtin_nodes): Mark unreachable and abort as cold.
26850 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
26852 * predict.c (unlikely_executed_stmt_p): Cleanup.
26854 2017-06-09 Richard Biener <rguenther@suse.de>
26856 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
26857 model if the ref is always written to.
26859 2017-06-09 Tamar Christina <tamar.christina@arm.com>
26861 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
26863 2017-06-09 Tamar Christina <tamar.christina@arm.com>
26865 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
26868 2017-06-09 Tom de Vries <tom@codesourcery.com>
26871 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
26872 "target cannot support label values" when encountering LABEL_REF.
26874 2017-06-09 Martin Liska <mliska@suse.cz>
26876 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
26877 (gimple_gen_ic_func_profiler): Emit direct comparison
26878 of __gcov_indirect_call_callee with NULL.
26879 (gimple_gen_time_profiler): Change probability from
26880 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
26882 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
26884 * profile.c (edge_gcov_counts): Turn to pointer.
26885 (compute_branch_probabilities, compute_branch_probabilities): Update.
26886 (branch_prob): Do not clear edge_gcov_count.
26887 * profile.h (edge_gcov_counts): Turn to pointer.
26888 (edge_gcov_count): Update.
26890 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
26892 * gimple.h (gimple_check_failed): Mark cold.
26894 2017-06-09 Richard Biener <rguenther@suse.de>
26896 PR tree-optimization/66623
26897 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
26898 refactor check_reduction into two parts, properly computing
26899 whether we have to check reduction validity for outer loop
26902 2017-06-09 Richard Biener <rguenther@suse.de>
26904 PR tree-optimization/79483
26905 * graphite-scop-detection.c (order): New global.
26906 (get_order): Compute bb to order mapping that satisfies code
26907 generation constraints.
26908 (cmp_pbbs): New helper.
26909 (build_scops): Start domwalk at entry block, sort generated
26912 2017-06-09 Richard Biener <rguenther@suse.de>
26914 PR middle-end/81007
26915 * ipa-polymorphic-call.c
26916 (ipa_polymorphic_call_context::restrict_to_inner_class):
26917 Skip FIELD_DECLs with error_mark_node type.
26918 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
26921 2017-06-09 Martin Liska <mliska@suse.cz>
26923 * predict.c (struct branch_predictor): New struct.
26924 (test_prediction_value_range): New test.
26925 (predict_c_tests): New function.
26926 * selftest-run-tests.c (selftest::run_tests): Run the function.
26927 * selftest.h: Declare new tests.
26929 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
26932 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
26933 gen_add3_insn did not fail.
26934 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
26935 r0, construct that number in a temporary reg and add that reg to r0.
26936 If asked to put the result in r0 as well, fail.
26938 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
26940 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
26941 for early expansion of vec_eqv.
26943 2017-06-08 Jakub Jelinek <jakub@redhat.com>
26945 PR middle-end/81005
26946 * ubsan.c (instrument_null): Avoid pointless code temporary.
26947 (pass_ubsan::execute): Instrument aggregate arguments of calls.
26949 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
26953 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
26956 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
26959 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
26961 * predict.c (unlikely_executed_edge_p): Move ahead.
26962 (probably_never_executed_edge_p): Use it.
26964 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
26966 PR middle-end/79988
26967 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
26968 gimple_call_builtin_p call.
26970 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
26972 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
26973 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
26974 rtl_check_failed_type2, rtl_check_failed_code1,
26975 rtl_check_failed_code2, rtl_check_failed_code_mode,
26976 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
26977 rtvec_check_failed_bounds, rtl_check_failed_flag,
26978 _fatal_insn_not_found, _fatal_insn): Likewise.
26979 * tree.h (tree_contains_struct_check_failed,
26980 tree_check_failed, tree_not_check_failed,
26981 tree_class_check_failed, tree_range_check_failed,
26982 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
26983 tree_vec_elt_check_failed, phi_node_elt_check_failed,
26984 tree_operand_check_failed, omp_clause_check_failed,
26985 omp_clause_operand_check_failed, omp_clause_range_check_failed):
26988 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
26990 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
26991 flag_branch_probabilities.
26992 * ipa-inline.c (edge_badness): Likewise.
26993 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
26994 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
26995 * predict.c (maybe_hot_frequency_p): Likewise.
26996 (probably_never_executed): Likewise.
26997 * sched-ebb.c (schedule_ebbs): Likewise.
26998 * sched-rgn.c (find_single_block_region): Likewise.
26999 * tracer.c (tail_duplicate): Likewise.
27001 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27003 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
27004 longer requires x_flag_profile_use.
27006 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27008 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
27009 instead of flag_reorder_blocks_and_partition.
27010 * dbxout.c (dbxout_function_end): Likewise.
27011 * dwarf2out.c (gen_subprogram_die): Likewise.
27012 * haifa-sched.c (sched_create_recovery_edges): Likewise.
27013 * hw-doloop.c (reorg_loops): Likewise.
27014 * varasm.c (assemble_start_function,
27015 assemble_end_function): Likewise.
27016 (decide_function_section): Do not check for
27017 flag_reorder_blocks_and_partition.
27019 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
27021 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
27023 (chkp_get_hard_register_fake_addr_expr): Ditto.
27024 (chkp_build_addr_expr): Add check for hard reg case.
27025 (chkp_parse_array_and_component_ref): Ditto.
27026 (chkp_find_bounds_1): Ditto.
27027 (chkp_process_stmt): Don't generate bounds store for
27030 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27032 * predict.c (maybe_hot_bb_p): Do not check profile status.
27033 (maybe_hot_edge_p): Likewise.
27034 (probably_never_executed): Check for zero counts even if profile
27036 (unlikely_executed_edge_p): New function.
27037 (unlikely_executed_stmt_p): New function.
27038 (unlikely_executed_bb_p): New function.
27039 (set_even_probabilities): Use unlikely predicates.
27040 (combine_predictions_for_bb): Likewise.
27041 (predict_paths_for_bb): Likewise.
27042 (predict_paths_leading_to_edge): Likewise.
27043 (determine_unlikely_bbs): New function.
27044 (estimate_bb_frequencies): Use it.
27045 (compute_function_frequency): Use zero counts even if profile is
27047 * profile-count.h: Fix typo.
27049 2017-08-08 Julia Koval <julia.koval@intel.com>
27051 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
27052 _mm512_mask_cvtsepi16_storeu_epi8,
27053 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
27054 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
27055 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
27056 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
27057 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
27058 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
27059 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
27060 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
27061 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
27062 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
27063 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
27064 __builtin_ia32_pmovuswb256mem_mask,
27065 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
27066 __builtin_ia32_pmovwb512mem_mask): New builtins.
27068 2017-08-08 Julia Koval <julia.koval@intel.com>
27070 PR target/73350,80862
27071 * config/i386/subst.md (round): Fix round pattern.
27072 * config/i386/i386.c (ix86_erase_embedded_rounding):
27073 Fix erasing rounding for the fixed pattern.
27075 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27077 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
27079 2017-06-08 Martin Liska <mliska@suse.cz>
27081 PR gcov-profile/80911
27082 * gcov.c (block_info::block_info): New constructor.
27084 2017-06-07 Carl Love <cel@us.ibm.com>
27086 * config/rs6000/rs6000-c: The return type of the following
27087 built-in functions was implemented as int not long long. Fix sign
27088 of return value for the unsigned version of vec_mulo and vec_mule.
27089 vector unsigned long long vec_bperm (vector unsigned long long,
27090 vector unsigned char)
27091 vector signed long long vec_mule (vector signed int,
27093 vector unsigned long long vec_mule (vector unsigned int,
27094 vector unsigned int)
27095 vector signed long long vec_mulo (vector signed int,
27097 vector unsigned long long vec_mulo (vector unsigned int,
27098 vector unsigned int)
27099 * doc/extend.texi: Fix the documentation for the built-in
27102 2017-06-07 Carl Love <cel@us.ibm.com>
27105 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
27108 2017-06-07 Carl Love <cel@us.ibm.com>
27110 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
27111 support, Generate doublehv for signed int/float for BE case only.
27113 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
27115 * doc/invoke.texi (mcx16): Rewrite.
27117 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27119 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
27120 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
27121 *mov<mode>_softfloat, and an anonymous splitter): Use
27122 nonimmediate_operand instead of rs6000_nonimmediate_operand.
27124 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27126 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
27128 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
27129 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
27130 (rs6000_debug_reg_global): Adjust.
27131 (rs6000_init_hard_regno_mode_ok): Adjust.
27132 (rs6000_dbx_register_number): Adjust.
27133 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
27134 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
27135 Remove SPE_ACC and SPEFSCR.
27136 (REG_ALLOC_ORDER): Ditto.
27137 (FRAME_POINTER_REGNUM): Change to 111.
27138 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
27139 (REG_CLASS_NAMES): Ditto.
27140 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
27141 (REGISTER_NAMES): Ditto.
27142 (ADDITIONAL_REG_NAMES): Ditto.
27143 (rs6000_reg_names): Ditto.
27144 * config/rs6000/rs6000.md: Renumber some register number
27147 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27149 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
27151 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
27152 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
27154 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
27155 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
27156 Delete the SPE high registers.
27157 (REG_ALLOC_ORDER): Ditto.
27158 (enum reg_class): Remove SPE_HIGH_REGS.
27159 (REG_CLASS_NAMES): Ditto.
27160 (REG_CLASS_CONTENTS): Delete the SPE high registers.
27161 (REGISTER_NAMES): Ditto.
27162 (rs6000_reg_names): Ditto.
27163 * doc/tm.texi.in: Remove SPE as example.
27164 * doc/tm.texi: Regenerate.
27166 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27168 * config/rs6000/8540.md (ppc8540_brinc): Delete.
27169 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
27170 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
27171 * config/rs6000/rs6000.md (type): Remove "brinc".
27173 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27175 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
27176 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
27177 * config/rs6000/linuxspe.h: Delete file.
27178 * config/rs6000/rs6000.md: Don't include spe.md.
27179 * config/rs6000/spe.h: Delete file.
27180 * config/rs6000/spe.md: Delete file.
27181 * config/rs6000/t-rs6000: Remove spe.md.
27183 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27185 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
27186 (reg_or_none500mem_operand): Delete.
27187 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
27188 instead of reg_or_none500mem_operand.
27190 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27192 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
27193 handling of SPE flags.
27194 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
27196 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27198 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
27200 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
27201 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
27202 paired_divv2sf3): Similar.
27203 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
27204 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
27205 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
27207 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
27208 Rename the paired_* instruction patterns.
27209 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
27211 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
27212 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
27213 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
27214 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
27215 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
27216 PAIRED_VECTOR_MODE.
27217 (struct machine_function): Delete field spe_insn_chain_scanned_p.
27218 (spe_func_has_64bit_regs_p): Delete.
27219 (spe_expand_predicate_builtin): Delete.
27220 (spe_expand_evsel_builtin): Delete.
27221 (TARGET_DWARF_REGISTER_SPAN): Do not define.
27222 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
27223 (invalid_e500_subreg): Delete.
27224 (rs6000_legitimize_address): Always force_reg op2 as well, for
27225 paired single memory accesses.
27226 (rs6000_member_type_forces_blk): Delete.
27227 (rs6000_spe_function_arg): Delete.
27228 (rs6000_expand_unop_builtin): Delete SPE handling.
27229 (rs6000_expand_binop_builtin): Ditto.
27230 (spe_expand_stv_builtin): Delete.
27231 (bdesc_2arg_spe): Delete.
27232 (spe_expand_builtin): Delete.
27233 (spe_expand_predicate_builtin): Delete.
27234 (spe_expand_evsel_builtin): Delete.
27235 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
27236 (spe_init_builtins): Delete.
27237 (spe_func_has_64bit_regs_p): Delete.
27238 (savres_routine_name): Delete "info" parameter. Adjust callers.
27239 (rs6000_emit_stack_reset): Ditto.
27240 (rs6000_dwarf_register_span): Delete.
27241 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
27242 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
27243 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
27245 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
27247 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
27248 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
27249 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
27250 mulv2sf3, divv2sf3): Delete expanders.
27252 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27254 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
27256 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27258 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
27259 * config/rs6000/rs6000.c: Ditto.
27261 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27263 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
27264 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
27265 comparison_operator.
27267 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27269 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
27270 * config/rs6000/rs6000.opt: Ditto.
27271 * config/rs6000/t-rtems: Ditto.
27273 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27275 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
27276 TARGET_E500_SINGLE by 0, simplify.
27277 * config/rs6000/rs6000.c: Ditto.
27278 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
27279 (spe_build_register_parallel): Delete.
27280 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
27281 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
27282 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
27283 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
27284 (E500_CONVERT): Delete.
27285 * config/rs6000/spe.md: Remove many patterns and all define_constants.
27287 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27289 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
27290 * config/rs6000/dfp.md: Ditto.
27291 (negdd2, *negdd2_fpr): Merge.
27292 (absdd2, *absdd2_fpr): Merge.
27293 (negtd2, *negtd2_fpr): Merge.
27294 (abstd2, *abstd2_fpr): Merge.
27295 * config/rs6000/e500.h: Delete file.
27296 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
27297 TARGET_FPRS by 1 and simplify.
27298 * config/rs6000/rs6000-c.c: Ditto.
27299 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
27300 TARGET_DF_SPE by 0.
27301 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
27303 * config/rs6000/rs6000.md: Ditto.
27304 (floatdidf2, *floatdidf2_fpr): Merge.
27305 (move_from_CR_gt_bit): Delete.
27306 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
27307 (E500_CR_IOR_COMPARE): Delete.
27308 (All patterns that require !TARGET_FPRS): Delete.
27309 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
27311 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27313 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
27315 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27317 * graphds.c (add_edge): Intitialize edge's attached data.
27318 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
27319 pointer parameter. Call pointed function on each edge during
27320 graph traversing. Skip traversing the edge when the function
27322 (graphds_dfs, graphds_scc): Ditto.
27323 (for_each_edge): New parameter. Pass the new parameter to callback
27325 * graphds.h (skip_edge_callback): New function pointer type.
27326 (graphds_dfs, graphds_scc): New function pointer parameter.
27327 (graphds_edge_callback, for_each_edge): New parameter.
27329 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27331 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
27332 out code checking if runtime alias check is possible to below ...
27333 Call the new function.
27334 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
27335 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
27337 2017-06-07 Marek Polacek <polacek@redhat.com>
27340 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
27341 TYPE_OVERFLOW_WRAPS check.
27343 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27345 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
27346 if versioning is required.
27347 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
27348 peeling with the check for versioning.
27350 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27352 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
27353 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
27354 Set true to new parameter if new ssa variable is defined.
27355 (vect_gen_vector_loop_niters): Refactor. Set range information
27356 for the new vector loop bound variable.
27357 (vect_do_peeling): Ditto.
27359 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27361 * tree-affine.c (ssa.h): Include header file.
27362 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
27363 has wrapping overflow behavior.
27365 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27367 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
27369 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27371 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
27372 (tree_to_aff_combination): ... here.
27374 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27376 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
27377 reg_pressure model function.
27378 (ivopts_global_cost_for_size): Delete.
27379 (determine_set_costs, iv_ca_recount_cost): Call new model function
27380 ivopts_estimate_reg_pressure.
27382 2017-06-07 Tamar Christina <tamar.christina@arm.com>
27384 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
27385 expensive than udiv. Remove floating point cases from mod.
27387 2017-06-07 Tamar Christina <tamar.christina@arm.com>
27389 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
27390 Increase idiv cost.
27392 2017-06-07 Tamar Christina <tamar.christina@arm.com>
27394 * config/aarch64/aarch64.md
27395 (copysignsf3): Fix mask generation.
27397 2017-06-07 Jakub Jelinek <jakub@redhat.com>
27399 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
27401 (class dump_manager): Add register_dumps method.
27402 * dumpfile.c: Include langhooks.h.
27403 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
27404 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
27405 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
27406 (dump_manager::dump_register): Start with 512 entries instead of 32.
27407 (dump_manager::register_dumps): New method.
27408 * toplev.c (general_init): Instead of invoking register_dumps
27409 langhook, invoke register_dumps method on the dump manager.
27410 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
27413 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
27415 * doc/md.texi: Clarify the restrictions on a define_insn condition.
27416 Say that # requires an associated define_split to exist, and that
27417 the define_split must be suitable for use after register allocation.
27419 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27421 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
27422 (compute_outgoing_frequencies): Also initialize zero counts.
27423 (find_many_sub_basic_blocks): Do not produce uninitialized profile
27424 around loops; preserve more of profile when nothing changes.
27426 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
27428 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
27430 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
27431 * config/arm/arm-cpu-cdata.h: Regenerate.
27432 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
27433 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
27434 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
27435 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
27436 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
27438 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
27440 * config/arm/t-rmprofile: Likewise.
27441 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
27443 2017-06-06 David S. Miller <davem@davemloft.net>
27446 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
27447 blockage if function uses alloca.
27449 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27451 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
27452 New "uid" fields to hold pretty-print IDs of group and ref.
27453 Memory references are now identified as <group_id>:<ref_id>
27454 instead of using [random] addresses.
27455 (dump_mem_details): Simplify, no functional change.
27456 (dump_mem_ref): Simplify and make output more concise.
27457 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
27458 (find_or_create_group): Initialize group uid.
27459 (record_ref): Initialize ref uid. Improve debug output.
27460 (prune_group_by_reuse, should_issue_prefetch_p,)
27461 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
27462 (mark_nontemporal_store, determine_loop_nest_reuse):
27463 Improve debug output.
27465 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27467 * dbgcnt.def (prefetch): New debug counter.
27468 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
27469 (schedule_prefetches): Stop issueing prefetches if debug counter
27472 2017-06-06 Tom de Vries <tom@codesourcery.com>
27474 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
27475 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
27477 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27479 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
27480 Use aarch64_reg_or_zero predicate for operand 4.
27481 (aarch64_compare_and_swap<mode> define_insn_and_split):
27482 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
27483 (aarch64_store_exclusive<mode>): Likewise for operand 2.
27485 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
27487 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
27488 (arm_compute_save_core_reg_mask): This.
27489 (thumb1_compute_save_reg_mask): Rename into ...
27490 (thumb1_compute_save_core_reg_mask): This.
27491 (arm_compute_save_reg0_reg12_mask): Adapt comment.
27492 (arm_compute_frame_layout): Likewise.
27494 2017-06-06 Richard Biener <rguenther@suse.de>
27496 PR tree-optimization/80974
27497 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
27498 keep or clear leaders SSA info.
27500 2017-06-06 Tom de Vries <tom@codesourcery.com>
27502 * config/nvptx/nvptx.c (split_mode_p): New function.
27503 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
27505 2017-06-06 Tom de Vries <tom@codesourcery.com>
27507 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
27509 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27512 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
27515 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27517 * shrink-wrap.c (handle_simple_exit): Update profile.
27518 (try_shrink_wrapping): Upate profile.
27520 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27522 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
27523 (tree_guess_outgoing_edge_probabilities): New.
27524 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
27525 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
27527 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27529 * ipa-split.c (split_function): Initialize return bb profile.
27531 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27533 * profile.c (compute_branch_probabilities): Also initialize
27534 EXIT_BLOCK profile.
27536 2017-06-06 Richard Biener <rguenther@suse.de>
27538 PR tree-optimization/80928
27539 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
27540 (vect_analyze_loop_operations): Properly guard analysis for
27542 (vect_transform_loop): Likewise.
27543 (vect_analyze_loop_2): Also reset SLP type on PHIs.
27544 (vect_model_induction_cost): Do not cost for pure SLP.
27545 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
27546 of induction in inner loop vectorization.
27547 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
27548 (vect_get_and_check_slp_defs): Handle vect_induction_def.
27549 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
27551 (vect_analyze_slp_cost_1): Cost induction.
27552 (vect_detect_hybrid_slp_stmts): Handle PHIs.
27553 (vect_get_slp_vect_defs): Likewise.
27554 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
27555 (vect_transform_stmt): Handle SLP reductions.
27556 * tree-vectorizer.h (vectorizable_induction): Adjust.
27558 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
27560 * config/rs6000/rs6000.c (make_resolver_func): Update
27561 init_lowered_empty_function call.
27563 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
27565 * doc/invoke.texi: Document the -fprofile-abs-path option.
27566 * common.opt (fprofile-abs-path): New option.
27567 * gcov-io.h (gcov_write_filename): Declare.
27568 * gcov-io.c (gcov_write_filename): New function.
27569 * coverage.c (coverage_begin_function): Use gcov_write_filename.
27570 * profile.c (output_location): Likewise.
27572 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
27574 * shring-wrap.c: Revert accidental commit.
27576 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
27578 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
27580 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
27582 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
27584 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
27585 profile in callgraph edge.
27586 * profile-count.h (apply_probability): If THIS is 0, then result is 0
27587 (apply_scale): Likewise.
27588 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
27589 Also scale profile when inlining function with zero profile.
27590 (initialize_cfun): Update exit block profile even when it is zero.
27591 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
27592 when profile is read.
27594 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
27596 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
27597 (CLONE_*): New constants to define the processors we can generate
27598 code for with the target_clone attribute.
27599 (rs6000_clone_map): New array to identify which clone processors
27600 the current program is running on.
27601 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
27602 target_clone attribute.
27603 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
27604 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
27605 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
27606 (cpu_expand_builtin): Add support for target_clone attribute.
27607 (rs6000_valid_attribute_p): Allow "default" attribute.
27608 (get_decl_name): New debug function to simplify printing the
27609 current function name in debugging statements.
27610 (rs6000_clone_priority): New functions to support the target_clone
27611 attribute, and be able to generate code to switch between ISA 2.05
27612 through ISA 3.0 (power6 through power9).
27613 (rs6000_compare_version_priority): Likewise.
27614 (rs6000_get_function_versions_dispatcher): Likewise.
27615 (make_resolver_func): Likewise.
27616 (add_condition_to_bb): Likewise.
27617 (dispatch_function_versions): Likewise.
27618 (rs6000_generate_version_dispatcher_body): Likewise.
27619 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
27620 (fusion_gpr_load_p): Fix a spacing issue.
27621 * doc/extend.texi (Common Function Attributes): Document that the
27622 PowerPC supports the target_clone attribute.
27624 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
27626 * config/arm/arm.h: explain F symbol found in description of ARM
27627 register allocation in its legend.
27629 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
27631 * config/mips/frame-header-opt.c: Include profile-count.h.
27632 * config/riscv/riscv.c: Include profile-count.h
27634 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
27636 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
27638 (sm_set_flag_if_changed): Add bbs field.
27639 (execute_sm_if_changed_flag_set): Pass BBS.
27640 (execute_sm): Update.
27642 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27644 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
27647 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27649 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
27650 (peephole2): New peephole2 to emit the above.
27651 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
27653 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27655 * config/aarch64/aarch64.c (define_peephole2 above
27656 *sub_<shift>_<mode>): New peephole.
27658 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
27660 * config/i386/i386.c (make_resolver_func): Update.
27661 * Makefile.in: Add profile-count.h and profile-count.o
27662 * auto-profile.c (afdo_indirect_call): Update to new API.
27663 (afdo_set_bb_count): Update.
27664 (afdo_propagate_edge): Update.
27665 (afdo_propagate_circuit): Update.
27666 (afdo_calculate_branch_prob): Update.
27667 (afdo_annotate_cfg): Update.
27668 * basic-block.h: Include profile-count.h
27669 (struct edge_def): Turn count to profile_count.
27670 (struct basic_block_def): Likewie.
27671 (REG_BR_PROB_BASE): Move to profile-count.h
27672 (RDIV): Move to profile-count.h
27673 * bb-reorder.c (max_entry_count): Turn to profile_count.
27674 (find_traces): Update.
27675 (rotate_loop):Update.
27676 (connect_traces):Update.
27677 (sanitize_hot_paths):Update.
27678 * bt-load.c (migrate_btr_defs): Update.
27679 * cfg.c (RDIV): Remove.
27680 (init_flow): Use alloc_block.
27681 (alloc_block): Uninitialize count.
27682 (unchecked_make_edge): Uninitialize count.
27683 (check_bb_profile): Update.
27684 (dump_edge_info): Update.
27685 (dump_bb_info): Update.
27686 (update_bb_profile_for_threading): Update.
27687 (scale_bbs_frequencies_int): Update.
27688 (scale_bbs_frequencies_gcov_type): Update.
27689 (scale_bbs_frequencies_profile_count): New.
27690 * cfg.h (update_bb_profile_for_threading): Update.
27691 (scale_bbs_frequencies_profile_count): Declare.
27692 * cfgbuild.c (compute_outgoing_frequencies): Update.
27693 (find_many_sub_basic_blocks): Update.
27694 * cfgcleanup.c (try_forward_edges): Update.
27695 (try_crossjump_to_edge): Update.
27696 * cfgexpand.c (expand_gimple_tailcall): Update.
27697 (construct_exit_block): Update.
27698 * cfghooks.c (verify_flow_info): Update.
27699 (dump_bb_for_graph): Update.
27700 (split_edge): Update.
27701 (make_forwarder_block): Update.
27702 (duplicate_block): Update.
27703 (account_profile_record): Update.
27704 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
27705 (get_estimated_loop_iterations): Update.
27706 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
27707 (single_likely_exit): Update.
27708 * cfgloopmanip.c (scale_loop_profile): Update.
27710 (set_zero_probability): Update.
27711 (lv_adjust_loop_entry_edge): Update.
27712 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
27713 (purge_dead_edges): Update.
27714 (rtl_account_profile_record): Update.
27715 * cgraph.c (cgraph_node::create): Uninitialize count.
27716 (symbol_table::create_edge): Uninitialize count.
27717 (cgraph_update_edges_for_call_stmt_node): Update.
27718 (cgraph_edge::dump_edge_flags): Update.
27719 (cgraph_node::dump): Update.
27720 (cgraph_edge::maybe_hot_p): Update.
27721 * cgraph.h: Include profile-count.h
27722 (create_clone), create_edge, create_indirect_edge): Update.
27723 (cgraph_node): Turn count to profile_count.
27724 (cgraph_edge0: Likewise.
27725 (make_speculative, clone): Update.
27726 (create_edge): Update.
27727 (init_lowered_empty_function): Update.
27728 * cgraphclones.c (cgraph_edge::clone): Update.
27729 (duplicate_thunk_for_node): Update.
27730 (cgraph_node::create_clone): Update.
27731 * cgraphunit.c (cgraph_node::analyze): Update.
27732 (cgraph_node::expand_thunk): Update.
27733 * final.c (dump_basic_block_info): Update.
27734 * gimple-streamer-in.c (input_bb): Update.
27735 * gimple-streamer-out.c (output_bb): Update.
27736 * graphite.c (print_global_statistics): Update.
27737 (print_graphite_scop_statistics): Update.
27738 * hsa-brig.c: Include basic-block.h.
27739 * hsa-dump.c: Include basic-block.h.
27740 * hsa-gen.c (T sum_slice): Update.
27741 (convert_switch_statements):Update.
27742 * hsa-regalloc.c: Include basic-block.h.
27743 * ipa-chkp.c (chkp_produce_thunks): Update.
27744 * ipa-cp.c (struct caller_statistics): Update.
27745 (init_caller_stats): Update.
27746 (gather_caller_stats): Update.
27747 (ipcp_cloning_candidate_p): Update.
27748 (good_cloning_opportunity_p): Update.
27749 (get_info_about_necessary_edges): Update.
27750 (dump_profile_updates): Update.
27751 (update_profiling_info): Update.
27752 (update_specialized_profile): Update.
27753 (perhaps_add_new_callers): Update.
27754 (decide_about_value): Update.
27755 (ipa_cp_c_finalize): Update.
27756 * ipa-devirt.c (struct odr_type_warn_count): Update.
27757 (struct decl_warn_count): Update.
27758 (struct final_warning_record): Update.
27759 (possible_polymorphic_call_targets): Update.
27760 (ipa_devirt): Update.
27761 * ipa-fnsummary.c (redirect_to_unreachable): Update.
27762 * ipa-icf.c (sem_function::merge): Update.
27763 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
27764 * ipa-inline.c (compute_uninlined_call_time): Update.
27765 (compute_inlined_call_time): Update.
27766 (want_inline_small_function_p): Update.
27767 (want_inline_self_recursive_call_p): Update.
27768 (edge_badness): Update.
27769 (lookup_recursive_calls): Update.
27770 (recursive_inlining): Update.
27771 (inline_small_functions): Update.
27772 (dump_overall_stats): Update.
27773 (dump_inline_stats): Update.
27774 * ipa-profile.c (ipa_profile_generate_summary): Update.
27775 (ipa_propagate_frequency): Update.
27776 (ipa_profile): Update.
27777 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
27778 * ipa-utils.c (ipa_merge_profiles): Update.
27779 * loop-doloop.c (doloop_modify): Update.
27780 * loop-unroll.c (report_unroll): Update.
27781 (unroll_loop_runtime_iterations): Update.
27782 * lto-cgraph.c (lto_output_edge): Update.
27783 (lto_output_node): Update.
27784 (input_node): Update.
27785 (input_edge): Update.
27786 (merge_profile_summaries): Update.
27787 * lto-streamer-in.c (input_cfg): Update.
27788 * lto-streamer-out.c (output_cfg): Update.
27789 * mcf.c (create_fixup_graph): Update.
27790 (adjust_cfg_counts): Update.
27791 (sum_edge_counts): Update.
27792 * modulo-sched.c (sms_schedule): Update.
27793 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
27794 * predict.c (maybe_hot_count_p): Update.
27795 (probably_never_executed): Update.
27796 (dump_prediction): Update.
27797 (combine_predictions_for_bb): Update.
27798 (propagate_freq): Update.
27799 (handle_missing_profiles): Update.
27800 (counts_to_freqs): Update.
27801 (rebuild_frequencies): Update.
27802 (force_edge_cold): Update.
27803 * predict.h: Include profile-count.h
27804 (maybe_hot_count_p, counts_to_freqs): UPdate.
27805 * print-rtl-function.c: Do not include cfg.h
27806 * print-rtl.c: Include basic-block.h
27807 * profile-count.c: New file.
27808 * profile-count.h: New file.
27809 * profile.c (is_edge_inconsistent): Update.
27810 (correct_negative_edge_counts): Update.
27811 (is_inconsistent): Update.
27812 (set_bb_counts): Update.
27813 (read_profile_edge_counts): Update.
27814 (compute_frequency_overlap): Update.
27815 (compute_branch_probabilities): Update; Initialize and deinitialize
27817 (branch_prob): Update.
27818 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
27819 (edge_gcov_count): New.
27820 (bb_gcov_count): New.
27821 * shrink-wrap.c (try_shrink_wrapping): Update.
27822 * tracer.c (better_p): Update.
27823 * trans-mem.c (expand_transaction): Update.
27824 (ipa_tm_insert_irr_call): Update.
27825 (ipa_tm_insert_gettmclone_call): Update.
27826 * tree-call-cdce.c: Update.
27827 * tree-cfg.c (gimple_duplicate_sese_region): Update.
27828 (gimple_duplicate_sese_tail): Update.
27829 (gimple_account_profile_record): Update.
27830 (execute_fixup_cfg): Update.
27831 * tree-inline.c (copy_bb): Update.
27832 (copy_edges_for_bb): Update.
27833 (initialize_cfun): Update.
27834 (freqs_to_counts): Update.
27835 (copy_cfg_body): Update.
27836 (expand_call_inline): Update.
27837 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
27838 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
27839 (try_unroll_loop_completely): Update.
27840 (try_peel_loop): Update.
27841 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
27842 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
27843 * tree-ssa-loop-split.c (connect_loops): Update.
27844 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
27845 * tree-ssa-reassoc.c (branch_fixup): Update.
27846 * tree-ssa-tail-merge.c (replace_block_by): Update.
27847 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
27848 (compute_path_counts): Update.
27849 (update_profile): Update.
27850 (recompute_probabilities): Update.
27851 (update_joiner_offpath_counts): Update.
27852 (estimated_freqs_path): Update.
27853 (freqs_to_counts_path): Update.
27854 (clear_counts_path): Update.
27855 (ssa_fix_duplicate_block_edges): Update.
27856 (duplicate_thread_path): Update.
27857 * tree-switch-conversion.c (case_bit_test_cmp): Update.
27858 (struct switch_conv_info): Update.
27859 * tree-tailcall.c (decrease_profile): Update.
27860 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
27861 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
27862 * value-prof.c (check_counter): Update.
27863 (gimple_divmod_fixed_value): Update.
27864 (gimple_mod_pow2): Update.
27865 (gimple_mod_subtract): Update.
27866 (gimple_ic_transform): Update.
27867 (gimple_stringop_fixed_value): Update.
27868 * value-prof.h (gimple_ic): Update.
27870 2017-06-02 Carl Love <cel@us.ibm.com>
27872 * config/rs6000/rs6000-c: Add support for built-in functions
27873 vector double vec_doublee (vector signed int);
27874 vector double vec_doublee (vector unsigned int);
27875 vector double vec_doublee (vector float);
27876 vector double vec_doubleh (vector signed int);
27877 vector double vec_doubleh (vector unsigned int);
27878 vector double vec_doubleh (vector float);
27879 vector double vec_doublel (vector signed int);
27880 vector double vec_doublel (vector unsigned int);
27881 vector double vec_doublel (vector float);
27882 vector double vec_doubleo (vector signed int);
27883 vector double vec_doubleo (vector unsigned int);
27884 vector double vec_doubleo (vector float);.
27885 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
27886 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
27888 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
27889 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
27890 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
27892 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
27893 vec_doublel, vec_doubleh.
27894 * doc/extend.texi: Update the built-in documentation file for the
27895 new built-in functions.
27897 2017-06-02 David Malcolm <dmalcolm@redhat.com>
27900 * ipa-inline-analysis.c (free_growth_caches): Set
27901 edge_removal_hook_holder to NULL after removing it.
27903 2017-06-02 Sudakshina Das <sudi.das@arm.com>
27905 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
27906 comparision with zero.
27908 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
27909 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
27910 for early expansion of vec_min and vec_max builtins.
27911 (builtin_function_type): Add min/max unsigned variants to those
27912 identified as having unsigned arguments.
27914 2017-06-02 Olivier Hainque <hainque@adacore.com>
27916 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
27918 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27920 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
27921 Use VALL_F16 iterator rather than VALL.
27923 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27925 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
27926 Emit CBNZ inside loop when doing a strong exchange and comparing
27927 against zero. Generate the CC flags after the loop.
27929 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
27931 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
27932 (dl_section_ref): New.
27933 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
27934 On AIX, append an expression to subtract the size of the
27935 section length to dl_section_ref.
27937 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
27939 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
27940 for early expansion of vector absolute builtins.
27942 2017-06-02 Richard Biener <rguenther@suse.de>
27944 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
27945 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
27947 2017-06-02 Richard Biener <rguenther@suse.de>
27949 PR tree-optimization/80948
27950 * tree-tailcall.c (find_tail_calls): Track stmts to move in
27951 stmt order as well.
27953 2017-06-02 Richard Biener <rguenther@suse.de>
27955 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
27957 * tree-vect-stmts.c (process_use): Do not mark backedge defs
27958 for inductions as relevant.
27960 2017-06-02 Richard Biener <rguenther@suse.de>
27962 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
27963 (vectorizable_induction): ... this. Remove dead code.
27965 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
27967 * builtins. (expand_builtin_alloca): Remove second parameter and
27968 infer its value from the first parameter instead.
27969 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
27971 2017-06-02 Jakub Jelinek <jakub@redhat.com>
27973 PR rtl-optimization/80903
27974 * loop-doloop.c (add_test): Unshare sequence.
27976 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
27978 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
27980 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
27982 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
27984 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
27985 xlogue_layout::get_instance, logue_layout::xlogue_layout,
27986 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
27987 (xlogue_layout::get_stub_rtx): Make static.
27988 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
27989 (xlogue_layout::compute_stub_managed_regs): Rename to...
27990 (xlogue_layout::count_stub_managed_regs): ...this.
27991 (xlogue_layout::is_stub_managed_reg): New function.
27992 (xlogue_layout::m_stub_names): Rename to...
27993 (xlogue_layout::s_stub_names): ...this, make static.
27994 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
27995 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
27996 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
27997 xlogue_layout::s_stub_names): Instantiate statics.
27998 (stub_managed_regs): Remove.
27999 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
28000 (disable_call_ms2sysv_xlogues): Rename to...
28001 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
28002 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
28004 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
28005 change after reload_completed.
28006 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
28008 (ix86_expand_prologue): Likewise.
28009 (ix86_expand_epilogue): Likewise.
28010 (ix86_expand_split_stack_prologue): Likewise.
28011 (ix86_compute_frame_layout): Remove frame parameter ...
28012 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
28013 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
28015 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
28016 (ix86_frame): Move from here ...
28017 * config/i386/i386.h (ix86_frame): ... to here.
28018 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
28019 complete ix86_frame data structure instead. Remove some_ld_name.
28021 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
28023 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
28024 symbols that hold a DECL_VALUE_EXPR.
28026 2017-06-01 Martin Jambor <mjambor@suse.cz>
28028 PR tree-optimization/80898
28029 * tree-sra.c (process_subtree_disqualification): Removed.
28030 (disqualify_candidate): Do not acll
28031 process_subtree_disqualification.
28032 (subtree_mark_written_and_enqueue): New function.
28033 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
28034 RHS has been disqualified and re-queue LHS if necessary. Apart
28035 from that, ignore disqualified RHS.
28037 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28039 * config/s390/s390.c (s390_emit_epilogue): Disable early return
28040 address fetch for z10 or later.
28042 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28044 * config/arc/arc.md (tst_movb): Add guard when splitting.
28046 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28048 * config/arc/arc.c (arc_can_eliminate): Test against
28049 arc_frame_pointer_needed.
28051 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28053 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
28054 to prevent store reordering.
28055 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
28056 (type): Add block type.
28057 (stack_tie): Define special instruction to be used in
28060 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28062 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
28063 constraint. It is not valid for the pattern.
28064 (noncommutative_binary_comparison): Likewise.
28066 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28068 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
28071 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28073 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
28074 be used by the reg-alloc.
28076 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28078 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
28079 reg-alloc when having mul64 or mul32x16 instructions.
28080 (mulsidi3): Likewise.
28081 (umulsidi3): Likewise.
28082 (mulsi32x16): New pattern.
28083 (mulsi64): Likewise.
28084 (mulsidi64): Likewise.
28085 (umulsidi64): Likewise.
28086 (MUL32x16_REG): Define.
28087 (mul64_600): Use MUL32x16_REG.
28088 (mac64_600): Likewise.
28089 (umul64_600): Likewise.
28090 (umac64_600): Likewise.
28092 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28094 * config/arc/arc.md (mulsi3_700): Make it commutative.
28096 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
28098 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
28100 (*sign_extendsidi2_insn): Likewise for movstosw.
28102 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
28104 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
28105 the type of the input discriminant value. Convert the
28106 discriminant value of signedness vary.
28108 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
28110 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
28111 Add to -Wall section.
28113 2017-06-01 Richard Biener <rguenther@suse.de>
28115 PR middle-end/66313
28116 * fold-const.c (fold_plusminus_mult_expr): If the factored
28117 factor may be zero use a wrapping type for the inner operation.
28118 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
28119 and handle moved defs.
28120 (process_assignment): Properly guard the unary op case. Return a
28121 tri-state indicating that moving the stmt before the call may allow
28122 to continue. Pass through to_move.
28123 (find_tail_calls): Handle moving unrelated defs before
28126 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
28129 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
28130 splitter result in the canonical way.
28132 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
28134 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
28135 also for 32bit target. Update insn attributes.
28136 (zero-extendsidi2 splitter): Allow all registers for operand 1.
28138 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
28140 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
28141 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
28142 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
28143 (_mm_maskz_min_ss): New intrinsics.
28145 2017-05-31 Martin Liska <mliska@suse.cz>
28147 * tree-vect-loop.c (vect_create_epilog_for_reduction):
28148 Change comment style to one we normally use.
28149 (vectorizable_reduction): Likewise.
28150 (vectorizable_induction): Likewise.
28151 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
28152 (vectorizable_call): Likewise.
28153 (vectorizable_simd_clone_call): Likewise.
28154 (vectorizable_conversion): Likewise.
28155 (vectorizable_assignment): Likewise.
28156 (vectorizable_shift): Likewise.
28157 (vectorizable_operation): Likewise.
28158 (vectorizable_store): Likewise.
28159 (vectorizable_load): Likewise.
28160 * tree-vectorizer.h: Likewise.
28162 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
28164 * passes.c (emergency_dump_function): New.
28165 * tree-pass.h (emergency_dump_function): Declare.
28166 * plugin.c (plugins_internal_error_function): Remove.
28167 * plugin.h (plugins_internal_error_function): Remove declaration.
28168 * toplev.c (internal_error_function): New static function. Use it...
28169 (general_init): ...here.
28171 2017-05-31 Graham Markall <graham.markall@embecosm.com>
28173 * config/arc/arc.c (arc_print_operand): Handle constant operands.
28174 (arc_rtx_costs): Add costs for new patterns.
28175 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
28176 * config/arc/predicates.md: Add _1_2_3_operand predicate.
28178 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
28180 * tree-ssa-strlen.c (get_next_strinfo): New function.
28181 (get_stridx_plus_constant): Use it.
28182 (zero_length_string): Likewise.
28183 (adjust_related_strinfos): Likewise.
28184 (adjust_last_stmt): Likewise.
28186 2017-05-31 Richard Biener <rguenther@suse.de>
28189 * config/i386/i386.c (ix86_expand_builtin): Remove assert
28190 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
28192 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
28194 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
28195 loop_vinfo argument and use of dependence distance vectors.
28196 Check instead whether the two references differ only in their
28197 initial value and assume that they have the same alignment if the
28198 difference is a multiple of the vector alignment.
28199 (vect_analyze_data_refs_alignment): Update call accordingly.
28201 2017-05-31 Martin Liska <mliska@suse.cz>
28204 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
28206 2017-05-31 Bin Cheng <bin.cheng@arm.com>
28208 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
28209 (create_intersect_range_checks): Move from ...
28210 * tree-data-ref.c (create_intersect_range_checks_index)
28211 (create_intersect_range_checks): ... to here.
28212 (create_runtime_alias_checks): New function factored from ...
28213 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
28214 here. Call above function.
28215 * tree-data-ref.h (create_runtime_alias_checks): New function.
28217 2017-05-31 Bin Cheng <bin.cheng@arm.com>
28219 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
28220 segment length for dr_b and compute it in wide_int.
28222 2017-05-31 Richard Biener <rguenther@suse.de>
28224 PR tree-optimization/80906
28225 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
28226 and pass through iv_map.
28227 (copy_bb_and_scalar_dependences): Adjust.
28228 (translate_pending_phi_nodes): Likewise.
28229 (copy_loop_close_phi_args): Handle code-generating IVs instead
28232 2017-05-30 David Malcolm <dmalcolm@redhat.com>
28234 * diagnostic-color.c (color_dict): Add "type-diff".
28235 (parse_gcc_colors): Update comment.
28236 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
28237 -fdiagnostics-show-template-tree and -fno-elide-type.
28238 (GCC_COLORS): Add type-diff to example.
28240 (-fdiagnostics-show-template-tree): New.
28241 (-fno-elide-type): New.
28242 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
28243 the pp_format_decoder callback. Call any m_format_postprocessor's
28245 (pretty_printer::pretty_printer): Initialize
28246 m_format_postprocessor.
28247 (pretty_printer::~pretty_printer): Delete any
28248 m_format_postprocessor.
28249 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
28250 (class format_postprocessor): New class.
28251 (struct pretty_printer::format_decoder): Document the new parameters.
28252 (struct pretty_printer::m_format_postprocessor): New field.
28253 * tree-diagnostic.c (default_tree_printer): Update for new
28254 bool and const char ** params.
28255 * tree-diagnostic.h (default_tree_printer): Likewise.
28257 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
28259 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
28260 (lwa_operand): Delete rs6000_gen_cell_microcode test.
28261 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
28262 rs6000_gen_cell_microcode code.
28263 (rs6000_final_prescan_insn): Delete.
28264 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
28265 "warn-cell-microcode" entries.
28266 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
28267 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
28268 throughout. Change cc_reg_not_micro_cr0_operand to
28269 cc_reg_not_cr0_operand throughout.
28270 (*extendhi<mode>2_noload): Delete.
28271 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
28272 (mwarn-cell-microcode): Delete.
28273 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
28274 -mgen-cell-microcode and -mwarn-cell-microcode.
28276 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
28279 * config/i386/constraints.md (Yd): New constraint.
28281 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
28282 and (?Yd, r) alternatives. Update insn attributes.
28283 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
28284 and (?*Yd, r) alternatives. Update insn attributes.
28285 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
28287 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
28289 * gimplify.c (gimplify_modify_expr): Don't create a
28290 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
28293 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
28295 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
28297 2017-05-30 Richard Biener <rguenther@suse.de>
28299 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
28300 and reduc_def fields.
28301 (STMT_VINFO_REDUC_TYPE): New define.
28302 (STMT_VINFO_REDUC_DEF): Likewise.
28303 (vect_force_simple_reduction): Adjust prototype.
28304 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
28305 (vect_is_simple_reduction): Remove check_reduction argument.
28306 (vect_force_simple_reduction): Adjust and set
28307 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
28308 (vectorizable_reduction): Do not re-do reduction analysis
28309 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
28310 * tree-parloops.c (gather_scalar_reductions): Adjust.
28312 2017-05-30 Richard Biener <rguenther@suse.de>
28314 PR middle-end/80901
28315 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
28318 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
28320 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
28321 Introduce unknown_misalignment parameter and remove vf.
28322 (vect_peeling_hash_get_lowest_cost):
28323 Pass unknown_misalignment parameter.
28324 (vect_enhance_data_refs_alignment):
28325 Fix unsupportable data ref treatment.
28327 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
28329 * tree-vect-data-refs.c (vect_get_data_access_cost):
28330 Workaround for SLP handling.
28331 (vect_enhance_data_refs_alignment):
28332 Compute costs for doing no peeling at all, compare to the best
28333 peeling costs so far and avoid peeling if cheaper.
28335 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
28337 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
28338 Return peeling info and set costs to zero for unlimited cost
28340 (vect_enhance_data_refs_alignment): Also inspect all datarefs
28341 with unknown misalignment. Compute and costs for unknown
28342 misalignment, compare them to the costs for known misalignment
28343 and choose the cheapest for peeling.
28345 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
28347 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
28348 (vect_get_peeling_costs_all_drs): Create function.
28349 (vect_peeling_hash_get_lowest_cost):
28350 Use vect_get_peeling_costs_all_drs.
28351 (vect_peeling_supportable): Create function.
28352 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
28354 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
28356 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
28357 DR_HAS_NEGATIVE_STEP.
28358 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
28359 (vect_enhance_data_refs_alignment): Use.
28360 (vect_duplicate_ssa_name_ptr_info): Use.
28361 * tree-vectorizer.h (dr_misalignment): Use.
28362 (known_alignment_for_access_p): Use.
28364 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
28367 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
28369 (has_section_name): New function.
28371 2017-05-30 Martin Liska <mliska@suse.cz>
28374 * auto-profile.c (get_function_decl_from_block): Fix
28377 2017-05-30 Richard Biener <rguenther@suse.de>
28379 PR middle-end/80876
28380 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
28382 2017-05-30 Martin Liska <mliska@suse.cz>
28384 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
28385 * dumpfile.h (struct dump_file_info): Remove ctors.
28387 2017-05-30 Martin Liska <mliska@suse.cz>
28389 * predict.def: Fix GNU coding style.
28391 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
28393 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
28394 Mark 'to' argument with ATTRIBUTE_UNUSED.
28396 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
28398 * config/xtensa/xtensa.c (xtensa_emit_call): Use
28399 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
28400 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
28403 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
28405 * doc/install.texi (Options specification): Restore entry of
28406 --enable-sjlj-exceptions.
28408 2017-05-27 Michael Eager <eager@eagercon.com>
28411 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
28413 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
28415 * config/microblaze/microblaze.h
28416 (FIXED_REGISTERS): Update in macro.
28417 (CALL_USED_REGISTERS): Update in macro.
28419 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
28421 * doc/install.texi: Add links to macOS binary distributions.
28423 2017-05-27 Jakub Jelinek <jakub@redhat.com>
28427 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
28429 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
28431 2017-05-26 Martin Liska <mliska@suse.cz>
28433 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
28435 2017-05-26 Martin Liska <mliska@suse.cz>
28437 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
28438 always leading ';; '.
28439 (dump_bb_info): Likewise.
28440 (brief_dump_cfg): Likewise.
28441 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
28442 * dumpfile.c: Remove usage of TDF_VERBOSE.
28443 * dumpfile.h (enum dump_kind): Likewise.
28444 (dump_gimple_bb_header): Do not use TDF_COMMENT.
28445 * print-tree.c (debug_verbose): Remove.
28446 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
28447 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
28448 * tree-diagnostic.c (default_tree_printer): Replace
28449 TDF_DIAGNOSTIC with TDF_SLIM.
28451 2017-05-26 Bin Cheng <bin.cheng@arm.com>
28453 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
28454 in parameter loop, rather than loop_vinfo.
28455 (create_intersect_range_checks): Ditto.
28456 (vect_create_cond_for_alias_checks): Update call to above functions.
28458 2017-05-26 Bin Cheng <bin.cheng@arm.com>
28460 PR tree-optimization/80815
28461 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
28462 for merging runtime alias checks. Handle negative DR_STEPs.
28464 2017-05-26 Bin Cheng <bin.cheng@arm.com>
28466 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
28468 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
28469 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
28470 out code pruning runtime alias checks.
28471 * tree-data-ref.c (prune_runtime_alias_test_list): New function
28472 factored out from above.
28473 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
28475 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
28477 (prune_runtime_alias_test_list): New decalaration.
28479 2017-05-26 Bin Cheng <bin.cheng@arm.com>
28481 * tree-vect-data-refs.c (compare_tree): Rename and move ...
28482 * tree-data-ref.c (data_ref_compare_tree): ... to here.
28483 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
28484 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
28485 (operator==, comp_dr_with_seg_len_pair): Ditto.
28486 (vect_prune_runtime_alias_test_list): Ditto.
28488 2017-05-26 Martin Liska <mliska@suse.cz>
28491 * params.def: Bound partial-inlining-entry-probability param.
28493 2017-05-26 Marek Polacek <polacek@redhat.com>
28496 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
28499 2017-05-26 Richard Biener <rguenther@suse.de>
28501 PR tree-optimization/80842
28502 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
28505 2017-05-26 Richard Biener <rguenther@suse.de>
28507 PR tree-optimization/80844
28508 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
28510 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
28512 * doc/md.texi (Machine Constraints): Update x86 family
28513 machine constraints section to match 'config/i386/constraints.md'.
28515 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
28517 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
28519 2017-05-25 Nathan Sidwell <nathan@acm.org>
28521 * doc/invoke.texi (--enable-languages): Update documentation.
28523 2017-05-25 Martin Liska <mliska@suse.cz>
28525 * dumpfile.c: Add TDF_FOLDING.
28526 * dumpfile.h (enum dump_kind): Likewise.
28527 * genmatch.c (dt_simplify::gen_1): Use it.
28529 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
28531 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
28533 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
28535 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
28536 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
28538 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
28540 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
28541 * match.pd (X == C): Rewrite it here.
28542 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
28543 with_certain_nonzero_bits2): New predicates.
28544 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
28546 2017-05-24 Nathan Sidwell <nathan@acm.org>
28548 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
28551 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
28554 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
28556 * config/powerpcspe: New port. Files are copied from the rs6000
28557 port, with "rs6000" in filenames replaced by "powerpcspe".
28559 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
28561 PR rtl-optimization/80754
28562 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
28564 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
28566 * config/sparc/sparc.md (length): Return the correct value for -mflat
28567 sibcalls to match output_sibcall.
28569 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
28573 * config/rs6000/rs6000.c (struct machine_function): Add new field
28575 (rs6000_get_separate_components): Init that field, use it.
28576 (rs6000_components_for_bb): Use the field.
28578 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
28580 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
28582 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
28584 PR middle-end/80823
28585 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
28587 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28590 * config/s390/s390.c (s390_check_qrst_address): Check incoming
28591 address against address_operand predicate.
28592 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
28594 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
28596 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
28597 parameters passed indirectly.
28599 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
28601 * config/i386/i386.md (*movdi_internal): Remove SSE4
28602 alternative 18 (?r, *v). Update insn attributes.
28603 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
28604 Update insn attributes.
28605 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
28606 Update insn attributes.
28607 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
28608 alternative 1 (r, v). Remove isa attribute.
28609 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
28610 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
28611 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
28613 2017-05-23 Tom de Vries <tom@codesourcery.com>
28615 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
28618 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
28620 * cgraphunit.c (symbol_table::process_new_functions): Update.
28621 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
28622 (inline_generate_summary): Rename to ...
28623 (ipa_fn_summary_generate): ... this one.
28624 (inline_read_summary): Rename to ...
28625 (ipa_fn_summary_read): ... this one.
28626 (inline_write_summary): Rename to ...
28627 (ipa_fn_summary_write): ... this one.
28628 (inline_free_summary): Rename to ...
28629 (ipa_free_fn_summary): ... this one.
28630 (pass_data_local_fn_summary, pass_local_fn_summary,
28631 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
28632 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
28633 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
28634 make_pass_ipa_fn_summary): New.
28635 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
28636 inline_write_summary, inline_free_summary): Remove.
28637 (ipa_free_fn_summary) : New.
28638 * ipa-inline.c (ipa_inline): Update.
28639 (pass_ipa_inline): Do not generate summaries.
28640 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
28642 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
28643 and add pass_ipa_fn_summary.
28644 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
28646 (make_pass_inline_parameters): Remove.
28648 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
28650 * omp-low.c (struct omp_context): Remove "default_kind" member.
28653 * omp-offload.c (execute_oacc_device_lower): Remove the
28654 parallelism dimensions function attributes for unparallelized
28655 OpenACC kernels constructs.
28657 2017-05-23 Martin Liska <mliska@suse.cz>
28659 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
28661 (cgraph_edge::make_speculative): Likewise.
28662 (cgraph_edge::resolve_speculation): Likewise.
28663 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
28664 (cgraph_node::dump): Likewise.
28665 * cgraph.h: Likewise.
28666 * cgraphunit.c (analyze_functions): Likewise.
28667 (symbol_table::compile): Likewise.
28668 * ipa-cp.c (print_all_lattices): Likewise.
28669 (determine_versionability): Likewise.
28670 (initialize_node_lattices): Likewise.
28671 (ipcp_verify_propagated_values): Likewise.
28672 (estimate_local_effects): Likewise.
28673 (update_profiling_info): Likewise.
28674 (create_specialized_node): Likewise.
28675 (perhaps_add_new_callers): Likewise.
28676 (decide_about_value): Likewise.
28677 (decide_whether_version_node): Likewise.
28678 (identify_dead_nodes): Likewise.
28679 (ipcp_store_bits_results): Likewise.
28680 * ipa-devirt.c (dump_targets): Likewise.
28681 (ipa_devirt): Likewise.
28682 * ipa-icf.c (sem_item::dump): Likewise.
28683 (sem_function::equals): Likewise.
28684 (sem_variable::equals): Likewise.
28685 (sem_item_optimizer::read_section): Likewise.
28686 (sem_item_optimizer::execute): Likewise.
28687 (congruence_class::dump): Likewise.
28688 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
28689 (dump_inline_summary): Likewise.
28690 (estimate_node_size_and_time): Likewise.
28691 (inline_analyze_function): Likewise.
28692 * ipa-inline-transform.c (inline_call): Likewise.
28693 * ipa-inline.c (report_inline_failed_reason): Likewise.
28694 (want_early_inline_function_p): Likewise.
28695 (edge_badness): Likewise.
28696 (update_edge_key): Likewise.
28697 (inline_small_functions): Likewise.
28698 * ipa-profile.c (ipa_profile): Likewise.
28699 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
28700 (ipa_make_edge_direct_to_target): Likewise.
28701 (remove_described_reference): Likewise.
28702 (ipa_impossible_devirt_target): Likewise.
28703 (propagate_controlled_uses): Likewise.
28704 (ipa_print_node_params): Likewise.
28705 (ipcp_transform_function): Likewise.
28706 * ipa-pure-const.c (pure_const_read_summary): Likewise.
28707 (propagate_pure_const): Likewise.
28708 * ipa-reference.c (generate_summary): Likewise.
28709 (read_write_all_from_decl): Likewise.
28710 (propagate): Likewise.
28711 (ipa_reference_read_optimization_summary): Likewise.
28712 * ipa-utils.c (ipa_merge_profiles): Likewise.
28713 * ipa.c (walk_polymorphic_call_targets): Likewise.
28714 (symbol_table::remove_unreachable_nodes): Likewise.
28715 (ipa_single_use): Likewise.
28716 * passes.c (execute_todo): Likewise.
28717 * predict.c (drop_profile): Likewise.
28718 * symtab.c (symtab_node::get_dump_name): New function.
28719 (symtab_node::dump_name): Likewise.
28720 (symtab_node::dump_asm_name): Likewise.
28721 (symtab_node::dump_references): Likewise.
28722 (symtab_node::dump_referring): Likewise.
28723 (symtab_node::dump_base): Likewise.
28724 (symtab_node::debug_symtab): Likewise.
28725 * tree-sra.c (convert_callers_for_node): Likewise.
28726 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
28727 * value-prof.c (init_node_map): Likewise.
28729 2017-05-23 Martin Liska <mliska@suse.cz>
28731 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
28732 and symtab_node::debug_symtab to symbol_table::debug.
28733 * cgraphunit.c (analyze_functions): Use the renamed function.
28734 (symbol_table::compile): Likewise.
28735 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
28736 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
28737 * passes.c (execute_todo): Likewise.
28738 * symtab.c (symbol_table::dump): New function.
28739 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
28741 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
28743 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
28744 that nonconst implies exec.
28746 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
28748 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
28749 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
28750 (inline_edge_summary_vec): Turn into ...
28751 (ipa_call_summaries): ... this one.
28752 (redirect_to_unreachable, edge_set_predicate,
28753 evaluate_properties_for_edge, inline_summary_alloc,
28754 reset_ipa_call_summary, reset_inline_summary,
28755 inline_summary_t::duplicate): Update.
28756 (inline_edge_duplication_hook): Turn to ...
28757 (ipa_call_summary_t::duplicate): ... this one.
28758 (inline_edge_removal_hook): Turn to ...
28759 (ipa_call_summary_t::remove): ... this one.
28760 (dump_inline_edge_summary): Turn to ...
28761 (dump_ipa_call_summary): ... this one.
28762 (estimate_function_body_sizes): Update.
28763 (inline_update_callee_summaries): Update.
28764 (remap_edge_change_prob): Update.
28765 (remap_edge_summaries): Update.
28766 (inline_merge_summary): Update.
28767 (do_estimate_edge_time): Update.
28768 (inline_generate_summary): Update.
28769 (inline_read_section): Update.
28770 (inline_read_summary): Update.
28771 (inline_free_summary): Update.
28772 * ipa-inline.c (can_inline_edge_p): Update.
28773 (compute_inlined_call_time): Update.
28774 (want_inline_small_function_p): Update.
28775 (edge_badness): Update.
28776 (early_inliner): Update.
28777 * ipa-inline.h (inline_edge_summary): Turn to ...
28778 (ipa_call_summary): ... this one.
28779 (ipa_call_summary_t): New class.
28780 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
28781 (ipa_call_summaries): New.
28782 (inline_edge_summary): Remove.
28783 (estimate_edge_growth): Update.
28784 * ipa-profile.c (ipa_propagate_frequency_1): Update.
28785 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
28786 * ipa-split.c (execute_split_functions): Update.
28787 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
28789 2017-05-23 Tom de Vries <tom@codesourcery.com>
28791 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
28792 attributes): Document rdrand effective target.
28794 2017-05-23 Tom de Vries <tom@codesourcery.com>
28796 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
28797 attributes): Sort alphabetically.
28799 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
28801 * config/avr/genmultilib.awk: Use gsub instead of gensub.
28803 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
28806 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
28807 V2DF/V2DI splat into two separate patterns, one that handles
28808 registers, and the other that only handles memory. Drop support
28809 for splatting from a GPR on ISA 2.07 and then splitting the
28810 splat into direct move and splat.
28811 (vsx_splat_<mode>_reg): Likewise.
28812 (vsx_splat_<mode>_mem): Likewise.
28814 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
28816 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
28818 2017-05-22 Jakub Jelinek <jakub@redhat.com>
28820 PR middle-end/80809
28821 * omp-low.c (finish_taskreg_remap): New function.
28822 (finish_taskreg_scan): If unit size of ctx->record_type
28823 is non-constant, unshare the size expression and replace
28824 decls in it with possible outer var refs.
28826 PR middle-end/80809
28827 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
28828 GOVD_SHARED rather than GOVD_PRIVATE with it.
28829 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
28830 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
28832 PR middle-end/80853
28833 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
28834 as last argument to build_outer_var_ref for pointer bases of array
28835 section reductions.
28837 2017-05-19 Martin Sebor <msebor@redhat.com>
28839 * print-tree.c (print_node): Print DECL_READ_P flag.
28841 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
28843 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
28844 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
28845 * cgraph.c: Likewise.
28846 * cgraphunit.c: Likewise.
28847 * gengtype.c: Likewise.
28848 * ipa-cp.c: Likewise.
28849 * ipa-devirt.c: Likewise.
28850 * ipa-icf.c: Likewise.
28851 * ipa-predicate.c: Likewise.
28852 * ipa-profile.c: Likewise.
28853 * ipa-prop.c: Likewise.
28854 * ipa-split.c: Likewise.
28856 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
28857 edge_predicate_pool, dump_inline_hints,
28858 inline_summary::account_size_time, redirect_to_unreachable,
28859 edge_set_predicate, set_hint_predicate,
28860 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
28861 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
28862 inline_summary_t::remove, remap_hint_predicate_after_duplication,
28863 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
28864 ipa_call_summary_t::remove, initialize_growth_caches,
28865 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
28866 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
28867 mark_modified, unmodified_parm_1, unmodified_parm,
28868 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
28869 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
28870 compute_bb_predicates, will_be_nonconstant_expr_predicate,
28871 will_be_nonconstant_predicate, record_modified_bb_info,
28872 get_minimal_bb, record_modified, param_change_prob,
28873 phi_result_unknown_predicate, predicate_for_phi_result,
28874 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
28875 estimate_function_body_sizes, compute_inline_parameters,
28876 compute_inline_parameters_for_curren, pass_data_inline_parameters,
28877 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
28878 inline_update_callee_summaries, remap_edge_change_prob,
28879 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
28880 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
28881 inline_analyze_function, inline_summary_t::insert,
28882 inline_generate_summary, read_ipa_call_summary, inline_read_section,
28883 inline_read_summary, write_ipa_call_summary, inline_write_summary,
28884 inline_free_summary): Move to ipa-fnsummary.h
28885 (predicate_t): Remove.
28886 * ipa-fnsummary.c: New file.
28887 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
28888 (enum inline_hints_vals, inline_hints, agg_position_info,
28889 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
28890 inline_summaries, ipa_call_summary, ipa_call_summary_t,
28891 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
28892 dump_inline_summary, dump_inline_hints, inline_generate_summary,
28893 inline_read_summary, inline_write_summary, inline_free_summary,
28894 inline_analyze_function, initialize_inline_failed,
28895 inline_merge_summary, inline_update_overall_summary,
28896 compute_inline_parameters): Move to ipa-fnsummary.h
28897 * ipa-fnsummary.h: New file.
28898 * ipa-inline-transform.h: Include ipa-inline.h.
28899 * ipa-inline.c: LIkewise.
28901 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
28903 * ipa-inline.c (edge_badness): Use inlined_time instead of
28904 inline_summaries->get.
28906 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
28908 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
28910 2017-05-22 Nathan Sidwell <nathan@acm.org>
28912 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
28913 (fdump-lang): Document 'raw' option.
28914 * dumpfile.h (TDI_tu): Delete.
28915 * dumpfile.c (dump_files): Remove translation-unit.
28916 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
28918 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
28920 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
28921 command option from $(AWK) call.
28922 * config/avr/genmultilib.awk: Simplify and rewrite so that it
28923 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
28924 [FORMAT]: Remove handling of variable.
28925 * config/avr/t-multilib: Regenerate.
28927 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
28929 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
28931 (dump_inline_summary): Do not print self_time.
28932 (estimate_function_body_sizes): Do not set self_time.
28933 (compute_inline_parameters): Likewise.
28934 (inline_read_section, inline_write_summary): Do not stream self_time.
28935 * ipa-inline.h (inline_summary): Drop self_time.
28937 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
28939 * ipa-inline-analysis.c (account_size_time): Rename to ...
28940 (inline_summary::account_size_time): ... this one.
28941 (reset_ipa_call_summary): Turn to ...
28942 (ipa_call_summary::reset): ... this one.
28943 (reset_inline_summary): Turn to ...
28944 (inline_summary::reset): ... this one.
28945 (inline_summary_t::remove): Update.
28946 (inline_summary_t::duplicate): Update.
28947 (ipa_call_summary_t::remove): Update.
28948 (dump_inline_summary): Update.
28949 (estimate_function_body_sizes): Update.
28950 (compute_inline_parameters): Update.
28951 (estimate_node_size_and_time): Update.
28952 (inline_merge_summary): Update.
28953 (inline_update_overall_summary): Update.
28954 (inline_read_section): Update.
28955 (inline_write_summary): Update.
28956 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
28957 add account_size_time and reset member functions.
28958 (ipa_call_summary): Add reset function.
28959 * ipa-predicate.h (predicate::operator &): Constify.
28961 2017-05-22 Richard Biener <rguenther@suse.de>
28963 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
28965 2017-05-19 Jason Merrill <jason@redhat.com>
28967 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
28969 2017-05-19 Marek Polacek <polacek@redhat.com>
28972 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
28973 TYPE_OVERFLOW_WRAPS checks.
28975 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
28977 * tree-core.h (enum omp_clause_default_kind): Add
28978 "OMP_CLAUSE_DEFAULT_PRESENT".
28979 * tree-pretty-print.c (dump_omp_clause): Handle it.
28980 * gimplify.c (enum gimplify_omp_var_data): Add
28981 "GOVD_MAP_FORCE_PRESENT".
28982 (gimplify_adjust_omp_clauses_1): Map it to
28983 "GOMP_MAP_FORCE_PRESENT".
28984 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
28986 * gimplify.c (oacc_default_clause): Clarify.
28988 2017-05-19 Nathan Sidwell <nathan@acm.org>
28990 LANG_HOOK_REGISTER_DUMPS
28991 * toplev.c (general_init): Call register dump lang hook.
28992 * doc/invoke.texi: Document -fdump-lang option family.
28993 * dumpfile.c (dump_files): Remove class dump here.
28994 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
28995 * dumpfile.h (tree_dump_index): Remove TDI_class.
28996 * langhooks-def.h (lhd_register_dumps): Declare.
28997 (LANG_HOOKS_REGISTER_DUMPS): Define.
28998 (LANG_HOOKS_INITIALIZER): Add it.
28999 * langhooks.c (lhd_register_dumps): Define.
29000 * langhooks.h (struct lang_hooks): Add register_dumps.
29002 2017-05-19 Nathan Sidwell <nathan@acm.org>
29004 * context.h (context::set_passes): New.
29005 * context.c (context::context): Do not create pass manager.
29006 * toplev.c (general_init): Create pass manager here.
29008 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
29010 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
29011 use this splitter if two add or or instructions would also work for
29012 the constant we want to generate.
29014 2017-05-19 Richard Biener <rguenther@suse.de>
29017 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
29018 predicate evaluation.
29020 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
29022 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
29024 * ipa-inline.c (want_inline_small_function_p): Do not cast to
29027 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
29029 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
29030 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
29031 (inline_edge_summary_vec): Turn into ...
29032 (ipa_call_summaries): ... this one.
29033 (redirect_to_unreachable, edge_set_predicate,
29034 evaluate_properties_for_edge, inline_summary_alloc,
29035 reset_ipa_call_summary, reset_inline_summary,
29036 inline_summary_t::duplicate): Update.
29037 (inline_edge_duplication_hook): Turn to ...
29038 (ipa_call_summary_t::duplicate): ... this one.
29039 (inline_edge_removal_hook): Turn to ...
29040 (ipa_call_summary_t::remove): ... this one.
29041 (dump_inline_edge_summary): Turn to ...
29042 (dump_ipa_call_summary): ... this one.
29043 (estimate_function_body_sizes): Update.
29044 (inline_update_callee_summaries): Update.
29045 (remap_edge_change_prob): Update.
29046 (remap_edge_summaries): Update.
29047 (inline_merge_summary): Update.
29048 (do_estimate_edge_time): Update.
29049 (inline_generate_summary): Update.
29050 (inline_read_section): Update.
29051 (inline_read_summary): Update.
29052 (inline_free_summary): Update.
29053 * ipa-inline.c (can_inline_edge_p): Update.
29054 (compute_inlined_call_time): Update.
29055 (want_inline_small_function_p): Update.
29056 (edge_badness): Update.
29057 (early_inliner): Update.
29058 * ipa-inline.h (inline_edge_summary): Turn to ...
29059 (ipa_call_summary): ... this one.
29060 (ipa_call_summary_t): New class.
29061 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
29062 (ipa_call_summaries): New.
29063 (inline_edge_summary): Remove.
29064 (estimate_edge_growth): Update.
29065 * ipa-profile.c (ipa_propagate_frequency_1): Update.
29066 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
29067 * ipa-split.c (execute_split_functions): Update.
29068 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
29070 2017-05-19 Richard Biener <rguenther@suse.de>
29072 PR middle-end/80764
29073 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
29075 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
29077 * config/rs6000/rs6000.c (struct machine_function): Add field
29078 fpr_is_wrapped_separately.
29079 (rs6000_get_separate_components): Use 64 components. Handle the
29080 new FPR components.
29081 (rs6000_components_for_bb): Handle the FPR components.
29082 (rs6000_emit_prologue_components): Handle the FPR components.
29083 (rs6000_emit_epilogue_components): Handle the FPR components.
29084 (rs6000_set_handled_components): Handle the FPR components.
29085 (rs6000_emit_prologue): Don't output prologue code for those FPRs
29086 that are already separately shrink-wrapped.
29087 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
29088 that are already separately shrink-wrapped.
29090 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
29093 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
29096 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
29097 (define_peephole2 for Altivec d-form load): Add peepholes to catch
29098 cases where the register allocator uses a move and an offsettable
29099 memory operation to/from a FPR register on ISA 2.06/2.07.
29100 (define_peephole2 for Altivec d-form store): Likewise.
29102 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
29105 * config/i386/mmx.md (*mov<mode>_internal): Enable
29106 alternatives 11, 12, 13 and 14 also for 32bit targets.
29107 Remove alternatives 15, 16, 17 and 18.
29108 * config/i386/sse.md (vec_concatv2di): Change
29109 alternative (!x, *y) to (x, ?!*Yn).
29111 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
29113 * dumpfile.h (enum dump_kind): Remove stray comma.
29115 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
29117 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
29118 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
29119 predicate::num_conditions
29120 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
29121 (CHANGED): turn into predicate::changed.
29122 (agg_position_info): Move to ipa-predicate.h
29123 (add_condition, predicate::add_clause, predicate::operator &=,
29124 predicate::or_with, predicate::evaluate, predicate::probability,
29125 dump_condition, dump_clause, predicate::dump,
29126 predicate::remap_after_duplication, predicate::remap_after_inlining,
29127 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
29128 (evaluate_conditions_for_known_args): Update.
29129 (set_cond_stmt_execution_predicate): Update.
29130 * ipa-inline.h: Include ipa-predicate.h
29131 (condition, inline_param_summary, conditions, agg_position_info,
29132 predicate): Move to ipa-predicate.h
29133 * ipa-predicate.c: New file.
29134 * ipa-predicate.h: New file.
29136 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
29138 * final.c (leaf_function_p): Check we are not in a sequence.
29140 2017-05-18 Martin Liska <mliska@suse.cz>
29142 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
29143 * dumpfile.c (dump_register): Use new enum dump_kind.
29144 (get_dump_file_name): Likewise.
29145 (dump_enable_all): Likewise.
29146 (dump_switch_p_1): Likewise.
29147 (enable_rtl_dump_file): Remove usage of TDF_RTL.
29148 * dumpfile.h (enum dump_kind): New enum type.
29149 (struct dump_file_info): Create constructor and
29150 format fields and comments.
29151 * passes.c (pass_manager::register_one_dump_file):
29153 * statistics.c (statistics_early_init): Likewise.
29154 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
29155 TDF_TREE with TDF_SLIM.
29156 (gather_memory_references_ref): Likewise.
29158 2017-05-18 Martin Liska <mliska@suse.cz>
29160 * vec.h (struct vnull): Use it.
29162 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
29164 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
29165 (true_predicate, false_predicate, true_predicate_p,
29166 false_predicate_p): Remove.
29167 (single_cond_predicate, not_inlined_predicate): Turn to member function
29169 (add_condition): Update.
29170 (add_clause): Turn to...
29171 (predicate::add_clause): ... this one; update; allow passing NULL
29173 (and_predicates): Turn to ...
29174 (predicate::operator &=): ... this one.
29175 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
29176 (or_predicates): Turn to ...
29177 (predicate::or_with): ... this one.
29178 (evaluate_predicate): Turn to ...
29179 (predicate::evaluate): ... this one.
29180 (predicate_probability): Turn to ...
29181 (predicate::probability): ... this one.
29182 (dump_condition): Update.
29183 (dump_predicate): Turn to ...
29184 (predicate::dump): ... this one.
29185 (account_size_time): Update.
29186 (edge_set_predicate): Update.
29187 (set_hint_predicate): UPdate.
29188 (evaluate_conditions_for_known_args): Update.
29189 (evaluate_properties_for_edge): Update.
29190 (remap_predicate_after_duplication): Turn to...
29191 (predicate::remap_after_duplication): ... this one.
29192 (remap_hint_predicate_after_duplication): Update.
29193 (inline_summary_t::duplicate): UPdate.
29194 (dump_inline_edge_summary): Update.
29195 (dump_inline_summary): Update.
29196 (set_cond_stmt_execution_predicate): Update.
29197 (set_switch_stmt_execution_predicate): Update.
29198 (compute_bb_predicates): Update.
29199 (will_be_nonconstant_expr_predicate): Update.
29200 (will_be_nonconstant_predicate): Update.
29201 (phi_result_unknown_predicate): Update.
29202 (predicate_for_phi_result): Update.
29203 (array_index_predicate): Update.
29204 (estimate_function_body_sizes): Update.
29205 (estimate_node_size_and_time): Update.
29206 (estimate_ipcp_clone_size_and_time): Update.
29207 (remap_predicate): Rename to ...
29208 (predicate::remap_after_inlining): ... this one.
29209 (remap_hint_predicate): Update.
29210 (inline_merge_summary): Update.
29211 (inline_update_overall_summary): Update.
29212 (estimate_size_after_inlining): Update.
29213 (read_predicate): Rename to ...
29214 (predicate::stream_in): ... this one.
29215 (read_inline_edge_summary): Update.
29216 (write_predicate): Rename to ...
29217 (predicate::stream_out): ... this one.
29218 (write_inline_edge_summary): Update.
29219 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
29220 (clause_t): Turn to uint32_t
29221 (predicate): Turn to class; implement constructor and operators
29223 (size_time_entry): Update.
29224 (inline_summary): Update.
29225 (inline_edge_summary): Update.
29227 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
29229 * fold-const.c (fold_binary_loc): Move transformation...
29230 * match.pd (C - X CMP X): ... here.
29232 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
29234 * config/sparc/sparc.c (sparc_option_override): Set function
29235 alignment for -mcpu=niagara7 to 64 to match the I$ line.
29236 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
29238 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
29240 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
29242 2017-05-18 Marek Polacek <polacek@redhat.com>
29245 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
29246 (pass_ubsan::execute): Call gimple_assign_single_p instead of
29247 gimple_assign_load_p.
29249 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
29251 PR middle-end/80692
29252 * real.c (do_compare): Give decimal_do_compare preference over
29253 comparing just the signs.
29255 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
29257 * doc/md.texi (Canonicalization of Instructions): Describe the
29258 canonical form of instructions that inherently set a condition
29261 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
29263 PR middle-end/80775
29264 * tree-cfg.c: Move deletion of unreachable case statements to after
29265 the merging of consecutive case labels.
29267 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
29269 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
29270 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
29271 restoring of callee-saved registers.
29273 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
29275 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
29276 * config/visium/visium.c (single_set_and_flags): Likewise.
29277 * config/visium/visium.md (Substitutions): Likewise.
29279 2017-05-17 Martin Liska <mliska@suse.cz>
29281 * cfg.c: Introduce dump_flags_t type and
29282 use it instead of int type.
29284 * cfghooks.c: Likewise.
29285 * cfghooks.h (struct cfg_hooks): Likewise.
29286 * cfgrtl.c: Likewise.
29287 * cfgrtl.h: Likewise.
29288 * cgraph.c (cgraph_node::get_body): Likewise.
29289 * coretypes.h: Likewise.
29290 * domwalk.c: Likewise.
29291 * domwalk.h: Likewise.
29292 * dumpfile.c (struct dump_option_value_info): Likewise.
29293 (dump_enable_all): Likewise.
29294 (dump_switch_p_1): Likewise.
29295 (opt_info_switch_p): Likewise.
29296 * dumpfile.h (enum tree_dump_index): Likewise.
29297 (struct dump_file_info): Likewise.
29298 * genemit.c: Likewise.
29299 * generic-match-head.c: Likewise.
29300 * gengtype.c (open_base_files): Likewise.
29301 * gimple-pretty-print.c: Likewise.
29302 * gimple-pretty-print.h: Likewise.
29303 * graph.c (print_graph_cfg): Likewise.
29304 * graphite-scop-detection.c (dot_all_sese): Likewise.
29305 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
29306 * loop-unroll.c (report_unroll): Likewise.
29307 * passes.c (pass_manager::register_one_dump_file): Likewise.
29308 * print-tree.c: Likewise.
29309 * statistics.c: Likewise.
29310 * tree-cfg.c: Likewise.
29311 * tree-cfg.h: Likewise.
29312 * tree-dfa.c: Likewise.
29313 * tree-dfa.h: Likewise.
29314 * tree-dump.c (dump_function): Likewise.
29315 * tree-dump.h (struct dump_info): Likewise.
29316 * tree-pretty-print.c: Likewise.
29317 * tree-pretty-print.h: Likewise.
29318 * tree-ssa-live.c: Likewise.
29319 * tree-ssa-live.h: Likewise.
29320 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
29321 * tree-vect-loop.c: Likewise.
29322 * tree-vect-slp.c: Likewise.
29324 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
29325 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29327 PR tree-optimization/80457
29328 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
29329 of all arguments to a statement as scalar_to_vec operations.
29330 (vectorizable_call): Adjust call to vect_model_simple_cost for
29332 (vectorizable_conversion): Likewise.
29333 (vectorizable_assignment): Likewise.
29334 (vectorizable_shift): Likewise.
29335 (vectorizable_operation): Likewise.
29336 (vectorizable_comparison): Likewise.
29337 (vect_is_simple_cond): Record the def types for operands.
29338 (vectorizable_condition): Likewise, call vect_model_simple_cost.
29339 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
29340 for statement argument count.
29342 2017-05-16 Carl Love <cel@us.ibm.com>
29344 * config/rs6000/rs6000-c: Add support for built-in functions
29345 vector unsigned long long vec_bperm (vector unsigned long long,
29346 vector unsigned char)
29347 vector signed long long vec_mule (vector signed int,
29349 vector unsigned long long vec_mule (vector unsigned int,
29350 vector unsigned int)
29351 vector signed long long vec_mulo (vector signed int,
29353 vector unsigned long long vec_mulo (vector unsigned int,
29354 vector unsigned int)
29355 vector signed char vec_sldw (vector signed char,
29356 vector signed char,
29358 vector unsigned char vec_sldw (vector unsigned char,
29359 vector unsigned char,
29361 vector signed short vec_sldw (vector signed short,
29362 vector signed short,
29364 vector unsigned short vec_sldw (vector unsigned short,
29365 vector unsigned short,
29367 vector signed int vec_sldw (vector signed int,
29370 vector unsigned int vec_sldw (vector unsigned int,
29371 vector unsigned int,
29373 vector signed long long vec_sldw (vector signed long long,
29374 vector signed long long,
29376 vector unsigned long long vec_sldw (vector unsigned long long,
29377 vector unsigned long long,
29379 * config/rs6000/rs6000-c: Add support for built-in functions
29380 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
29381 * config/rs6000/altivec.h: Add defintion for vec_sldw.
29382 * doc/extend.texi: Update the built-in documentation for the
29383 new built-in functions.
29385 2017-05-16 Marek Polacek <polacek@redhat.com>
29389 * tree.c (save_expr): Don't fold the expression.
29391 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
29393 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
29394 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
29395 and (?*y,m). Update insn attributes.
29397 2017-05-16 Martin Liska <mliska@suse.cz>
29399 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
29400 flags argument of print_gimple_stmt, print_gimple_expr,
29401 print_generic_stmt and print_generic_expr.
29402 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
29403 * coretypes.h: Likewise.
29404 * except.c (dump_eh_tree): Likewise.
29405 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
29406 * gimple-pretty-print.h: Likewise.
29407 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
29408 (backprop::push_to_worklist): Likewise.
29409 (backprop::pop_from_worklist): Likewise.
29410 (backprop::process_use): Likewise.
29411 (backprop::intersect_uses): Likewise.
29412 (note_replacement): Likewise.
29413 * gimple-ssa-store-merging.c
29414 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
29415 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
29416 (pass_store_merging::execute): Likewise.
29417 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
29418 (ssa_base_cand_dump_callback): Likewise.
29419 (dump_incr_vec): Likewise.
29420 (replace_refs): Likewise.
29421 (replace_mult_candidate): Likewise.
29422 (create_add_on_incoming_edge): Likewise.
29423 (create_phi_basis): Likewise.
29424 (insert_initializers): Likewise.
29425 (all_phi_incrs_profitable): Likewise.
29426 (introduce_cast_before_cand): Likewise.
29427 (replace_one_candidate): Likewise.
29428 * gimplify.c (gimplify_expr): Likewise.
29429 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
29430 (set_rename): Likewise.
29431 (rename_uses): Likewise.
29432 (copy_loop_phi_nodes): Likewise.
29433 (add_close_phis_to_merge_points): Likewise.
29434 (copy_loop_close_phi_args): Likewise.
29435 (copy_cond_phi_args): Likewise.
29436 (graphite_copy_stmts_from_block): Likewise.
29437 (translate_pending_phi_nodes): Likewise.
29438 * graphite-poly.c (print_pdr): Likewise.
29439 (dump_gbb_cases): Likewise.
29440 (dump_gbb_conditions): Likewise.
29441 (print_scop_params): Likewise.
29442 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
29443 (build_cross_bb_scalars_use): Likewise.
29444 (gather_bbs::before_dom_children): Likewise.
29445 * hsa-dump.c (dump_hsa_immed): Likewise.
29446 * ipa-cp.c (print_ipcp_constant_value): Likewise.
29447 (get_replacement_map): Likewise.
29448 * ipa-inline-analysis.c (dump_condition): Likewise.
29449 (estimate_function_body_sizes): Likewise.
29450 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
29451 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
29452 * ipa-prop.c (ipa_dump_param): Likewise.
29453 (ipa_print_node_jump_functions_for_edge): Likewise.
29454 (ipa_modify_call_arguments): Likewise.
29455 (ipa_modify_expr): Likewise.
29456 (ipa_dump_param_adjustments): Likewise.
29457 (ipa_dump_agg_replacement_values): Likewise.
29458 (ipcp_modif_dom_walker::before_dom_children): Likewise.
29459 * ipa-pure-const.c (check_stmt): Likewise.
29460 (pass_nothrow::execute): Likewise.
29461 * ipa-split.c (execute_split_functions): Likewise.
29462 * omp-offload.c (dump_oacc_loop_part): Likewise.
29463 (dump_oacc_loop): Likewise.
29464 * trans-mem.c (tm_log_emit): Likewise.
29465 (tm_memopt_accumulate_memops): Likewise.
29466 (dump_tm_memopt_set): Likewise.
29467 (dump_tm_memopt_transform): Likewise.
29468 * tree-cfg.c (gimple_verify_flow_info): Likewise.
29469 (print_loop): Likewise.
29470 * tree-chkp-opt.c (chkp_print_addr): Likewise.
29471 (chkp_gather_checks_info): Likewise.
29472 (chkp_get_check_result): Likewise.
29473 (chkp_remove_check_if_pass): Likewise.
29474 (chkp_use_outer_bounds_if_possible): Likewise.
29475 (chkp_reduce_bounds_lifetime): Likewise.
29476 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
29477 (chkp_mark_completed_bounds): Likewise.
29478 (chkp_register_incomplete_bounds): Likewise.
29479 (chkp_mark_invalid_bounds): Likewise.
29480 (chkp_maybe_copy_and_register_bounds): Likewise.
29481 (chkp_build_returned_bound): Likewise.
29482 (chkp_get_bound_for_parm): Likewise.
29483 (chkp_build_bndldx): Likewise.
29484 (chkp_get_bounds_by_definition): Likewise.
29485 (chkp_generate_extern_var_bounds): Likewise.
29486 (chkp_get_bounds_for_decl_addr): Likewise.
29487 * tree-chrec.c (chrec_apply): Likewise.
29488 * tree-data-ref.c (dump_data_reference): Likewise.
29489 (dump_subscript): Likewise.
29490 (dump_data_dependence_relation): Likewise.
29491 (analyze_overlapping_iterations): Likewise.
29492 * tree-inline.c (expand_call_inline): Likewise.
29493 (tree_function_versioning): Likewise.
29494 * tree-into-ssa.c (dump_defs_stack): Likewise.
29495 (dump_currdefs): Likewise.
29496 (dump_names_replaced_by): Likewise.
29497 (dump_update_ssa): Likewise.
29498 (update_ssa): Likewise.
29499 * tree-object-size.c (pass_object_sizes::execute): Likewise.
29500 * tree-parloops.c (build_new_reduction): Likewise.
29501 (try_create_reduction_list): Likewise.
29502 (ref_conflicts_with_region): Likewise.
29503 (oacc_entry_exit_ok_1): Likewise.
29504 (oacc_entry_exit_single_gang): Likewise.
29505 * tree-pretty-print.h: Likewise.
29506 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
29507 (get_scalar_evolution): Likewise.
29508 (add_to_evolution): Likewise.
29509 (get_loop_exit_condition): Likewise.
29510 (analyze_evolution_in_loop): Likewise.
29511 (analyze_initial_condition): Likewise.
29512 (analyze_scalar_evolution): Likewise.
29513 (instantiate_scev): Likewise.
29514 (number_of_latch_executions): Likewise.
29515 (gather_chrec_stats): Likewise.
29516 (final_value_replacement_loop): Likewise.
29517 (scev_const_prop): Likewise.
29518 * tree-sra.c (dump_access): Likewise.
29519 (disqualify_candidate): Likewise.
29520 (create_access): Likewise.
29521 (reject): Likewise.
29522 (maybe_add_sra_candidate): Likewise.
29523 (create_access_replacement): Likewise.
29524 (analyze_access_subtree): Likewise.
29525 (analyze_all_variable_accesses): Likewise.
29526 (sra_modify_assign): Likewise.
29527 (initialize_constant_pool_replacements): Likewise.
29528 (find_param_candidates): Likewise.
29529 (decide_one_param_reduction): Likewise.
29530 (replace_removed_params_ssa_names): Likewise.
29531 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
29532 * tree-ssa-copy.c (dump_copy_of): Likewise.
29533 (copy_prop_visit_cond_stmt): Likewise.
29534 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
29535 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
29536 (record_equivalences_from_stmt): Likewise.
29537 * tree-ssa-dse.c (compute_trims): Likewise.
29538 (delete_dead_call): Likewise.
29539 (delete_dead_assignment): Likewise.
29540 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
29541 (forward_propagate_into_cond): Likewise.
29542 (pass_forwprop::execute): Likewise.
29543 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
29544 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
29546 (move_computations_worker): Likewise.
29547 (execute_sm): Likewise.
29548 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
29549 (remove_exits_and_undefined_stmts): Likewise.
29550 (remove_redundant_iv_tests): Likewise.
29551 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
29552 (adjust_iv_update_pos): Likewise.
29553 * tree-ssa-math-opts.c (bswap_replace): Likewise.
29554 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
29555 (value_replacement): Likewise.
29556 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
29557 * tree-ssa-pre.c (print_pre_expr): Likewise.
29558 (get_representative_for): Likewise.
29559 (create_expression_by_pieces): Likewise.
29560 (insert_into_preds_of_block): Likewise.
29561 (eliminate_insert): Likewise.
29562 (eliminate_dom_walker::before_dom_children): Likewise.
29563 (eliminate): Likewise.
29564 (remove_dead_inserted_code): Likewise.
29565 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
29566 * tree-ssa-reassoc.c (get_rank): Likewise.
29567 (eliminate_duplicate_pair): Likewise.
29568 (eliminate_plus_minus_pair): Likewise.
29569 (eliminate_not_pairs): Likewise.
29570 (undistribute_ops_list): Likewise.
29571 (eliminate_redundant_comparison): Likewise.
29572 (update_range_test): Likewise.
29573 (optimize_range_tests_var_bound): Likewise.
29574 (optimize_vec_cond_expr): Likewise.
29575 (rewrite_expr_tree): Likewise.
29576 (rewrite_expr_tree_parallel): Likewise.
29577 (linearize_expr): Likewise.
29578 (break_up_subtract): Likewise.
29579 (linearize_expr_tree): Likewise.
29580 (attempt_builtin_powi): Likewise.
29581 (attempt_builtin_copysign): Likewise.
29582 (transform_stmt_to_copy): Likewise.
29583 (transform_stmt_to_multiply): Likewise.
29584 (dump_ops_vector): Likewise.
29585 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
29586 (print_scc): Likewise.
29587 (set_ssa_val_to): Likewise.
29588 (visit_reference_op_store): Likewise.
29589 (visit_use): Likewise.
29590 (sccvn_dom_walker::before_dom_children): Likewise.
29591 (run_scc_vn): Likewise.
29592 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
29594 (expr_hash_elt::print): Likewise.
29595 (const_and_copies::pop_to_marker): Likewise.
29596 (const_and_copies::record_const_or_copy_raw): Likewise.
29597 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
29598 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
29599 (dump_predicates): Likewise.
29600 (find_uninit_use): Likewise.
29601 (warn_uninitialized_phi): Likewise.
29602 (pass_late_warn_uninitialized::execute): Likewise.
29603 * tree-ssa.c (verify_vssa): Likewise.
29604 (verify_ssa): Likewise.
29605 (maybe_optimize_var): Likewise.
29606 * tree-vrp.c (dump_value_range): Likewise.
29607 (dump_all_value_ranges): Likewise.
29608 (dump_asserts_for): Likewise.
29609 (register_edge_assert_for_2): Likewise.
29610 (vrp_visit_cond_stmt): Likewise.
29611 (vrp_visit_switch_stmt): Likewise.
29612 (vrp_visit_stmt): Likewise.
29613 (vrp_visit_phi_node): Likewise.
29614 (simplify_cond_using_ranges_1): Likewise.
29615 (fold_predicate_in): Likewise.
29616 (evrp_dom_walker::before_dom_children): Likewise.
29617 (evrp_dom_walker::push_value_range): Likewise.
29618 (evrp_dom_walker::pop_value_range): Likewise.
29619 (execute_early_vrp): Likewise.
29621 2017-05-16 Richard Biener <rguenther@suse.de>
29623 * dwarf2out.c (loc_list_from_tree_1): Do not create
29624 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
29626 2017-05-16 Richard Biener <rguenther@suse.de>
29628 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
29630 (note_variable_value_in_expr): If we resolved the decl ref
29631 do not push to the stack.
29633 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
29635 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
29636 operations in fast-math mode.
29637 (vaddq_f16): Likewise.
29638 (vmul_f16): Likewise.
29639 (vmulq_f16): Likewise.
29640 (vsub_f16): Likewise.
29641 (vsubq_f16): Likewise.
29642 * config/arm/neon.md (add<mode>3): New.
29644 (fma:<VH:mode>3): New. Also remove outdated comment.
29647 2017-05-16 Martin Liska <mliska@suse.cz>
29651 * ipa-devirt.c (warn_types_mismatch): Fix typo.
29652 (odr_types_equivalent_p): Likewise.
29654 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
29656 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
29658 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
29661 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
29662 non-interunit SSE move alternatives with '?'.
29663 (zero-extendsidi peephole2): New peephole to skip intermediate
29664 general register in SSE zero-extend sequence.
29666 2017-05-15 Jeff Law <law@redhat.com>
29668 * reorg.c (relax_delay_slots): Create a new variable to hold
29669 the temporary target rather than clobbering TARGET_LABEL.
29671 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
29672 missing argument to extract_bit_field call.
29673 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
29675 2017-05-15 Martin Liska <mliska@suse.cz>
29678 * gcc.c (process_command): Do not allow empty argument of -o option.
29680 2017-05-15 Renlin Li <renlin.li@arm.com>
29682 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
29683 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
29684 * config/aarch64/constraints.md (Usf): Add long call check.
29685 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
29686 (call_value): Likewise.
29687 (sibcall): Likewise.
29688 (sibcall_value): Likewise.
29690 (call_value_insn): New.
29691 (sibcall_insn): Update rtx pattern.
29692 (sibcall_value_insn): Likewise.
29693 (call_internal): Remove.
29694 (call_value_internal): Likewise.
29695 (sibcall_internal): Likewise.
29696 (sibcall_value_internal): Likewise.
29697 (call_reg): Likewise.
29698 (call_symbol): Likewise.
29699 (call_value_reg): Likewise.
29700 (call_value_symbol): Likewise.
29702 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
29705 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
29707 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
29709 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
29710 compatible with CCGOCmode and with CCZmode.
29712 2017-05-14 Martin Sebor <msebor@redhat.com>
29714 PR middle-end/77671
29715 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
29716 (gimple_fold_builtin_snprintf): Same.
29717 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
29718 (gimple_fold_builtin_snprintf): Same.
29719 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
29720 of character types.
29721 (is_call_safe): New function.
29722 (try_substitute_return_value): Call it.
29723 (try_simplify_call): New function.
29724 (pass_sprintf_length::handle_gimple_call): Call it.
29726 2017-05-14 Martin Sebor <msebor@redhat.com>
29728 PR middle-end/80669
29729 * builtins.c (expand_builtin_stpncpy): Simplify.
29731 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
29733 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
29734 * config/i386/i386.h
29735 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
29736 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
29737 (struct machine_function): Add new members call_ms2sysv,
29738 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
29739 (struct machine_frame_state): New fields sp_realigned and
29740 sp_realigned_offset.
29741 * config/i386/i386.c
29742 (enum xlogue_stub): New enum.
29743 (enum xlogue_stub_sets): New enum.
29744 (class xlogue_layout): New class.
29745 (struct ix86_frame): New fields stack_realign_allocate_offset,
29746 stack_realign_offset and outlined_save_offset. Modify comments to
29747 detail stack layout when using out-of-line stubs.
29748 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
29749 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
29750 -mcall-ms2sysv-xlogues.
29751 (stub_managed_regs): New static variable.
29752 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
29753 registers managed by out-of-line stub.
29754 (disable_call_ms2sysv_xlogues): New function.
29755 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
29756 m->call_ms2sysv when appropriate and compute frame layout for
29758 (sp_valid_at, fp_valid_at): New inline functions.
29759 (choose_basereg): New function.
29760 (choose_baseaddr): Add align parameter, use choose_basereg and modify
29762 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
29763 Use align parameter of choose_baseaddr to generated aligned SSE movs
29765 (pro_epilogue_adjust_stack): Modify to track
29766 machine_frame_state::sp_realigned.
29767 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
29768 (ix86_nsaved_sseregs): Likewise.
29769 (ix86_emit_save_regs): Likewise.
29770 (ix86_emit_save_regs_using_mov): Likewise.
29771 (ix86_emit_save_sse_regs_using_mov): Likewise.
29772 (get_scratch_register_on_entry): Likewise.
29773 (gen_frame_set): New function.
29774 (gen_frame_load): Likewise.
29775 (gen_frame_store): Likewise.
29776 (emit_outlined_ms2sysv_save): Likewise.
29777 (emit_outlined_ms2sysv_restore): Likewise.
29778 (ix86_expand_prologue): Modify stack re-alignment code and call
29779 emit_outlined_ms2sysv_save when appropriate.
29780 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
29781 parameter rtx_insn *insn, which allows the function to be used to only
29782 generate the notes.
29783 (ix86_expand_epilogue): Modify validity checks of frame and stack
29784 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
29785 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
29786 * config/i386/predicates.md
29787 (save_multiple): New predicate.
29788 (restore_multiple): Likewise.
29789 * config/i386/sse.md
29790 (save_multiple<mode>): New pattern.
29791 (save_multiple_realign<mode>): Likewise.
29792 (restore_multiple<mode>): Likewise.
29793 (restore_multiple_and_return<mode>): Likewise.
29794 (restore_multiple_leave_return<mode>): Likewise.
29795 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
29797 2017-05-14 Julia Koval <julia.koval@intel.com>
29799 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
29800 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
29801 (__builtin_ia32_xsetbv): New builtins.
29802 * config/i386/i386.c (ix86_expand_special_args_builtin):
29804 (ix86_expand_builtin): Special expand for new intrinsics.
29805 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
29806 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
29807 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
29809 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29811 * cfganal.c (inverted_post_order_compute): Change argument type
29813 * cfganal.h (inverted_post_order_compute): Adjust prototype.
29814 * df-core.c (rest_of_handle_df_initialize): Adjust.
29815 (rest_of_handle_df_finish): Likewise.
29816 (df_analyze_1): Likewise.
29817 (df_analyze): Likewise.
29818 (loop_inverted_post_order_compute): Change argument to be a vec *.
29819 (df_analyze_loop): Adjust.
29820 (df_get_n_blocks): Likewise.
29821 (df_get_postorder): Likewise.
29822 * df.h (struct df_d): Change field to be a vec.
29823 * lcm.c (compute_laterin): Adjust.
29824 (compute_available): Likewise.
29825 * lra-lives.c (lra_create_live_ranges_1): Likewise.
29826 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
29827 * tree-ssa-pre.c (compute_antic): Likewise.
29829 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29831 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
29832 (depth_first_search::depth_first_search): Change structure init
29833 function to this constructor.
29834 (depth_first_search::add_bb): Rename function to this member.
29835 (depth_first_search::execute): Likewise.
29836 (flow_dfs_compute_reverse_finish): Adjust.
29838 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29840 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
29841 (longest_simple_path): Likewise.
29842 * shrink-wrap.c (spread_components): Likewise.
29843 (disqualify_problematic_components): Likewise.
29844 (emit_common_heads_for_components): Likewise.
29845 (emit_common_tails_for_components): Likewise.
29846 (insert_prologue_epilogue_for_components): Likewise.
29848 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29850 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
29853 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29855 * df-core.c (df_set_blocks): Start using auto_bitmap.
29856 (df_compact_blocks): Likewise.
29857 * df-problems.c (df_rd_confluence_n): Likewise.
29858 * df-scan.c (df_insn_rescan_all): Likewise.
29859 (df_process_deferred_rescans): Likewise.
29860 (df_update_entry_block_defs): Likewise.
29861 (df_update_exit_block_uses): Likewise.
29862 (df_entry_block_bitmap_verify): Likewise.
29863 (df_exit_block_bitmap_verify): Likewise.
29864 (df_scan_verify): Likewise.
29865 * lra-constraints.c (lra_constraints): Likewise.
29866 (undo_optional_reloads): Likewise.
29867 (lra_undo_inheritance): Likewise.
29868 * lra-remat.c (calculate_gen_cands): Likewise.
29869 (do_remat): Likewise.
29870 * lra-spills.c (assign_spill_hard_regs): Likewise.
29871 (spill_pseudos): Likewise.
29872 * tree-ssa-pre.c (bitmap_set_and): Likewise.
29873 (bitmap_set_subtract_values): Likewise.
29875 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29877 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
29878 management with auto_bitmap.
29879 (fix_inter_tick): Likewise.
29880 (fix_recovery_deps): Likewise.
29881 * ira.c (add_store_equivs): Likewise.
29882 (find_moveable_pseudos): Likewise.
29883 (split_live_ranges_for_shrink_wrap): Likewise.
29884 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
29885 (rtx_reuse_manager::seen_def_p): Likewise.
29886 (rtx_reuse_manager::set_seen_def): Likewise.
29887 * print-rtl.h (class rtx_reuse_manager): Likewise.
29889 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29891 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
29893 (migrate_btr_def): Likewise.
29894 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
29895 * df-core.c (loop_post_order_compute): Likewise.
29896 (loop_inverted_post_order_compute): Likewise.
29897 * hsa-common.h: Likewise.
29898 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
29899 * init-regs.c (initialize_uninitialized_regs): Likewise.
29900 * ipa-inline.c (resolve_noninline_speculation): Likewise.
29901 (inline_small_functions): Likewise.
29902 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
29903 * ira.c (combine_and_move_insns): Likewise.
29904 (build_insn_chain): Likewise.
29905 * loop-invariant.c (find_invariants): Likewise.
29906 * lower-subreg.c (propagate_pseudo_copies): Likewise.
29907 * predict.c (tree_predict_by_opcode): Likewise.
29908 (predict_paths_leading_to): Likewise.
29909 (predict_paths_leading_to_edge): Likewise.
29910 (estimate_loops_at_level): Likewise.
29911 (estimate_loops): Likewise.
29912 * shrink-wrap.c (try_shrink_wrapping): Likewise.
29913 (spread_components): Likewise.
29914 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
29915 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
29916 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
29917 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
29918 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
29919 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
29920 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
29921 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
29922 (mark_threaded_blocks): Likewise.
29923 (thread_through_all_blocks): Likewise.
29924 * tree-ssa.c (verify_ssa): Likewise.
29925 (execute_update_addresses_taken): Likewise.
29926 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
29928 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29930 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
29932 (post_order_compute): Likewise.
29933 (inverted_post_order_compute): Likewise.
29934 (pre_and_rev_post_order_compute_fn): Likewise.
29936 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29938 * genrecog.c (int_set::int_set): Explicitly construct our
29939 auto_vec base class.
29940 * vec.h (auto_vec::auto_vec): New constructor.
29942 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29944 * bitmap.h (class auto_bitmap): New constructor taking
29945 bitmap_obstack * argument.
29947 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29949 * bitmap.h (class auto_bitmap): Change type of m_bits to
29950 bitmap_head, and adjust ctor / dtor and member operators.
29952 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
29954 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
29955 when returned register mode doesn't match original mode.
29957 2017-05-12 Jeff Law <law@redhat.com>
29958 Jakub Jelinek <jakub@redhat.com>
29960 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
29961 we look for cc setter after the compare-elim changes.
29962 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
29963 within the vector to match what compare-elim now expects.
29964 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
29965 (xorsi3_flags, one_cmplsi2_flags): Likewise.
29967 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
29968 after the compare-elim changes.
29969 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
29970 the vector to match what compare-elim now expects.
29971 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
29972 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
29973 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
29974 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
29975 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
29977 * config/visium/visium.c (single_set_and_flags): Fix where
29978 we look for cc setter after the compare-elim changes.
29979 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
29980 with the vector to match what compare-elim now expects.
29981 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
29982 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
29983 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
29984 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
29985 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
29986 (neg<mode>2_insn_set_overflow): Likewise.
29988 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
29990 PR middle-end/79794
29991 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
29992 maybe_expand_insn call, set ops[0].target. If still set after call,
29993 set alt_rtl. Add extra arg to recursive calls.
29994 (extract_bit_field): Add alt_rtl argument. Pass to
29996 * expmed.h (extract_bit_field): Fix prototype.
29997 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
29998 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
29999 to extract_bit_field_calls.
30000 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
30001 Pass alt_rtl to extract_bit_field calls.
30002 * calls.c (store_unaligned_arguments_into_psuedos)
30003 load_register_parameters): Pass extra NULL to extract_bit_field calls.
30004 * optabs.c (maybe_legitimize_operand): Clear op->target when call
30006 * optabs.h (struct expand_operand): Add target bitfield.
30008 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
30010 * compare-elim.c (try_eliminate_compare): Canonicalize
30011 operation with embedded compare to
30012 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
30013 (set (reg) (operation)].
30015 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
30017 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
30020 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
30021 cost of adding a carry flag for ADC instruction.
30022 [case MINUS]: Ignore the cost of subtracting a carry flag
30023 for SBB instruction.
30025 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
30027 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
30029 * config/rs6000/bmiintrin.h: New file.
30030 * config/rs6000/bmi2intrin.h: New file.
30031 * config/rs6000/x86intrin.h: New file.
30033 2017-05-12 Jeff Law <law@redhat.com>
30035 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
30038 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
30040 PR middle-end/80707
30041 * tree-cfg.c: Remove cfg edges of unreachable case statements.
30043 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30045 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
30046 early expansion of vector divide builtins.
30047 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
30048 builtins identified as having unsigned arguments.
30050 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30052 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
30053 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
30054 expansion of vector logical operations (and, andc, or, xor,
30057 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30059 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
30060 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
30062 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30064 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
30065 early GIMPLE expansion of vector multiplies.
30067 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
30069 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
30070 TARGET_HAVE_MOVT conditional.
30071 (movt splitter): Likewise.
30073 2017-05-12 Richard Biener <rguenther@suse.de>
30075 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
30076 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
30077 Fold all stmts not inplace.
30079 2017-05-12 Richard Biener <rguenther@suse.de>
30081 PR tree-optimization/80713
30082 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
30083 inserted_exprs bit for not removed stmts.
30085 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
30087 PR middle-end/69921
30088 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
30089 parallelized" attribute for parallelized OpenACC kernels.
30090 * omp-offload.c (execute_oacc_device_lower): Use it.
30092 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
30093 Set "oacc kernels" attribute.
30094 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
30095 parameter. Adjust all users.
30096 (oacc_fn_attrib_kernels_p): Remove function.
30097 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
30098 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
30099 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
30100 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
30101 assert "oacc kernels" attribute is set.
30103 2017-05-11 Carl Love <cel@us.ibm.com>
30105 * config/rs6000/rs6000-c: Add support for built-in functions
30106 vector unsigned char vec_popcnt (vector signed char)
30107 vector unsigned char vec_popcnt (vector unsigned char)
30108 vector unsigned short vec_popcnt (vector signed short)
30109 vector unsigned short vec_popcnt (vector unsigned short)
30110 vector unsigned int vec_popcnt (vector signed int)
30111 vector unsigned int vec_popcnt (vector unsigned int)
30112 vector unsigned long long vec_popcnt (vector signed long long)
30113 vector unsigned long long vec_popcnt (vector unsigned long long)
30114 vector signed long long vec_slo (vector signed long long,
30115 vector signed char)
30116 vector signed long long vec_slo (vector signed long long,
30117 vector unsigned char)
30118 vector unsigned long long vec_slo (vector unsigned long long,
30119 vector signed char)
30120 vector unsigned long long vec_slo (vector unsigned long long,
30121 vector unsigned char)
30122 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
30123 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
30124 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
30125 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
30126 * doc/extend.texi: Update the built-in documentation file for the
30127 new built-in functions.
30129 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
30131 * attribs.h (sorted_attr_string): Move machine independent
30132 functions for target clone support from the i386 port to common
30133 code. Rename ix86_function_versions to common_function_versions.
30134 Rename make_name to make_unique_name.
30135 (common_function_versions): Likewise.
30136 (make_unique_name): Likewise.
30137 (make_dispatcher_decl): Likewise.
30138 (is_function_default_version): Likewise.
30139 * attribs.c (attr_strcmp): Likewise.
30140 (sorted_attr_string): Likewise.
30141 (common_function_versions): Likewise.
30142 (make_unique_name): Likewise.
30143 (make_dispatcher_decl): Likewise.
30144 (is_function_default_version): Likewise.
30145 * config/i386/i386.c (attr_strcmp): Likewise.
30146 (sorted_attr_string): Likewise.
30147 (ix86_function_versions): Likewise.
30148 (make_name): Likewise.
30149 (make_dispatcher_decl): Likewise.
30150 (is_function_default_version): Likewise.
30151 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
30153 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30156 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
30157 Account for direct move costs for vec_construct of integer
30160 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
30163 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
30164 (UNSPEC_STX_ATOMIC): Ditto.
30165 (loaddi_via_sse): New insn.
30166 (storedi_via_sse): Ditto.
30167 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
30168 Update corresponding peephole2 patterns.
30169 (atomic_storedi_fpu): Ditto.
30171 2017-05-11 Julia Koval <julia.koval@intel.com>
30173 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
30174 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
30176 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
30177 (__builtin_ia32_rsqrt14ss_mask): New builtins.
30178 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
30180 2017-05-11 Nathan Sidwell <nathan@acm.org>
30182 * graphite-poly.c: Include dumpfile.h.
30184 * dumpfle.h (dump_function): Declare here ...
30185 * tree-dump.h (dump_function): ... not here.
30186 * dumpfile.c: #include tree-cfg.h.
30187 (dump_function): Move here from ...
30188 * tree-dump.c (dump_function): ... here.
30189 * gimplify.c: #include splay-tree.h, not tree-dump.h.
30190 * graphite-poly.c: Don't include tree-dump.h.
30191 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
30192 * print-tree.c: Likewise.
30193 * stor-layout.c: Likewise.
30194 * tree-nested.c: Likewise.
30196 * dumpfile.c (dump_start): Use TDF_FLAGS.
30197 (dump_enable_all): Fix TDF_KIND check thinko.
30199 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
30201 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
30202 array entries to represent two legal parameterizations of the
30203 overloaded __builtin_cmpb function, as represented by the
30204 P6_OV_BUILTIN_CMPB constant.
30205 (altivec_resolve_overloaded_builtin): Add special case handling
30206 for the __builtin_cmpb function, as represented by the
30207 P6_OV_BUILTIN_CMPB constant.
30208 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
30209 (BU_P6_64BIT_2): New macro.
30210 (BU_P6_OVERLOAD_2): New macro
30211 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
30212 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
30213 (CMPB): Add overload support to represent both 32-bit and 64-bit
30214 compare-bytes function.
30215 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
30216 support for TARGET_CMPB.
30217 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
30218 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
30219 documentation of the __builtin_cmpb overloaded built-in function.
30221 2017-05-11 Richard Biener <rguenther@suse.de>
30223 PR tree-optimization/80705
30224 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
30225 bases are not vectorizable.
30227 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30229 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
30230 when counting register pressure.
30232 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30234 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
30235 (iv_ca_more_deps): Renamed to ...
30236 (iv_ca_compare_deps): ... this.
30237 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
30239 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30241 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
30243 (determine_group_iv_costs): ... here.
30244 (find_inv_vars_cb): Record inv var if it's not recorded before.
30246 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30248 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
30249 (get_shiftadd_cost): Ditto.
30251 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30253 * tree-ssa-address.c: Include header file.
30254 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
30256 (add_to_parts): Refactor.
30257 (addr_to_parts): New parameter. Update use of move_hint_to_base.
30258 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
30261 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30263 PR tree-optimization/53090
30264 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
30266 (extract_cond_operands): Detect condition with IV on both sides
30267 and return COMP_IV_EXPR_2.
30268 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
30269 (rewrite_use_compare): Simplify by removing call to function
30270 extract_cond_operands.
30272 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30274 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
30275 (extract_cond_operands): Detect condition comparing against non-
30276 invariant bound and return appropriate enum value.
30277 (find_interesting_uses_cond): Update use of extract_cond_operands.
30278 Handle its return value accordingly.
30279 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
30281 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30283 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
30284 nonlinear iv_use computation in loop invariant sensitive way.
30286 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30288 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
30289 (find_iv_candidates): Call relate_compare_use_with_all_cands.
30291 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30293 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
30294 (dump_cand): Support iv_cand.inv_exprs.
30295 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
30297 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
30300 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30302 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
30304 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
30305 as local function. Include necessary header files.
30306 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
30308 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30310 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
30312 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30314 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
30315 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
30316 RSHIFT_EXPR and BIT_NOT_EXPR.
30318 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30320 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
30321 (adjust_setup_cost): New parameter supporting round up adjustment.
30322 (struct address_cost_data): Delete.
30323 (force_expr_to_var_cost): Don't bound cost with spill_cost.
30324 (split_address_cost, ptr_difference_cost): Delete.
30325 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
30326 (struct ainc_cost_data): New struct.
30327 (get_address_cost_ainc): New function.
30328 (get_address_cost, get_computation_cost): Reimplement.
30329 (determine_group_iv_cost_address): Record inv_expr for all uses of
30331 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
30332 (iv_ca_has_deps): Reimplemented to ...
30333 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
30335 (iv_ca_extend): Call iv_ca_more_deps.
30337 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30339 * tree-ssa-address.c (struct mem_address): Move to header file.
30340 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
30341 * tree-ssa-address.h (struct mem_address): Move from C file.
30342 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
30344 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30346 * tree-affine.h (aff_combination_type): New interface.
30347 (aff_combination_zero_p): Remove static.
30348 (aff_combination_const_p): New interface.
30349 (aff_combination_singleton_var_p): New interfaces.
30351 2017-05-11 Richard Biener <rguenther@suse.de>
30353 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
30354 Skip unreachable blocks and destinations.
30355 (eliminate): Move stmt removal and fixup ...
30356 (fini_eliminate): ... here. Skip inserted exprs.
30357 (pass_pre::execute): Move fini_pre after fini_eliminate.
30358 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
30359 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
30360 PRE to get rid of dead code that has invalid SSA form and
30361 split critical edges again.
30363 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30365 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
30367 2017-05-11 Richard Biener <rguenther@suse.de>
30369 * passes.c (execute_function_todo): Verify loops if they are
30370 said to be up-to-date.
30371 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
30372 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
30374 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
30377 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
30378 handle calling assemble_external ourself.
30381 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
30382 modes with zero size. Enhance comment.
30384 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30386 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
30387 built-ins for vec_xl and vec_xst with short and char pointer
30390 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
30392 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
30393 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
30394 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
30395 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
30396 (_mm_maskz_min_round_ss): New intrinsics.
30397 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
30398 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
30399 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
30400 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
30401 (__builtin_ia32_minss_mask_round): New builtins.
30402 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
30403 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
30404 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
30406 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
30407 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
30409 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
30412 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
30414 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
30415 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
30416 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
30417 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
30418 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
30419 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
30420 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
30421 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
30422 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
30423 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
30424 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
30425 (__builtin_ia32_mulss_mask_round): New builtins.
30426 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
30427 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
30428 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
30430 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
30431 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
30433 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
30436 2017-05-10 Julia Koval <julia.koval@intel.com>
30438 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
30439 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
30440 (_mm256_setr_m128i): New intrinsics.
30442 2017-05-10 Julia Koval <julia.koval@intel.com>
30444 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
30445 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
30446 (_mm_maskz_rcp14_ss): New intrinsics.
30447 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
30448 (__builtin_ia32_rcp14ss_mask): New builtins.
30449 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
30451 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
30453 PR tree-optimization/51513
30454 * tree-cfg.c (gimple_seq_unreachable_p): New function.
30455 (assert_unreachable_fallthru_edge_p): Use it.
30456 (group_case_labels_stmt): Likewise.
30457 * tree-cfg.h: Prototype it.
30458 * stmt.c: Include cfghooks.h and tree-cfg.h.
30459 (emit_case_dispatch_table) <gap_label>: New local variable.
30460 Use it to fill dispatch table gaps.
30461 Test for default_label before updating probabilities.
30462 (expand_case) <default_label>: Remove unneeded initialization.
30463 Test for unreachable default case statement and remove its edge.
30464 Set default_label accordingly.
30465 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
30467 2017-05-10 Carl Love <cel@us.ibm.com>
30469 * config/rs6000/rs6000-c: Add support for built-in functions
30470 vector signed char vec_neg (vector signed char)
30471 vector signed short int vec_neg (vector short int)
30472 vector signed int vec_neg (vector signed int)
30473 vector signed long long vec_neg (vector signed long long)
30474 vector float vec_neg (vector float)
30475 vector double vec_neg (vector double)
30476 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
30478 * config/rs6000/altivec.h: Add define for vec_neg
30479 * doc/extend.texi: Update the built-in documentation for the
30480 new built-in functions.
30482 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30484 PR tree-optimization/77644
30485 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
30487 2017-05-10 Nathan Sidwell <nathan@acm.org>
30489 * dumpfile.h (TDI_lang_all): New.
30490 (TDF_KIND): New. Renumber others
30491 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
30493 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
30495 (get_dump_file_name): Adjust suffix generation.
30496 (dump_enable_all): Use TDF_KIND.
30497 * doc/invoke.texi (-fdump-lang-all): Document.
30499 * dumpfile.h: Tabify.
30501 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
30504 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
30505 Move member access before delete.
30507 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
30509 * tree-inline.c (expand_call_inline): Split block at stmt
30512 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
30515 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
30516 are now unused after splitting mov{sf,sd}_hardfloat.
30517 (f32_lr2): Likewise.
30518 (f32_lm): Likewise.
30519 (f32_lm2): Likewise.
30520 (f32_li): Likewise.
30521 (f32_li2): Likewise.
30522 (f32_lv): Likewise.
30523 (f32_sr): Likewise.
30524 (f32_sr2): Likewise.
30525 (f32_sm): Likewise.
30526 (f32_sm2): Likewise.
30527 (f32_si): Likewise.
30528 (f32_si2): Likewise.
30529 (f32_sv): Likewise.
30530 (f32_dm): Likewise.
30531 (f32_vsx): Likewise.
30532 (f32_av): Likewise.
30533 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
30534 For movsf, order stores so the VSX stores occur before the GPR
30535 store which encourages the register allocator to use a traditional
30536 FPR instead of a GPR. For movsd, order the stores so that the GPR
30537 store comes before the VSX stores to allow the power6 to work.
30538 This is due to the power6 not having a 32-bit integer store
30539 instruction from a FPR.
30540 (movsf_hardfloat): Likewise.
30541 (movsd_hardfloat): Likewise.
30543 2017-05-09 Martin Sebor <msebor@redhat.com>
30545 PR translation/80280
30546 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
30549 PR translation/80280
30550 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
30551 data member added in r247778.
30552 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
30554 2017-05-09 Nathan Sidwell <nathan@acm.org>
30556 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
30558 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
30561 2017-05-09 Marek Polacek <polacek@redhat.com>
30563 * doc/invoke.texi: Fix typo.
30565 2017-05-09 Richard Biener <rguenther@suse.de>
30567 * tree-vrp.c (vrp_val_is_max): Adjust comment.
30568 (vrp_val_is_min): Likewise.
30569 (set_value_range_to_value): Likewise.
30570 (set_value_range_to_nonnegative): Likewise.
30571 (gimple_assign_nonzero_p): Likewise.
30572 (gimple_stmt_nonzero_p): Likewise.
30573 (vrp_int_const_binop): Likewise. Remove unreachable case.
30574 (adjust_range_with_scev): Adjust comments.
30575 (compare_range_with_value): Likewise.
30576 (extract_range_from_phi_node): Likewise.
30577 (test_for_singularity): Likewise.
30579 2017-05-09 Richard Biener <rguenther@suse.de>
30581 * tree-vrp.c (get_single_symbol): Add assert that we don't
30582 get overflowed constants as invariant part.
30583 (compare_values_warnv): Add comment before the TREE_NO_WARNING
30584 checks. Use wi::cmp instead of recursing for integer constants.
30585 (compare_values): Just ignore whether we assumed undefined
30586 overflow instead of failing the compare.
30587 (extract_range_for_var_from_comparison_expr): Add comment before the
30588 TREE_NO_WARNING sets.
30589 (test_for_singularity): Likewise.
30590 (extract_range_from_comparison): Do not disable optimization
30591 when we assumed undefined overflow.
30592 (extract_range_basic): Remove init of unused var.
30594 2017-05-09 Richard Biener <rguenther@suse.de>
30596 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
30597 (extract_range_from_multiplicative_op_1): Adjust.
30598 (extract_range_from_binary_expr_1): Use int_const_binop.
30600 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
30603 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
30604 rs6000_store_data_bypass_p in seven define_bypass directives and
30605 in several comments.
30606 * config/rs6000/rs6000-protos.h: Add prototype for
30607 rs6000_store_data_bypass_p function.
30608 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
30609 function implements slightly different (rs6000-specific) semantics
30610 than store_data_bypass_p, returning false rather than aborting
30611 with assertion error when arguments do not satisfy the
30612 requirements of store data bypass.
30613 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
30614 rs6000_store_data_bypass_p.
30616 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
30618 * config/xtensa/xtensa-protos.h
30619 (xtensa_initial_elimination_offset): New declaration.
30620 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
30621 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
30622 macro definition, add case for FRAME_POINTER_REGNUM when
30623 FRAME_GROWS_DOWNWARD.
30624 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
30625 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
30626 xtensa_initial_elimination_offset.
30628 2017-05-08 Nathan Sidwell <nathan@acm.org>
30630 * doc/invoke.texi: Alphabetize -fdump options.
30632 2017-05-08 Martin Sebor <msebor@redhat.com>
30634 PR translation/80280
30635 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
30637 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
30639 * target.def (compute_frame_layout): New optional target hook.
30640 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
30641 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
30642 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
30644 * reload1.c (verify_initial_elim_offsets): Likewise.
30645 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
30646 (use_simple_return_p): Call arm_compute_frame_layout if needed.
30647 (arm_get_frame_offsets): Split up into this ...
30648 (arm_compute_frame_layout): ... and this function.
30650 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
30652 * config/aarch64/constraints.md (Usa): New constraint.
30653 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
30655 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
30657 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
30658 with_multilib_list after it has been checked.
30660 2017-05-08 Richard Biener <rguenther@suse.de>
30662 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
30663 (bitmap_set_subtract_values): Likewise.
30665 2017-05-08 Richard Biener <rguenther@suse.de>
30667 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
30668 (gimple_assign_nonzero): ... this and remove strict_overflow_p
30670 (gimple_stmt_nonzero_warnv_p): Rename to ...
30671 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
30673 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
30674 (extract_range_basic): Adjust, do not disable propagation on
30675 strict overflow sensitive simplification.
30676 (vrp_visit_cond_stmt): Likewise.
30678 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
30680 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
30681 body size unconditionally.
30683 2017-05-07 Jeff Law <law@redhat.com>
30686 2017-05-06 Jeff Law <law@redhat.com>
30687 PR tree-optimization/78496
30688 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
30691 PR tree-optimization/78496
30692 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
30693 (simplify_stmt_using_ranges): Call it.
30694 (vrp_dom_walker::before_dom_children): Extract equivalences
30695 from an ASSERT_EXPR with an equality comparison against a
30698 2017-05-06 Jeff Law <law@redhat.com>
30700 PR tree-optimization/78496
30701 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
30704 PR tree-optimization/78496
30705 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
30706 (simplify_stmt_using_ranges): Call it.
30707 (vrp_dom_walker::before_dom_children): Extract equivalences
30708 from an ASSERT_EXPR with an equality comparison against a
30711 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
30713 * lra-constraints.c (lra_copy_reg_equiv): New function.
30714 (split_reg): Use it to copy equivalence information from the
30715 original register to the spill register.
30717 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
30719 PR rtl-optimization/75964
30720 * simplify-rtx.c (simplify_const_relational_operation): Remove
30721 invalid handling of comparisons of integer ABS.
30723 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
30725 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
30726 initialize to zero.
30727 (init_regs): Remove declaration.
30728 (function_arg_advance_32): Initialize error_p as boolean variable.
30730 2017-05-05 Nathan Sidwell <nathan@acm.org>
30732 * store-motion.c (remove_reachable_equiv_notes): Reformat long
30733 lines. Use for (;;).
30735 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30737 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
30738 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
30739 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
30740 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
30741 VF=2 that require versioning.
30743 2017-05-05 David Malcolm <dmalcolm@redhat.com>
30745 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
30748 2017-05-05 David Malcolm <dmalcolm@redhat.com>
30750 * diagnostic.h (diagnostic_override_option_index): Convert from
30751 macro to inline function.
30753 2017-05-05 David Malcolm <dmalcolm@redhat.com>
30755 * diagnostic.c (last_module_changed_p): New function.
30756 (set_last_module): New function.
30757 (diagnostic_report_current_module): Convert macro usage to
30758 the above functions.
30759 * diagnostic.h (diagnostic_context::last_module): Strengthen
30760 from const line_map * to const line_map_ordinary *.
30761 (diagnostic_last_module_changed): Delete macro.
30762 (diagnostic_set_last_module): Delete macro.
30764 2017-05-05 David Malcolm <dmalcolm@redhat.com>
30766 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
30767 with diagnostic_report_diagnostic.
30768 (diagnostic_n_impl_richloc): Likewise.
30769 * diagnostic.h (report_diagnostic): Delete macro.
30770 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
30771 with diagnostic_report_diagnostic.
30772 * substring-locations.c (format_warning_va): Likewise.
30774 2017-05-05 David Malcolm <dmalcolm@redhat.com>
30776 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
30777 save/restor of format_spec. Move option-printing code to...
30778 (print_option_information): ...this new function, and
30779 reimplement by simply printing to the pretty_printer,
30780 rather than appending to the format string.
30782 2017-05-05 David Malcolm <dmalcolm@redhat.com>
30784 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
30785 handling logic into...
30786 (update_effective_level_from_pragmas): ...this new function.
30788 2017-05-04 Andrew Waterman <andrew@sifive.com>
30790 * config/riscv/riscv.opt (mstrict-align): New option.
30791 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
30792 (SLOW_UNALIGNED_ACCESS): Define.
30793 (riscv_slow_unaligned_access): Declare.
30794 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
30796 (riscv_slow_unaligned_access): New variable.
30797 (rocket_tune_info): Set slow_unaligned_access to true.
30798 (optimize_size_tune_info): Set slow_unaligned_access to false.
30799 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
30800 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
30801 (riscv_option_override): Set riscv_slow_unaligned_access.
30802 * doc/invoke.texi: Add -mstrict-align to RISC-V.
30804 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
30806 * config/riscv/riscv.md: Unify indentation.
30808 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
30813 * config/rs6000/rs6000.md (u code attribute): Add FIX and
30815 (extendsi<mode>2): Add support for doing sign extension via
30816 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
30817 don't have ISA 3.0 instructions.
30818 (extendsi<mode>2 splitter): Likewise.
30819 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
30820 generate the normal insns since SImode can now go in vector
30821 registers. Disallow the special UNSPECs needed for previous
30822 machines to hide SImode being used. Add new insns
30823 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
30824 (fix_trunc<mode>si2_stfiwx): Likewise.
30825 (fix_trunc<mode>si2_internal): Likewise.
30826 (fixuns_trunc<mode>si2): Likewise.
30827 (fixuns_trunc<mode>si2_stfiwx): Likewise.
30828 (fctiw<u>z_<mode>_smallint): Likewise.
30829 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
30830 of floating point to 32-bit integer from doing a direct move to
30831 the GPR registers to do a store.
30832 (fctiwz_<mode>): Break long line.
30834 2017-05-05 Bin Cheng <bin.cheng@arm.com>
30836 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
30837 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
30838 (addr_list, addr_offset_valid_p): New.
30839 (split_address_groups): Check offset validity with above function.
30840 (gt-tree-ssa-loop-ivopts.h): Include header file.
30842 2017-05-05 Nathan Sidwell <nathan@acm.org>
30844 * config.gcc (arm*-*-*): Add missing 'fi'.
30846 2017-05-05 Steve Ellcey <sellcey@cavium.com>
30848 * doc/invoke.texi (-fopt-info): Explicitly say order of options
30849 included in -fopt-info does not matter.
30850 * doc/optinfo.texi (-fopt-info): Fix description of default
30851 behavour. Explicitly say order of options included in -fopt-info
30854 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
30856 * config.gcc: Allow combinations of aprofile and rmprofile values for
30857 --with-multilib-list.
30858 * config/arm/t-multilib: New file.
30859 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
30860 variables. Remove setting of ISA and floating-point ABI in
30861 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
30862 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
30863 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
30864 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
30866 * config/arm/t-rmprofile: Likewise except for the matches changes.
30867 * doc/install.texi (--with-multilib-list): Document the combination of
30868 aprofile and rmprofile values and warn about pitfalls in doing that.
30870 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
30872 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
30873 (movdi_aarch64): Likewise.
30875 2017-05-05 Jakub Jelinek <jakub@redhat.com>
30877 PR tree-optimization/80632
30878 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
30880 (build_arrays): Initialize it for virtual phis.
30881 (fix_phi_nodes): Use it for virtual phis.
30883 PR tree-optimization/80558
30884 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
30885 [x, y] op z into [x op, y op z] for op & or | if conditions
30888 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
30889 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
30892 * config/arm/arm.md (use_literal_pool): Remove.
30893 (64-bit immediate split): No longer takes cost into consideration
30894 if arm_disable_literal_pool is enabled.
30895 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
30896 used when arm_disable_literal_pool is enabled.
30897 (arm_max_const_double_inline_cost): Remove use of
30898 arm_disable_literal_pool.
30899 (push_minipool_fix): Add assert.
30900 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
30901 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
30902 (no_literal_pool_sf_immediate): New.
30904 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30906 PR tree-optimization/80613
30907 * tree-ssa-dce.c (propagate_necessity): Remove cases for
30908 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
30910 2017-05-05 Richard Biener <rguenther@suse.de>
30912 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
30914 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
30916 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
30917 of this flag from insn conditions due to removal from r247495.
30919 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
30921 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
30923 (arm_early_store_addr_dep_ptr): Likewise.
30924 * config/arm/aarch-common-protos.h
30925 (arm_early_load_addr_dep_ptr): Add prototype.
30926 (arm_early_store_addr_dep_ptr): Likewise.
30927 * config/arm/cortex-a53.md: Add new bypasses.
30929 2017-05-05 Jakub Jelinek <jakub@redhat.com>
30931 * tree.c (next_type_uid): Change type to unsigned.
30932 (type_hash_canon): Decrement back next_type_uid if
30933 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
30934 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
30937 2017-05-04 Martin Sebor <msebor@redhat.com>
30939 * builtins.c: Fix a trivial typo in a comment.
30941 PR middle-end/79234
30942 * builtins.c (check_sizes): Adjust to handle reading past the end.
30943 Avoid printing excessive upper bound of ranges. Use %E to print
30944 tree nodes instead of converting them to %wu.
30945 (expand_builtin_memchr): New function.
30946 (compute_dest_size): Rename...
30947 (compute_objsize): ...to this.
30948 (expand_builtin_memcpy): Adjust.
30949 (expand_builtin_mempcpy): Adjust.
30950 (expand_builtin_strcat): Adjust.
30951 (expand_builtin_strcpy): Adjust.
30952 (check_strncat_sizes): Adjust.
30953 (expand_builtin_strncat): Adjust.
30954 (expand_builtin_strncpy): Adjust and simplify.
30955 (expand_builtin_memset): Adjust.
30956 (expand_builtin_bzero): Adjust.
30957 (expand_builtin_memcmp): Adjust.
30958 (expand_builtin): Handle memcmp.
30959 (maybe_emit_chk_warning): Check strncat just once.
30961 2017-05-04 Martin Sebor <msebor@redhat.com>
30963 PR preprocessor/79214
30964 PR middle-end/79222
30965 PR middle-end/79223
30966 * builtins.c (check_sizes): Add inlining context and issue
30967 warnings even when -Wno-system-headers is set.
30968 (check_strncat_sizes): Same.
30969 (expand_builtin_strncat): Same.
30970 (expand_builtin_memmove): New function.
30971 (expand_builtin_stpncpy): Same.
30972 (expand_builtin): Handle memmove and stpncpy.
30974 2017-05-04 Bin Cheng <bin.cheng@arm.com>
30976 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
30977 which is not used any more.
30979 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
30981 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
30983 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
30985 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
30986 (cortexa53_tunings): Likewise.
30987 (cortexa57_tunings): Likewise.
30988 (cortexa72_tunings): Likewise.
30989 (cortexa73_tunings): Likewise.
30991 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
30993 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
30994 Set loop alignment to 8.
30996 2017-05-04 Martin Sebor <msebor@redhat.com>
30998 PR translation/80280
30999 * builtins.c (expand_builtin_object_size): Add missing quoting to
31000 %D and like directives.
31001 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
31002 (hsa_type_for_tree_type): Same.
31003 (verify_function_arguments): Same.
31004 * symtab.c (symbol_table::change_decl_assembler_name): Same.
31005 * varasm.c (get_section): Same.
31008 2017-05-04 Martin Sebor <msebor@redhat.com>
31010 PR translation/80280
31011 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
31013 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31015 * config/aarch64/aarch64.c (generic_addrcost_table):
31016 Change HI/TI mode setting.
31018 2017-05-04 Martin Jambor <mjambor@suse.cz>
31020 PR tree-optimization/80622
31021 * tree-sra.c (comes_initialized_p): New function.
31022 (build_accesses_from_assign): Only set write lazily when
31023 comes_initialized_p is false.
31024 (analyze_access_subtree): Use comes_initialized_p.
31025 (propagate_subaccesses_across_link): Assert !comes_initialized_p
31026 instead of testing for PARM_DECL.
31028 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31030 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
31031 constraint on operand 0 to allow more general addressing modes.
31032 Adjust output template.
31033 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
31035 * config/aarch64/aarch64-protos.h
31036 (aarch64_address_valid_for_prefetch_p): Declare prototype.
31037 * config/aarch64/constraints.md (Dp): New address constraint.
31038 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
31041 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
31043 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
31044 update use of estimate_ipcp_clone_size_and_time.
31045 (estimate_local_effects): Update use of
31046 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
31047 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
31048 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
31049 Return nonspecialized time.
31051 2017-05-04 Richard Biener <rguenther@suse.de>
31053 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
31054 for the last VUSE which def dominates the PHI. Directly call
31056 (get_continuation_for_phi_1): Remove.
31058 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
31060 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
31061 to explain the use of truncating division. Cap the number of
31062 iterations to the maximum given by nb_iterations_upper_bound,
31065 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
31067 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
31068 * configure: Regenerate.
31069 * config.in: Regenerate.
31070 * config/i386/driver-mingw32.c: new file.
31071 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
31072 * config.host: Link driver-mingw32.o on MinGW host.
31073 * doc/install.texi: Document new --enable-mingw-wildcard configure
31076 2017-05-04 Marek Polacek <polacek@redhat.com>
31078 PR tree-optimization/80612
31079 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
31081 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31082 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
31084 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
31085 (movt splitter): Likewise.
31086 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
31087 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
31088 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
31089 block for Thumb-1 with MOVT.
31090 (thumb2_legitimate_address_p): Move code block ...
31091 (can_avoid_literal_pool_for_label_p): ... into this new function.
31092 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
31094 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
31095 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
31096 "M-profile targets with the MOVT instruction".
31098 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31100 * config/arm/arm-builtins.c (arm_init_builtins): Rename
31101 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
31102 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
31104 2017-05-04 Martin Liska <mliska@suse.cz>
31106 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
31107 variable cond_code.
31109 2017-05-04 Richard Biener <rguenther@suse.de>
31111 * tree.c (array_at_struct_end_p): Handle arrays at struct
31112 end with flexarrays more conservatively. Refactor and treat
31113 arrays of arrays or aggregates more strict. Fix
31114 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
31115 * tree.c (array_at_struct_end_p): Adjust prototype.
31116 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
31117 * gimple-fold.c (get_range_strlen): Likewise.
31118 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
31120 2017-05-04 Richard Biener <rguenther@suse.de>
31122 PR tree-optimization/31130
31123 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
31125 (supports_overflow_infinity): Likewise.
31126 (is_negative_overflow_infinity): Likewise.
31127 (is_positive_overflow_infinity): Likewise.
31128 (is_overflow_infinity): Likewise.
31129 (stmt_overflow_infinity): Likewise.
31130 (overflow_infinity_range_p): Likewise.
31131 (usable_range_p): Remove as always returning true.
31132 (make_overflow_infinity): Remove.
31133 (negative_overflow_infinity): Likewise.
31134 (positive_overflow_infinity): Likewise.
31135 (avoid_overflow_infinity): Likewise.
31136 (set_value_range): Adjust accordingly.
31137 (set_value_range_to_nonnegative): Likewise, remove now unused
31138 overflow_infinity arg.
31139 (vrp_operand_equal_p): Adjust.
31140 (update_value_range): Likewise.
31141 (range_int_cst_singleton_p): Likewise.
31142 (operand_less_p): Likewise.
31143 (compare_values_warnv): Likewise.
31144 (extract_range_for_var_from_comparison_expr): Likewise.
31145 (vrp_int_const_binop): Likewise.
31146 (zero_nonzero_bits_from_vr): Likewise.
31147 (extract_range_from_multiplicative_op_1): Likewise.
31148 (extract_range_from_binary_expr_1): Likewise.
31149 (extract_range_from_unary_expr): Likewise.
31150 (extract_range_from_comparison): Likewise.
31151 (extract_range_basic): Likewise.
31152 (adjust_range_with_scev): Likewise.
31153 (compare_ranges): Likewise.
31154 (compare_range_with_value): Likewise.
31155 (dump_value_range): Likewise.
31156 (test_for_singularity): Likewise, remove strict_overflow_p parameter
31158 (simplify_cond_using_ranges): Adjust.
31160 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
31162 * brig-builtins.def: Added a builtin for class_f64.
31163 * builtin-types.def: Added a builtin type needed by class_f64.
31165 2017-05-03 Jason Merrill <jason@redhat.com>
31167 * timevar.def: Add TV_CONSTEXPR.
31169 2017-05-03 David Malcolm <dmalcolm@redhat.com>
31171 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
31173 2017-05-03 Martin Jambor <mjambor@suse.cz>
31175 * ipa-prop.c (ipa_update_after_lto_read): Removed.
31176 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
31177 * ipa-cp.c (ipcp_propagate_stage): Do not call
31178 ipa_update_after_lto_read.
31179 * ipa-inline.c (ipa_inline): Likewise.
31181 2017-05-03 Martin Jambor <mjambor@suse.cz>
31183 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
31184 tag. Added a default constructor and a destructor.
31185 (ipa_edge_args_sum_t): New class;
31186 (ipa_edge_args_sum): Declare.
31187 (ipa_edge_args_vector): Remove declaration.
31188 (IPA_EDGE_REF): Use ipa_edge_args_sum.
31189 (ipa_free_edge_args_substructures): Remove declaration.
31190 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
31191 (ipa_edge_args_info_available_for_edge_p): Likewise.
31192 * ipa-prop.c (ipa_edge_args_vector): Removed.
31193 (edge_removal_hook_holder): Likewise.
31194 (edge_duplication_hook_holder): Likewise.
31195 (ipa_edge_args_sum): New variable.
31196 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
31197 ipa_edge_args_vector.
31198 (ipa_free_edge_args_substructures): Likewise.
31199 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
31200 ipa_edge_args_vector.
31201 (ipa_edge_removal_hook): Turned into method
31202 ipa_edge_args_sum_t::remove.
31203 (ipa_edge_duplication_hook): Turned into method
31204 ipa_edge_args_sum_t::duplicate.
31205 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
31206 registering edge hooks.
31207 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
31208 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
31209 ipa_edge_args_sum instead of ipa_edge_args_vector.
31210 * ipa-profile.c (ipa_profile): Likewise.
31212 2017-05-03 Martin Jambor <mjambor@suse.cz>
31214 * symbol-summary.h (function_summary): New method exists.
31215 (function_summary::symtab_removal): Deallocate through release.
31216 (call_summary): New class.
31217 (gt_ggc_mx): New overload.
31218 (gt_pch_nx): Likewise.
31219 (gt_pch_nx): Likewise.
31221 2017-05-03 Jeff Law <law@redhat.com>
31223 PR tree-optimization/78496
31224 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
31225 from simplify_cond_using_ranges. Split off code to walk
31226 backwards through casts into ...
31227 (simplify_cond_using_ranges_2): New function.
31228 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
31229 (execute_vrp): After identifying jump threads, call
31230 simplify_cond_using_ranges_2.
31232 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
31235 * ipa-inline.h (inline_summary): Add ctor.
31236 (create_ggc): Do not use ggc_cleared_alloc.
31238 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
31239 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
31241 * gcc.c (handle_braces): Support escaping in switch matching
31243 * doc/invoke.texi (Spec Files): Document it.
31244 Remove superfluous @code markup in items.
31246 2017-05-03 David Malcolm <dmalcolm@redhat.com>
31248 * diagnostic-show-locus.c (struct column_range): New struct.
31249 (get_affected_columns): New function.
31250 (get_printed_columns): New function.
31251 (struct correction): New struct.
31252 (correction::ensure_capacity): New function.
31253 (correction::ensure_terminated): New function.
31254 (struct line_corrections): New struct.
31255 (line_corrections::~line_corrections): New dtor.
31256 (line_corrections::add_hint): New function.
31257 (layout::print_trailing_fixits): Reimplement in terms of the new
31259 (selftest::test_overlapped_fixit_printing): New function.
31260 (selftest::diagnostic_show_locus_c_tests): Call it.
31262 2017-05-03 Nathan Sidwell <nathan@acm.org>
31264 Canonicalize canonical type hashing
31265 * tree.h (type_hash_canon_hash): Declare.
31266 * tree.c (type_hash_list, attribute_hash_list): Move into
31267 type_hash_canon_hash.
31268 (build_type_attribute_qual_variant): Break out hash code calc into
31269 type_hash_canon_hash.
31270 (type_hash_canon_hash): New. Generic type hash computation.
31271 (build_range_type_1, build_array_type_1, build_function_type,
31272 build_method_type_directly, build_offset_type, build_complex_type,
31273 make_vector_type): Call it.
31275 2017-05-03 Richard Biener <rguenther@suse.de>
31277 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
31278 When all DRs have unknown misaligned do not always peel
31279 when there is a store but apply the same costing model as if
31280 there were only loads.
31282 2017-05-03 Richard Biener <rguenther@suse.de>
31285 PR tree-optimization/80492
31286 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
31287 compare_base_decls returning dont-know properly.
31289 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
31291 * config/arm/iterators.md (CCSI): New mode iterator.
31292 (arch): New mode attribute.
31293 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
31294 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
31295 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
31296 code iterator for success result mode.
31297 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
31298 the corresponding new insn generators.
31300 2017-05-03 Bin Cheng <bin.cheng@arm.com>
31303 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31304 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
31306 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
31308 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
31309 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
31310 (DDR_A): Wrap DDR argument in brackets.
31311 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
31312 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
31313 (DDR_REVERSED_P): Likewise.
31315 2017-05-03 Jakub Jelinek <jakub@redhat.com>
31317 PR tree-optimization/79472
31318 * tree-switch-conversion.c (struct switch_conv_info): Add
31319 contiguous_range and default_case_nonstandard fields.
31320 (collect_switch_conv_info): Compute contiguous_range and
31321 default_case_nonstandard fields, don't clear final_bb if
31322 contiguous_range and only the default case doesn't have the required
31324 (check_all_empty_except_final): Set default_case_nonstandard instead
31325 of failing if contiguous_range and the default case doesn't have empty
31327 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
31328 and only the default case doesn't have the required constants. Skip
31330 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
31331 if default_case_nonstandard.
31332 (build_constructors): Build constant 1 just once. Assert that default
31333 values aren't inserted in between cases if contiguous_range. Skip
31335 (build_arrays): Skip virtual phis.
31336 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
31337 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
31338 Handle virtual phis.
31339 (gen_inbound_check): Handle default_case_nonstandard case.
31340 (process_switch): Adjust check_final_bb caller. Call
31341 gather_default_values with the first non-default case instead of
31342 default case if default_case_nonstandard.
31344 2017-05-02 Nathan Sidwell <nathan@acm.org>
31346 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
31347 check. Fix formatting.
31349 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
31351 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
31352 errors when comparing specialized and unspecialized times.
31354 2017-05-02 David Malcolm <dmalcolm@redhat.com>
31356 * diagnostic-show-locus.c
31357 (layout::should_print_annotation_line_p): Make private.
31358 (layout::print_annotation_line): Make private.
31359 (layout::annotation_line_showed_range_p): Make private.
31360 (layout::show_ruler): Make private.
31361 (layout::print_source_line): Make private. Pass in line and
31362 line_width, rather than calling location_get_source_line. Drop
31364 (layout::print_leading_fixits): New method.
31365 (layout::print_any_fixits): Rename to...
31366 (layout::print_trailing_fixits): ...this, and make private.
31367 Don't print newline fixits.
31368 (diagnostic_show_locus): Move logic for printing one row into...
31369 (layout::print_line): ...this new function. Move the
31370 location_get_source_line call and error-handling from
31371 print_source_line to here. Call print_leading_fixits, and rename
31372 print_any_fixits to print_trailing_fixits.
31373 (selftest::test_fixit_insert_containing_newline): Update now that
31374 newlines are partially supported.
31375 (selftest::test_fixit_insert_containing_newline_2): New test.
31376 (selftest::test_fixit_replace_containing_newline): Update comments.
31377 (selftest::diagnostic_show_locus_c_tests): Call the new test.
31378 * edit-context.c (class added_line): New class.
31379 (class edited_line): Describe newline handling in comment.
31380 (edited_line::actually_edited_p): New method.
31381 (edited_line::print_content): Delete redundant decl.
31382 (edited_line::m_predecessors): New field.
31383 (edited_file::print_content): Call edited_line::print_content.
31384 (edited_file::print_diff): Update to support newlines.
31385 (edited_file::print_diff_hunk): Likewise.
31386 (edited_file::print_run_of_changed_lines): New function.
31387 (edited_file::print_diff_line): Convert to...
31388 (print_diff_line): ...this.
31389 (edited_file::get_effective_line_count): New function.
31390 (edited_line::edited_line): Initialize new field m_predecessors.
31391 (edited_line::~edited_line): Clean up m_predecessors.
31392 (edited_line::apply_fixit): Handle newlines.
31393 (edited_line::get_effective_line_count): New function.
31394 (edited_line::print_content): New function.
31395 (edited_line::print_diff_lines): New function.
31396 (selftest::test_applying_fixits_insert_containing_newline): New
31398 (selftest::test_applying_fixits_replace_containing_newline): New
31400 (selftest::insert_line): New function.
31401 (selftest::test_applying_fixits_multiple_lines): Add example of
31403 (selftest::edit_context_c_tests): Call the new tests.
31405 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31407 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
31408 parameter cand. Update dump information.
31409 (get_computation_cost): Update uses.
31411 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31413 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
31414 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
31415 (get_computation_at, rewrite_use_address): Update use of
31416 get_computation_aff.
31418 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31420 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
31421 (get_computation): Delete.
31422 (get_computation_cost): Implement like get_computation_cost_at.
31423 Use get_computation_at.
31424 (get_computation_cost_at): Delete.
31425 (rewrite_use_nonlinear_expr): Use get_computation_at.
31426 (rewrite_use_compare, remove_unused_ivs): Ditto.
31428 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31430 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
31432 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31434 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
31435 (ivopts_global_cost_for_size): Rename parameter and update uses.
31436 (iv_ca_recount_cost): Update uses.
31437 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
31438 candidates seperately in n_invs and n_cands.
31439 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
31441 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31443 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
31444 (find_inv_vars_cb): New.
31445 (find_depends): Renamed to ...
31446 (find_inv_vars): ... this.
31447 (add_candidate_1, force_var_cost): Call find_inv_vars.
31448 (split_address_cost, determine_group_iv_cost_cond): Ditto.
31450 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31452 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
31453 inv_vars. Add inv_exprs.
31454 (struct iv_cand): Rename depends_on to inv_vars.
31455 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
31456 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
31457 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
31458 (dump_cand): Dump inv_vars.
31459 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
31460 (record_invariant, find_depends, add_candidate_1): Ditto.
31461 (set_group_iv_cost, force_var_cost): Ditto.
31462 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
31463 (get_computation_cost_at, get_computation_cost): Ditto.
31464 (determine_group_iv_cost_generic): Ditto.
31465 (determine_group_iv_cost_address): Ditto.
31466 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
31467 (determine_group_iv_costs): Ditto.
31468 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
31469 (iv_ca_set_remove_invariants): Renamed to ...
31470 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
31471 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
31472 (iv_ca_set_add_invariants): Renamed to ...
31473 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
31474 (iv_ca_set_cp): Use iv_ca_set_add_invs.
31475 (iv_ca_has_deps): Support inv_vars and inv_exprs.
31476 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
31477 (create_new_ivs): Remove useless dump.
31479 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31481 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
31483 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
31484 (iv_ca_set_no_cp, create_new_iv): Ditto.
31486 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31488 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
31490 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
31492 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
31493 function tree_check2.
31495 2017-05-02 Martin Liska <mliska@suse.cz>
31497 * doc/gcov.texi: Add missing preposition.
31498 * gcov.c (function_info::function_info): Properly fill up
31499 all member variables.
31501 2017-05-02 Tamar Christina <tamar.christina@arm.com>
31503 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
31505 2017-05-02 Tamar Christina <tamar.christina@arm.com>
31507 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
31509 2017-05-02 Martin Liska <mliska@suse.cz>
31512 * lto-streamer-in.c (lto_read_tree_1): Remove
31513 LTO_STREAMER_DEBUG.
31514 * lto-streamer.c (struct tree_hash_entry): Likewise.
31515 (struct tree_entry_hasher): Likewise.
31516 (tree_entry_hasher::hash): Likewise.
31517 (tree_entry_hasher::equal): Likewise.
31518 (lto_streamer_init): Likewise.
31519 (lto_orig_address_map): Likewise.
31520 (lto_orig_address_get): Likewise.
31521 (lto_orig_address_remove): Likewise.
31522 * lto-streamer.h: Likewise.
31523 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
31524 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
31526 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
31528 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
31529 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
31530 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
31531 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
31532 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
31533 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
31534 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
31535 (mm_maskz_sub_ss): New intrinsics.
31536 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31537 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
31538 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
31539 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
31540 (__builtin_ia32_subss_mask_round): New builtins.
31541 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31542 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
31543 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
31545 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
31546 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
31548 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
31551 2017-05-02 Martin Jambor <mjambor@suse.cz>
31553 PR tree-optimization/78687
31554 * tree-sra.c (access): New field parent.
31555 (process_subtree_disqualification): New function.
31556 (disqualify_candidate): Call it.
31557 (build_accesses_from_assign): Reset write flag if creating an
31559 (build_access_subtree): Fill in parent field and also prpagate
31560 down grp_write flag.
31561 (create_artificial_child_access): New parameter set_grp_write, set
31562 grp_write to its value.
31563 (propagate_subaccesses_across_link): Also propagate grp_write flag
31565 (propagate_all_subaccesses): Push the closest parent back to work
31566 queue if add_access_to_work_queue returned true.
31568 2017-05-02 Richard Biener <rguenther@suse.de>
31570 * common.opt (fstrict-overflow): Alias negative to fwrapv.
31571 * doc/invoke.texi (fstrict-overflow): Remove all traces of
31572 -fstrict-overflow documentation.
31573 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
31574 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
31575 flag_strict_overflow.
31576 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
31577 * lto-opts.c (lto_write_options): Do not stream it.
31578 * lto-wrapper.c (merge_and_complain): Do not handle it.
31579 * opts.c (default_options_table): Do not set -fstrict-overflow.
31580 (finish_options): Likewise do not clear it when sanitizing.
31581 * simplify-rtx.c (simplify_const_relational_operation): Do not
31582 test flag_strict_overflow.
31584 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
31586 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
31587 using enabled attribute.
31588 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
31589 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
31590 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
31591 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
31592 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
31593 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
31594 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
31595 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
31596 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
31597 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
31599 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
31601 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
31603 2017-05-02 Richard Biener <rguenther@suse.de>
31605 PR tree-optimization/80591
31607 2017-04-10 Richard Biener <rguenther@suse.de>
31609 * tree-ssa-structalias.c (find_func_aliases): Properly handle
31612 2017-05-02 Richard Biener <rguenther@suse.de>
31614 PR tree-optimization/80549
31615 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
31616 (cleanup_tree_cfg_noloop): Create forwarders to known loop
31617 headers if they do not have a preheader.
31619 2017-05-02 Martin Liska <mliska@suse.cz>
31622 * common.opt: Fix typo.
31623 * doc/invoke.texi: Likewise.
31625 2017-05-01 Jan Beulich <jbeulich@suse.com>
31627 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
31628 swapping, add (x,x,m,x,n) alternative.
31630 2017-05-01 Nathan Sidwell <nathan@acm.org>
31632 * calls.c (combine_pending_stack_adjustment_and_call): Remove
31633 unnecessary unadjusted_alignment check.
31635 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
31638 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
31640 * gimplify.c (gimplify_cilk_detach): New function.
31641 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
31642 * tree-core.h: Document EXPR_CILK_SPAWN.
31643 * tree.h (EXPR_CILK_SPAWN): Define.
31645 2017-05-01 David Malcolm <dmalcolm@redhat.com>
31647 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
31648 to use new fixit_hint representation, using the "replace" logic.
31649 (get_line_span_for_fixit_hint): Likewise.
31650 (layout::print_any_fixits): Likewise.
31651 (selftest::test_one_liner_many_fixits): Rename to...
31652 (selftest::test_one_liner_many_fixits_1): ...this, and update
31653 comment and expected output to reflect that the multiple fix-it
31654 hints are now consolidated into one insertion.
31655 (selftest::test_one_liner_many_fixits_2): New test.
31656 (selftest::test_diagnostic_show_locus_one_liner): Update for
31658 (selftest::test_fixit_consolidation): Update for fix-it API
31660 * diagnostic.c (print_parseable_fixits): Likewise.
31661 * edit-context.c (edited_line::m_line_events): Convert from
31662 auto_vec <line_event *> to auto_vec <line_event>.
31663 (class line_event): Convert from abstract base class to a concrete
31664 class, taking over the role of replace_event.
31665 (class insert_event): Delete.
31666 (class replace_event): Rename to class line_event. Convert to
31668 (edit_context::add_fixits): Reimplement.
31669 (edit_context::apply_insert): Delete.
31670 (edit_context::apply_replace): Rename to...
31671 (edit_context::apply_fixit): ...this. Convert to half-open range.
31672 (edited_file::apply_insert): Delete.
31673 (edited_file::apply_replace): Rename to...
31674 (edited_file::apply_fixit): ...this.
31675 (edited_line::~edited_line): Drop deletion of events.
31676 (edited_line::apply_insert): Delete.
31677 (edited_line::apply_replace): Rename to...
31678 (edited_line::apply_fixit): ...this. Convert to half-open range.
31679 Update for change to type of m_line_events.
31680 * edit-context.h (edit_context::apply_insert): Delete.
31681 (edit_context::apply_replace): Rename to...
31682 (edit_context::apply_fixit): ...this.
31684 2017-05-01 Martin Sebor <msebor@redhat.com>
31686 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
31689 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
31692 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
31693 __get_cpuid_max returns 0.
31694 (__get_cpuid_count): Ditto.
31696 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
31698 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
31699 replacement expression is another instance of one of its arguments.
31701 2017-05-01 Jakub Jelinek <jakub@redhat.com>
31704 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
31705 check for stack push/pop autoinc.
31706 * config/i386/i386.c (ix86_agi_dependent): Return false
31707 if the only reason why modified_in_p returned true is that
31708 addr is SP based and set_insn is a push or pop.
31710 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
31712 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
31715 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
31718 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
31719 (account_size_time): Use two predicates - exec_pred and
31721 (evaluate_conditions_for_known_args): Compute both clause and
31723 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
31724 (inline_summary_t::duplicate): Update.
31725 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
31727 (compute_inline_parameters): Likewise.
31728 (estimate_edge_size_and_time): Update caluclation of time.
31729 (estimate_node_size_and_time): Compute both time and nonspecialized
31731 (estimate_ipcp_clone_size_and_time): Update.
31732 (inline_merge_summary): Update.
31733 (do_estimate_edge_time): Update.
31734 (do_estimate_edge_size): Update.
31735 (do_estimate_edge_hints): Update.
31736 (inline_read_section, inline_write_summary): Stream both new predicates.
31737 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
31739 (compute_inlined_call_time): Cleanup.
31740 (big_speedup_p): Update.
31741 (edge_badness): Update.
31742 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
31743 (size_time_entry): Replace predicate by exec_predicate and
31744 nonconst_predicate.
31745 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
31746 (estimate_edge_time): Return also nonspec_time.
31747 (reset_edge_growth_cache): Update.
31749 2017-04-29 Jakub Jelinek <jakub@redhat.com>
31751 PR rtl-optimization/80491
31752 * ifcvt.c (noce_process_if_block): When looking for x setter
31753 with missing else_bb, don't check only the insn right before
31754 cond_earliest, but look for the last insn that x is modified in
31755 within the same bb.
31757 PR rtl-optimization/80491
31758 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
31760 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
31762 PR tree-optimization/80487
31763 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
31765 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31767 PR tree-optimization/79697
31768 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
31769 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
31770 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
31772 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
31773 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
31775 2017-04-28 Martin Sebor <msebor@redhat.com>
31777 PR tree-optimization/80523
31778 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
31779 (init_target_to_host_charmap, target_to_host, target_strtol10): New
31781 (maybe_warn, format_directive, parse_directive): Use new functions.
31782 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
31784 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
31786 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
31788 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
31790 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
31791 target_header_dir): Set correctly.
31792 * configure: Regenerated.
31793 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
31794 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
31795 instead of SYSTEM_HEADER_DIR.
31797 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
31799 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
31800 (estimate_local_effects): Likewise.
31801 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
31802 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
31803 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
31804 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
31805 do_estimate_edge_time, estimate_edge_time): Likewise.
31806 * ipa-inline-analysis.c (estimate_node_size_and_time,
31807 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
31808 (estimate_time_after_inlining): Remove.
31810 2017-04-28 Martin Liska <mliska@suse.cz>
31812 * doc/gcov.texi: Enhance documentation of gcov.
31814 2017-04-28 Martin Liska <mliska@suse.cz>
31816 * doc/gcov.texi: Sort options in alphabetic order.
31817 * doc/gcov-dump.texi: Likewise.
31818 * doc/gcov-tool.texi: Likewise.
31819 * gcov.c (print_usage): Likewise.
31820 * gcov-dump.c (print_usage): Likewise.
31821 * gcov-tool.c (print_merge_usage_message): Likewise.
31822 (print_rewrite_usage_message): Likewise.
31823 (print_overlap_usage_message): Likewise.
31825 2017-04-28 Martin Liska <mliska@suse.cz>
31827 PR gcov-profile/53915
31828 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
31830 2017-04-28 Martin Liska <mliska@suse.cz>
31832 PR gcov-profile/79891
31833 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
31834 is marked by compiler as living on a line.
31835 (get_cycles_count): Remove usage of the union.
31836 (output_intermediate_file): Likewise.
31837 (find_source): Fix GNU coding style.
31838 (accumulate_line_counts): Remove old non-all block mode.
31839 (output_lines): Remove usage of the union.
31840 * profile.c (output_location): Include all BBs, even if
31841 belonging to a same line (and file) as a previous BB.
31843 2017-04-28 Martin Liska <mliska@suse.cz>
31845 * gcov.c (process_args): Handle new argument 'w'.
31846 (read_graph_file): Assign ID to BBs.
31847 (output_branch_count): Display BB # if verbose flag is set.
31848 (output_lines): Likewise for arcs.
31849 (print_usage): Add '--verbose' option help.
31850 * doc/gcov.texi: Document --verbose (-w) option.
31852 2017-04-28 Martin Liska <mliska@suse.cz>
31854 * gcov.c (struct block_location_info): New struct.
31855 (process_file): Fill up the new structure.
31856 (read_graph_file): Replace usage of encoding by the newly added
31858 (add_line_counts): Likewise.
31859 (accumulate_line_counts): Remove usage of the union.
31860 (function_info::function_info): New function.
31861 (function_info::~function_info): Likewise.
31862 (process_file): Call delete instead of release_function.
31863 (release_function): Release the function.
31864 (release_structures): Call delete instead of release_function.
31865 (solve_flow_graph): Replace usage of num_blocks.
31866 (find_exception_blocks): Likewise.
31867 (output_lines): Fix GNU coding style.
31869 2017-04-28 Martin Liska <mliska@suse.cz>
31872 * coverage.c (coverage_remove_note_file): New function.
31873 * coverage.h: Declare the function.
31874 * toplev.c (finalize): Clean if an error has been seen.
31876 2017-04-28 Martin Liska <mliska@suse.cz>
31878 PR gcov-profile/80031
31879 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
31880 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
31881 * gcov.c (read_graph_file): Read just number of blocks.
31882 * profile.c (branch_prob): Do not stream 0 flags per a basic
31885 2017-04-28 Martin Liska <mliska@suse.cz>
31887 * gcov-dump.c (tag_*): Add new argument to declarations.
31888 (dump_gcov_file): Likewise.
31889 (tag_blocks): Add and use new argument depth.
31890 (tag_arcs): Likewise.
31891 (tag_lines): Likewise.
31892 (tag_counters): Likewise.
31893 (tag_summary): Likewise.
31894 (dump_working_sets): Use depth to do a proper indentation.
31896 2017-04-28 Jakub Jelinek <jakub@redhat.com>
31899 * cgraph.h (symtab_node::debug_symtab): No longer inline.
31900 * symtab.c (symtab_node::debug_symtab): Move definition here.
31902 2017-04-28 Richard Biener <rguenther@suse.de>
31904 * lto-streamer.h (LTO_major_version): Bump to 7.
31906 2017-04-28 Richard Biener <rguenther@suse.de>
31908 * tree-vrp.c (assert_info): New struct.
31909 (add_assert_info): New helper.
31910 (register_edge_assert_for_2): Refactor to add asserts to a vector
31912 (register_edge_assert_for_1): Likewise.
31913 (register_edge_assert_for): Likewise.
31914 (finish_register_edge_assert_for): New helper actually registering
31915 asserts where live on edge.
31916 (find_conditional_asserts): Adjust.
31917 (find_switch_asserts): Likewise.
31918 (evrp_dom_walker::try_find_new_range): Generalize.
31919 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
31921 2017-04-27 Marek Polacek <polacek@redhat.com>
31924 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
31925 arg10 and arg11 to itype.
31927 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
31929 * doc/extend.texi (Object Size Checking): Improve grammar.
31931 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
31934 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
31935 that the logic for permitting reciprocal estimates matches that
31938 2017-04-27 Jakub Jelinek <jakub@redhat.com>
31941 * tree.c (type_cache_hasher::equal): Only compare
31942 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
31943 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
31944 non-aggregate element types.
31945 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
31946 about the flag on ARRAY_TYPEs in the comment, formatting fix.
31948 2017-04-27 Richard Biener <rguenther@suse.de>
31950 PR middle-end/80533
31951 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
31952 stripping ARRAY_REFs from MEM_EXPR make sure we're not
31953 keeping a reference to a trailing array.
31955 2017-04-27 Richard Biener <rguenther@suse.de>
31957 PR middle-end/80539
31958 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
31959 being in loop-closed SSA form conservatively.
31960 (chrec_fold_multiply_poly_poly): Likewise.
31962 2017-04-27 Tamar Christina <tamar.christina@arm.com>
31964 PR middle-end/79665
31965 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
31966 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
31968 2017-04-27 Jakub Jelinek <jakub@redhat.com>
31971 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
31972 (aarch64_function_arg_alignment): Return unsigned int again, but still
31973 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
31974 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
31975 Don't emit -Wpsabi note.
31976 (aarch64_function_arg_boundary): Likewise.
31977 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
31980 2017-04-26 Nathan Sidwell <nathan@acm.org>
31982 * tree.h (crc32_unsigned_n): Declare.
31983 (crc32_unsigned, crc32_unsigned): Make inline.
31984 * tree.c (crc32_unsigned_bits): Replace with ...
31985 (crc32_unsigned_n): ... this.
31986 (crc32_unsigned, crc32_byte): Remove.
31987 (crc32_string): Remove unnecessary braces.
31989 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
31991 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
31992 * ipa-inline-analysis.c (MAX_TIME): Remove.
31993 (account_size_time): Use sreal for time.
31994 (dump_inline_summary): Update.
31995 (estimate_function_body_sizes): Update.
31996 (estimate_edge_size_and_time): Update.
31997 (estimate_calls_size_and_time): Update.
31998 (estimate_node_size_and_time): Update.
31999 (inline_merge_summary): Update.
32000 (inline_update_overall_summary): Update.
32001 (estimate_time_after_inlining): Update.
32002 (inline_read_section): Update.
32003 (inline_write_summary): Update.
32004 * ipa-inline.c (compute_uninlined_call_time): Update.
32005 (compute_inlined_call_time): Update.
32006 (recursive_inlining): Update.
32007 (inline_small_functions): Update.
32008 (dump_overall_stats): Update.
32009 * ipa-inline.h: Include sreal.h.
32010 (size_time_entry): Turn time to sreal.
32011 (inline_summary): Turn self_time nad time to sreal.
32013 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
32015 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
32017 (sreal::stream_out, sreal::stream_in): New.
32018 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
32020 2017-04-25 Jakub Jelinek <jakub@redhat.com>
32022 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
32025 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
32028 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
32030 (dimode_scalar_chain::compute_convert_gain): Ditto.
32031 (dimode_scalar_chain::make_vector_copies): Ditto.
32032 (dimode_scalar_chain::convert_reg): Ditto.
32033 (dimode_scalar_chain::convert_insn): Ditto.
32034 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
32035 (VI248_AVX512BW_1): New mode iterator.
32036 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
32037 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
32040 2017-04-25 Martin Sebor <msebor@redhat.com>
32042 PR tree-optimization/80497
32043 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
32044 constants are representable in HOST_WIDE_INT.
32045 (parse_directive): Ditto.
32047 2017-04-25 Martin Sebor <msebor@redhat.com>
32050 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
32051 (new_zero_array): Adjust signature.
32052 (dom_info::dom_init): Used unsigned rather that size_t.
32053 (dom_info::dom_info): Same.
32055 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
32056 Jakub Jelinek <jakub@redhat.com>
32059 * config/arm/arm.c: Include gimple.h.
32060 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
32061 returns negative, increment ncrn only if it returned positive.
32062 (arm_needs_doubleword_align): Return int instead of bool,
32063 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
32064 members, but if there is any such non-FIELD_DECL
32065 > PARM_BOUNDARY aligned decl, return -1 instead of false.
32066 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
32067 returns negative, increment nregs only if it returned positive.
32068 (arm_setup_incoming_varargs): Likewise.
32069 (arm_function_arg_boundary): Emit -Wpsabi note if
32070 arm_needs_doubleword_align returns negative, return
32071 DOUBLEWORD_ALIGNMENT only if it returned positive.
32073 2017-04-25 Marek Polacek <polacek@redhat.com>
32076 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
32077 first argument to type.
32079 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
32082 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
32083 type checks to test for compatibility instead of equality.
32085 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
32086 Jakub Jelinek <jakub@redhat.com>
32089 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
32091 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
32092 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
32093 the alignment computation, but return their maximum in warn_alignment.
32094 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
32095 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
32097 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
32098 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
32101 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32103 * config/arc/simdext.md (dmpyh): Fix typo.
32105 2017-04-25 Richard Biener <rguenther@suse.de>
32107 PR tree-optimization/80492
32108 * alias.c (compare_base_decls): Handle registers with asm
32109 specification conservatively.
32110 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
32111 compare_base_decls returning dont-know properly.
32113 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32115 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
32116 (legitimate_offset_address_p): New function.
32117 (arc_legitimate_address_p): Use above function.
32119 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32121 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
32123 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32125 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
32126 ACCH registers whenever they are available.
32128 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32130 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
32131 double regs fix when not used.
32133 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32135 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
32137 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
32138 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
32140 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32142 * config/arc/arc.c (arc_output_addsi): Check for h-register class
32143 when emitting short ADD instructions.
32145 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32147 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
32149 (cmpsi_cc_c_insn): Likewise.
32150 (cbranchsi4_scratch): Compute proper instruction length using
32151 compact_hreg_operand.
32152 * config/arc/predicates.md (compact_hreg_operand): New predicate.
32154 2017-04-25 Richard Biener <rguenther@suse.de>
32156 PR middle-end/80509
32157 * passes.c (pass_manager::pass_manager): Initialize
32158 m_name_to_pass_map.
32160 2017-04-25 Richard Biener <rguenther@suse.de>
32162 PR tree-optimization/79201
32163 * tree-ssa-sink.c (statement_sink_location): Handle calls.
32165 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32168 * config/s390/vector.md: Split MEM->GPR vector moves for
32169 non-s_operand addresses.
32171 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32174 * config/s390/predicates.md (reload_const_wide_int_operand): New
32176 * config/s390/s390.md ("movti"): Remove d/P alternative.
32177 ("movti_bigconst"): New pattern definition.
32179 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
32182 * s390-protos.h (s390_expand_cs_hqi): Removed.
32183 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
32184 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
32185 modes as well as CCZ1mode and CCZmode.
32186 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
32187 signature of s390_emit_compare_and_swap.
32188 (s390_expand_cs_hqi): Likewise, make static.
32189 (s390_expand_cs_tdsi): Generate an explicit compare before trying
32190 compare-and-swap, in some cases.
32191 (s390_expand_cs): Wrapper function.
32192 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
32194 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
32195 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
32196 patterns for small and large integers. Forbid symref memory operands.
32197 Move expander to s390.c. Require cc register.
32198 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
32199 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
32200 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
32201 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
32202 symref memory operands. Remove CC mode and call s390_match_ccmode
32204 ("atomic_exchange<mode>"): Allow and implement all integer modes.
32206 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
32208 * config/s390/s390.md (define_peephole2): New peephole to help
32209 combining the load-and-test pattern with volatile memory.
32211 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
32213 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
32214 with CCZmode for TARGET_Z196.
32216 2017-04-25 Jakub Jelinek <jakub@redhat.com>
32218 PR rtl-optimization/80501
32219 * combine.c (make_compound_operation_int): Set subreg_code to SET
32220 even for AND with mask of the sign bit of mode.
32222 PR rtl-optimization/80500
32223 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
32224 sum's initial value.
32226 2017-04-25 Julian Brown <julian@codesourcery.com>
32227 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32229 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
32231 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
32233 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
32235 2017-04-25 Julian Brown <julian@codesourcery.com>
32236 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32238 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
32239 (thunderx2t99_sha): New Reservation.
32241 2017-04-25 Julian Brown <julian@codesourcery.com>
32242 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32244 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
32245 type for 1-element load.
32247 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
32249 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
32251 2017-04-24 Martin Jambor <mjambor@suse.cz>
32253 PR tree-optimization/80293
32254 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
32255 char arrays not totally scalarizable if it is false.
32256 (analyze_all_variable_accesses): Pass correct value in the new
32257 parameter. Add a statistics counter.
32259 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
32261 PR middle-end/79931
32262 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
32264 2017-04-24 Richard Biener <rguenther@suse.de>
32266 PR tree-optimization/80494
32267 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
32268 out for complex types.
32270 2017-04-24 Richard Biener <rguenther@suse.de>
32272 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
32273 * tree-ssa-sccvn.c (print_scc): Print SCC size.
32274 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
32275 (DFS): Adjust and never fail.
32276 (sccvn_dom_walker::fail): Remove.
32277 (sccvn_dom_walker::before_dom_children): Adjust.
32278 (run_scc_vn): Likewise and never fail.
32279 * tree-ssa-pre.c (pass_pre::execute): Adjust.
32280 (pass_fre::execute): Likewise.
32282 2017-04-24 Richard Biener <rguenther@suse.de>
32284 PR tree-optimization/79725
32285 * tree-ssa-sink.c (statement_sink_location): Return whether
32286 failure reason was zero uses. Move that check later.
32287 (sink_code_in_bb): Deal with zero uses by removing the stmt
32290 2017-04-24 Richard Biener <rguenther@suse.de>
32293 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
32294 pointer-based references.
32296 2017-04-24 Richard Biener <rguenther@suse.de>
32299 * pass_manager.h (pass_manager::operator new): Remove.
32300 (pass_manager::operator delete): Likewise.
32301 * passes.c (pass_manager::operator new): Remove.
32302 (pass_manager::operator delete): Likewise.
32303 (pass_manager::pass_manager): Zero individual pass members.
32305 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
32308 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
32309 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
32310 Check "XEXP (src, 1)" operand here.
32311 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
32312 Check "XEXP (src, 1)" operand here.
32313 (dimode_scalar_chain::make_vector_copies): Detect count register
32314 of a shift instruction. Zero extend count register from QImode
32315 to DImode to satisfy vector shift pattern count operand predicate.
32316 Substitute vector shift count operand with a DImode copy.
32317 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
32320 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
32322 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
32323 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
32324 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
32325 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
32326 (UNSPEC_NOREX_MEM): Remove definition.
32328 2017-04-21 Richard Biener <rguenther@suse.de>
32330 PR tree-optimization/79547
32331 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
32332 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
32333 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
32334 without any constraints.
32336 2017-04-21 Richard Biener <rguenther@suse.de>
32338 PR tree-optimization/78847
32339 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
32341 2017-04-21 Richard Biener <rguenther@suse.de>
32343 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
32344 (build_distinct_type_copy): Likewise.
32345 (build_variant_type_copy): Likewise.
32346 * tree.c (build_qualified_type): Pass down mem-stat info.
32347 (build_distinct_type_copy): Likewise.
32348 (build_variant_type_copy): Likewise.
32350 2017-04-21 Richard Biener <rguenther@suse.de>
32352 PR tree-optimization/80237
32353 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
32355 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
32356 for a simplified result.
32358 2016-04-21 Richard Biener <rguenther@suse.de>
32360 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
32361 sth as strict as a simple_iv but a chrec without symbols and an
32362 operand defined in the loop we are peeling (and not some subloop).
32363 (propagate_constants_for_unrolling): Propagate all constants.
32365 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
32368 * config/i386/i386.c (print_reg): Remove assert for disalowed
32369 regno values, call output_operand_lossage instead.
32371 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
32374 * config/i386/constraints.md (Yc): New register constraint.
32375 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
32376 Use Yc constraint for alternative 2 of operand 0. Remove
32377 preferred_for_speed attribute.
32379 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
32381 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
32382 lastprivate clauses in SIMT case.
32384 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
32386 * doc/invoke.texi (-Wextra-semi): Document new warning option.
32388 2017-04-20 Richard Biener <rguenther@suse.de>
32390 PR tree-optimization/57796
32391 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
32392 as N scalar stores.
32393 (vect_model_load_cost): Cost gathers as N scalar loads.
32395 2017-04-20 Richard Biener <rguenther@suse.de>
32397 * ggc-page.c (ggc_allocated_p): Rename to ...
32398 (safe_lookup_page_table_entry): ... this and return the lookup
32400 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
32402 2017-04-20 Richard Biener <rguenther@suse.de>
32404 PR tree-optimization/80453
32405 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
32406 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
32407 from the conditions.
32408 (vn_phi_eq): Pass them down.
32409 (vn_phi_lookup): Record them.
32410 (vn_phi_insert): Likewise.
32412 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
32414 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
32415 uninitialized variable warning to avoid buffer overrun.
32417 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
32420 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
32421 is suppressed for '{ 0 }' in C.
32423 2017-04-20 Jakub Jelinek <jakub@redhat.com>
32425 * BASE-VER: Set to 8.0.0.
32427 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
32429 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
32430 priority .init_array and .fini_array section with SECTION_NOTYPE
32433 2017-04-20 Jakub Jelinek <jakub@redhat.com>
32435 PR middle-end/80423
32436 * tree.h (build_array_type): Add typeless_storage default argument.
32437 * tree.c (type_cache_hasher::equal): Also compare
32438 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
32439 (build_array_type): Add typeless_storage argument, set
32440 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
32442 (build_nonshared_array_type): Adjust build_array_type_1 caller.
32443 (build_array_type): Likewise. Add typeless_storage argument.
32445 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
32446 Jakub Jelinek <jakub@redhat.com>
32448 PR tree-optimization/80426
32449 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
32450 operation on symbolic operands, also compute the overflow for the
32451 invariant part when the operation degenerates into a negation.
32453 2017-04-19 Jakub Jelinek <jakub@redhat.com>
32456 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
32457 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
32460 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
32462 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
32465 * config/avr/avr.c (tree.h): Include it.
32466 (cgraph.h): Include it.
32467 (avr_encode_section_info): Don't warn for uninitialized progmem
32468 variable if it's just an alias.
32470 2017-04-19 Richard Biener <rguenther@suse.de>
32473 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
32474 when needed by AutoPGO.
32476 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
32479 * doc/lto.texi: Remove an extra 'that'.
32481 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
32483 PR rtl-optimization/80429
32484 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
32485 are only used in debug insns.
32487 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
32488 Vladimir Makarov <vmakarov@redhat.com>
32490 * config/sparc/predicates.md (input_operand): Add comment. Return
32491 true for any memory operand when LRA is in progress.
32492 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
32494 2017-04-18 Jeff Law <law@redhat.com>
32497 * mips.md ({return,simple_return}_internal): Do not overwrite
32500 2017-04-18 Jakub Jelinek <jakub@redhat.com>
32502 PR tree-optimization/80443
32503 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
32504 instead of adding 1, subtract -1 and similarly instead of subtracting
32507 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
32509 PR rtl-optimization/80357
32510 * haifa-sched.c (tmp_bitmap): New variable.
32511 (model_recompute): Handle duplicate use records.
32512 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
32513 (free_global_sched_pressure_data): Free it.
32515 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
32518 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
32519 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
32520 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
32521 instead of SYSTEM_HEADER_DIR.
32523 2017-04-18 Jeff Law <law@redhat.com>
32525 PR middle-end/80422
32526 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
32527 predecessors after walking up the insn chain.
32529 2017-04-18 Jakub Jelinek <jakub@redhat.com>
32532 * dwarf2out.c (modified_type_die): Try harder not to emit internal
32533 sizetype type into debug info.
32535 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
32538 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
32539 unneeded test for TARGET_UPPER_REGS_SF.
32540 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
32542 2017-04-18 Jakub Jelinek <jakub@redhat.com>
32545 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
32546 instead of gsi_after_labels.
32548 2017-04-18 Jeff Law <law@redhat.com>
32550 * regcprop.c (maybe_mode_change): Avoid creating copies of the
32554 2017-04-13 Jeff Law <law@redhat.com>
32555 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
32556 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
32558 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
32561 * config/avr/avr.c (intl.h): Include it.
32562 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
32564 2017-04-18 Martin Liska <mliska@suse.cz>
32566 PR gcov-profile/78783
32567 * gcov-tool.c (gcov_output_files): Validate that destination
32568 file is either removed by the tool or by a user.
32570 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
32571 Guy Benyei <guybe@mellanox.com>
32573 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
32574 block, and do not negate it, the stored id is already negative.
32576 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
32578 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
32580 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
32583 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
32584 masks of options that should be turned off if the VSX vector
32585 options are turned off.
32586 (OTHER_P8_VECTOR_MASKS): Likewise.
32587 (OTHER_VSX_VECTOR_MASKS): Likewise.
32588 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
32589 rs6000_disable_incompatible_switches to validate no type switches
32591 (rs6000_incompatible_switch): New function to disallow turning on
32592 other vector options if -mno-vsx, -mno-power8-vector, or
32593 -mno-power9-vector are specified.
32595 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
32597 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
32599 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
32601 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
32602 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
32603 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
32604 (ARG_POINTER_CFA_OFFSET): Likewise.
32606 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
32608 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
32609 conditions to take advantage of various optimizations.
32611 2017-04-13 Jeff Law <law@redhat.com>
32613 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
32614 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
32615 (zero_extendsidi2_dext): Likewise.
32617 2017-04-13 Jakub Jelinek <jakub@redhat.com>
32620 * fold-const.c (fold_ternary_loc): Revert
32621 use op0 instead of fold_convert_loc (loc, type, arg0) part of
32624 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
32626 PR rtl-optimization/80343
32627 * lra-remat.c (update_scratch_ops): Assign original hard reg to
32628 new scratch pseudo.
32630 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
32633 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
32634 to ubsan_encode_value.
32636 2017-04-13 Jeff Law <law@redhat.com>
32638 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
32639 appearing in DEBUG_INSNs.
32641 2017-04-13 Martin Liska <mliska@suse.cz>
32643 PR gcov-profile/80413
32644 * gcov-io.c (gcov_write_string): Copy to buffer just when
32645 allocated size is greater than zero.
32647 2017-04-13 Jakub Jelinek <jakub@redhat.com>
32650 * dwarf2out.c (decls_for_scope): Ignore declarations of
32651 current_function_decl in BLOCK_NONLOCALIZED_VARS.
32653 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
32656 * ipa-visibility.c (non_local_p): Fix typos.
32657 (localize_node): When localizing symbol in same comdat group,
32658 dissolve the group only when we know external symbols are going
32660 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
32662 2017-04-12 Jakub Jelinek <jakub@redhat.com>
32664 PR tree-optimization/79390
32665 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
32666 order does not result in usable sequence, retry with reversed operand
32672 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
32673 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
32674 op0 instead of fold_convert_loc (loc, type, arg0).
32676 2017-04-12 Jeff Law <law@redhat.com>
32678 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
32679 has a delay slot in the generated code.
32681 * config/cris/cris.md (cris_preferred_reload_class): Return
32682 GENNONACR_REGS rather than GENERAL_REGS.
32684 2017-04-12 Jakub Jelinek <jakub@redhat.com>
32687 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
32688 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
32689 signedness of the result type.
32691 2017-04-12 Richard Biener <rguenther@suse.de>
32692 Jeff Law <law@redhat.com>
32694 PR tree-optimization/80359
32695 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
32696 trim stores to TARGET_MEM_REFs.
32698 2017-04-12 Richard Biener <rguenther@suse.de>
32700 PR tree-optimization/79390
32701 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
32702 threading case even more.
32704 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
32707 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
32708 for quad_address_p for TImode, instead of just not indexed_address.
32710 2017-04-12 Richard Biener <rguenther@suse.de>
32711 Bernd Edlinger <bernd.edlinger@hotmail.de>
32713 PR middle-end/79671
32714 * alias.c (component_uses_parent_alias_set_from): Handle
32715 TYPE_TYPELESS_STORAGE.
32716 (get_alias_set): Likewise.
32717 * tree-core.h (tree_type_common): Add typeless_storage flag.
32718 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
32719 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
32720 for types containing members with TYPE_TYPELESS_STORAGE.
32721 (place_field): Likewise.
32722 (layout_type): Likewise for ARRAY_TYPE.
32723 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
32724 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
32725 TYPE_TYPELESS_STORAGE.
32726 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
32728 2017-04-12 Jakub Jelinek <jakub@redhat.com>
32731 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
32732 first argument to type.
32734 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32738 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
32739 CONST0_RTX (mode) rather than const0_rtx where appropriate.
32740 (rs6000_expand_binop_builtin): Likewise.
32741 (rs6000_expand_ternop_builtin): Likewise; also add missing
32742 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
32743 vshasigma built-ins.
32744 * doc/extend.texi: Document that vec_xxpermdi's third argument
32745 must be a constant.
32747 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
32749 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
32750 Use shift_const cost parameter when calculating gain of STV shifts.
32752 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
32754 PR rtl-optimization/70478
32755 * lra-constraints.c (process_alt_operands): Check memory for
32756 disfavoring memory insn operand.
32758 2017-04-11 Jakub Jelinek <jakub@redhat.com>
32760 PR middle-end/80100
32761 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
32762 left shift in unsigned HOST_WIDE_INT type.
32764 PR rtl-optimization/80385
32765 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
32766 (not (neg X)) into (plus X -1) for complex or non-integral modes.
32769 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
32770 if they have any depend clauses.
32772 2017-04-11 Martin Liska <mliska@suse.cz>
32775 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
32776 * ipa-split.c (split_function): Create a local comdat symbol
32777 if caller is in a comdat group.
32779 2017-04-11 Martin Liska <mliska@suse.cz>
32782 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
32785 2017-04-11 Martin Sebor <msebor@redhat.com>
32787 PR middle-end/80364
32788 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
32789 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
32791 (directive::set_width, directive::set_precision, format_character):
32793 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
32796 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
32799 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
32800 conflict, set target->arch_name instead of target->cpu_name.
32802 2017-04-11 Richard Biener <rguenther@suse.de>
32804 PR tree-optimization/80374
32805 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
32806 build_zero_cst, remove fold_convertible_p check again.
32808 2017-04-11 Martin Liska <mliska@suse.cz>
32811 * ubsan.c (instrument_object_size): Do not instrument register
32814 2017-04-11 Jakub Jelinek <jakub@redhat.com>
32817 * config/i386/i386-builtin-types.def
32818 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
32819 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
32820 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
32821 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
32822 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
32823 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
32824 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
32825 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
32826 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
32827 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
32828 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
32829 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
32830 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
32831 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
32832 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
32833 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
32834 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
32835 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
32836 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
32837 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
32838 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
32839 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
32840 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
32841 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
32842 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
32843 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
32844 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
32845 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
32846 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
32847 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
32848 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
32849 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
32850 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
32851 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
32852 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
32853 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
32854 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
32855 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
32856 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
32857 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
32858 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
32859 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
32860 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
32861 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
32862 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
32863 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
32865 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
32866 flag to second_arg_count, handle 4 argument function type _COUNT
32867 aliases, handle second_arg_count on second argument rather than last.
32869 2017-04-10 Jeff Law <law@redhat.com>
32871 PR tree-optimization/80374
32872 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
32873 record anything if we can not convert integer_zero_node to the
32876 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
32879 * config/rs6000/rs6000.c (rs6000_option_override_internal):
32880 Enhance special handling given to the TARGET_P9_MINMAX option in
32881 relation to certain other options.
32883 2017-04-10 Bin Cheng <bin.cheng@arm.com>
32885 PR tree-optimization/80153
32886 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
32887 remove POINTER_PLUS_EXPR's base part directly, rather than through
32890 2017-04-10 Richard Biener <rguenther@suse.de>
32891 Bin Cheng <bin.cheng@arm.com>
32893 PR tree-optimization/80153
32894 * tree-affine.c (aff_combination_to_tree): Get base pointer from
32895 the first element of pointer type aff_tree. Build result expr in
32897 (add_elt_to_tree): Convert to type unconditionally. Remove other
32898 fold_convert calls.
32899 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
32900 (rewrite_use_nonlinear_expr): Check invariant using iv information.
32902 2017-04-10 Richard Biener <rguenther@suse.de>
32904 * tree-ssa-structalias.c (find_func_aliases): Properly handle
32907 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
32909 PR rtl-optimization/70478
32910 * lra-constraints.c (curr_small_class_check): New.
32911 (update_and_check_small_class_inputs): New.
32912 (process_alt_operands): Update curr_small_class_check. Disfavor
32913 alternative insn memory operands. Check available regs for small
32916 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
32919 * config/mips/mips.opt (-mvirt): Update description.
32920 * doc/invoke.texi (-mvirt): Likewise.
32922 2017-04-10 Richard Biener <rguenther@suse.de>
32924 PR middle-end/80362
32925 * fold-const.c (fold_binary_loc): Look at unstripped ops when
32926 looking for NEGATE_EXPR in -A / -B to A / B folding.
32928 2017-04-10 Martin Liska <mliska@suse.cz>
32930 PR gcov-profile/80224
32931 * gcov.c (print_usage): Fix usage string.
32932 (get_gcov_intermediate_filename): Remove.
32933 (output_gcov_file): Use both for normal and intermediate format.
32934 (generate_results): Do not initialize special file for
32935 intermediate format.
32937 2017-04-10 Richard Biener <rguenther@suse.de>
32939 PR tree-optimization/80304
32940 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
32943 2017-04-10 Nathan Sidwell <nathan@acm.org>
32946 * config/rs6000/rs6000.c (rs6000_vector_type): New.
32947 (rs6000_init_builtins): Use it.
32949 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32951 * config/arm/arm.md (<mrc>): Add mode to SET source.
32952 (<mrrc>): Likewise.
32954 2017-04-10 Richard Biener <rguenther@suse.de>
32956 PR middle-end/80344
32957 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
32959 2017-04-10 Jakub Jelinek <jakub@redhat.com>
32962 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
32963 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
32964 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
32965 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
32966 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
32967 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
32968 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
32969 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
32970 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
32971 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
32972 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
32973 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
32974 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
32975 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
32976 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
32977 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
32978 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
32979 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
32980 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
32981 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
32982 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
32983 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
32984 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
32986 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
32988 PR rtl-optimization/70478
32989 * lra-constraints.c: Reverse the last patch.
32991 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
32993 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
32994 Add comment for WCHAR_T.
32996 2017-04-08 Martin Liska <mliska@suse.cz>
32999 2017-04-07 Martin Liska <mliska@suse.cz>
33002 * ipa-split.c (split_function): Add function part to a same comdat
33005 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33008 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
33010 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
33012 * rs6000/rs6000.c (vec_load_pendulum): Rename...
33013 (vec_pairing): ...to this.
33014 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
33015 (rs6000_sched_init): Adjust for name change.
33016 (struct rs6000_sched_context): Likewise.
33017 (rs6000_init_sched_context): Likewise.
33018 (rs6000_set_sched_context): Likewise.
33020 2017-04-07 Jakub Jelinek <jakub@redhat.com>
33026 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
33028 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
33029 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
33030 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
33032 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
33034 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
33036 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
33038 PR rtl-optimization/70703
33039 * ira-color.c (update_conflict_hard_regno_costs): Use
33040 int64_t instead of HOST_WIDE_INT.
33042 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
33044 PR rtl-optimization/70478
33045 * lra-constraints.c (process_alt_operands): Disfavor alternative
33046 insn memory operands.
33048 2017-04-07 Jeff Law <law@redhat.com>
33050 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
33051 CALL and NOTE_INSN_CALL_ARG_LOCATION.
33053 2017-04-07 Martin Liska <mliska@suse.cz>
33056 * config/aarch64/aarch64.c (aarch64_process_target_attr):
33057 Show error message instead of an ICE.
33059 2017-04-07 Martin Liska <mliska@suse.cz>
33062 * ipa-split.c (split_function): Add function part to a same comdat
33065 2017-04-07 Richard Biener <rguenther@suse.de>
33067 PR middle-end/80341
33068 * tree.c (get_unwidened): Also handle ! for_type case for
33070 * convert.c (do_narrow): Split out from ...
33071 (convert_to_integer_1): ... here. Do not pass final truncation
33072 type to get_unwidened for TRUNC_DIV_EXPR.
33074 2017-04-07 Richard Biener <rguenther@suse.de>
33076 * tree-affine.c (wide_int_ext_for_comb): Take type rather
33078 (aff_combination_const): Adjust.
33079 (aff_combination_scale): Likewise.
33080 (aff_combination_add_elt): Likewise.
33081 (aff_combination_add_cst): Likewise.
33082 (aff_combination_convert): Likewise.
33083 (add_elt_to_tree): Likewise. Remove unused argument.
33084 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
33086 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
33088 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
33090 * config/arm/arm.c (arm_default_short_enums): Use
33091 ARM_DEFAULT_SHORT_ENUMS.
33092 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
33094 2017-04-06 Jakub Jelinek <jakub@redhat.com>
33097 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
33098 members with redundant out-of-class redeclaration.
33100 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
33103 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
33104 * config/i386/i386.md (*zero_extendsidi2):
33105 Add (?*x,*x) and (?*v,*v) alternatives.
33107 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
33110 * config/i386/i386.c (ix86_expand_builtin)
33111 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
33112 mode from insn data. Convert operands to insn operand mode.
33113 Copy operands that don't satisfy insn predicate to a register.
33115 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
33117 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
33120 2017-04-06 Richard Biener <rguenther@suse.de>
33122 PR tree-optimization/80334
33123 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
33124 preserve alignment of accesses.
33126 2017-04-06 Richard Biener <rguenther@suse.de>
33128 PR tree-optimization/80262
33129 * tree-sra.c (build_ref_for_offset): Preserve address-space
33131 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
33132 Drop useless address-space information on MEM_REF offsets.
33134 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
33136 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
33138 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
33140 PR rtl-optimization/70703
33141 * ira-color.c (update_conflict_hard_regno_costs): Use
33142 HOST_WIDE_INT instead of long.
33144 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
33147 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
33148 not defined for x86_64 target. Add -mmmx target option when __SSE2__
33150 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
33151 for x86_64 target. Handle -m3dnowa option.
33153 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
33155 PR rtl-optimization/70703
33156 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
33157 (update_conflict_hard_regno_costs): Use long instead of unsigned
33158 arithmetic for cost calculation.
33160 2017-04-05 Jakub Jelinek <jakub@redhat.com>
33161 Bernd Edlinger <bernd.edlinger@hotmail.de>
33164 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
33167 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
33170 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
33171 ptr_mode with Pmode throughout.
33172 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
33173 into probe_stack_range and use DImode.
33175 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
33178 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
33179 call_eh_return is true.
33181 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33183 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
33184 Initialize last_match_fntype_index.
33186 2017-04-05 Jakub Jelinek <jakub@redhat.com>
33189 * tree-nvr.c: Include internal-fn.h.
33190 (pass_return_slot::execute): Ignore internal calls without
33193 2017-04-04 Jakub Jelinek <jakub@redhat.com>
33194 Richard Biener <rguenther@suse.de>
33197 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
33198 captures used multiple times, except for the last use.
33199 * generic-match-head.c: Include gimplify.h.
33201 2017-04-04 Jakub Jelinek <jakub@redhat.com>
33203 PR tree-optimization/79390
33204 * target.h (struct noce_if_info): Declare.
33205 * targhooks.h (default_noce_conversion_profitable_p): Declare.
33206 * target.def (noce_conversion_profitable_p): New target hook.
33207 * ifcvt.h (struct noce_if_info): New type, moved from ...
33208 * ifcvt.c (struct noce_if_info): ... here.
33209 (noce_conversion_profitable_p): Renamed to ...
33210 (default_noce_conversion_profitable_p): ... this. No longer
33212 (noce_try_store_flag_constants, noce_try_addcc,
33213 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
33214 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
33215 instead of noce_conversion_profitable_p.
33216 * config/i386/i386.c: Include ifcvt.h.
33217 (ix86_option_override_internal): Don't override
33218 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
33219 (ix86_noce_conversion_profitable_p): New function.
33220 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
33221 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
33222 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
33223 * doc/tm.texi: Regenerated.
33225 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33227 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
33230 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
33233 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
33234 instructions for small multiply cores.
33236 2017-04-04 Jeff Law <law@redhat.com>
33238 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
33240 (mips_expand_vec_perm_const): Initialize elements in orig_perm
33241 that are not set by the loop over the elements.
33243 2017-04-04 Jakub Jelinek <jakub@redhat.com>
33246 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
33247 int mode, convert_modes it to mode as unsigned, otherwise use
33248 lowpart_subreg to mode rather than SImode.
33249 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
33250 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
33251 Use DImode instead of SImode for the shift count operand.
33252 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
33255 2017-04-04 Richard Biener <rguenther@suse.de>
33257 PR middle-end/80281
33258 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
33259 arithmetic done for the negate or the plus. Simplify.
33260 (A - (-B) -> A + B): Likewise.
33261 * fold-const.c (split_tree): Make sure to not negate pointers.
33263 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
33265 PR rtl-optimization/60818
33266 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
33267 a compare of comparisons with the thing compared if this results
33268 in a different machine mode.
33270 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
33272 * alias.c (base_alias_check): Fix typo in comment.
33273 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
33274 * cgraphunit.c (symbol_table::compile): Likewise.
33275 * collect2.c (maybe_run_lto_and_relink): Likewise.
33276 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
33277 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
33278 * config/avr/avr.c (avr_map_op_t): Likewise.
33279 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
33280 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
33281 * config/epiphany/epiphany.md (movcc): Likewise.
33282 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
33283 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
33285 * config/mips/mips.c (mips_save_restore_reg): Likewise.
33286 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
33287 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
33288 * config/sh/sh.c (sh_rtx_costs): Likewise.
33289 * fold-const.c (fold_truth_andor): Likewise.
33290 * genautomata.c (collapse_flag): Likewise.
33291 * gengtype.h (struct type::u::s): Likewise.
33292 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
33293 * input.c (FORMAT_AMOUNT): Likewise.
33294 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
33295 (known_aggs_to_agg_replacement_list): Likewise.
33296 * ipa-inline-analysis.c: Likewise.
33297 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
33298 * ipa-polymorphic-call.c
33299 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
33300 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
33301 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
33303 * modulo-sched.c (apply_reg_moves): Likewise.
33304 * omp-expand.c (build_omp_regions_1): Likewise.
33305 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
33306 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
33307 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
33308 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
33309 * value-prof.c: Likewise.
33310 * var-tracking.c (val_reset): Likewise.
33312 2017-04-03 Richard Biener <rguenther@suse.de>
33314 PR tree-optimization/80275
33315 * fold-const.c (split_address_to_core_and_offset): Handle
33318 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
33320 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
33321 descriptors is at least equal to that of functions.
33323 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
33325 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
33327 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
33330 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
33331 (mov<IMOD4:mode>): New expander.
33332 (*mov<IMOD4:mode>_internal): New insn and split pattern.
33334 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
33336 PR rtl-optimization/79405
33337 * fwprop.c (propagations_left): New variable.
33338 (forward_propagate_into): Decrement it.
33339 (fwprop_init): Initialize it.
33340 (fw_prop): If the variable has reached zero, stop propagating.
33341 (fwprop_addr): Ditto.
33343 2017-03-31 Jakub Jelinek <jakub@redhat.com>
33346 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
33347 a FUNCTION_DECL, pass it as decl instead of origin to
33350 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
33352 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
33355 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
33358 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
33359 TARGET_VSX_SMALL_INTEGER.
33361 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33363 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
33364 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
33366 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
33368 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
33369 extraction from odd-numbered MSA register.
33371 2017-03-31 Jakub Jelinek <jakub@redhat.com>
33373 PR middle-end/80173
33374 * expmed.c (store_bit_field_1): Don't attempt to create
33375 a word subreg out of hard registers wider than word if they
33376 have HARD_REGNO_NREGS of 1 for their mode.
33378 PR middle-end/80163
33379 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
33380 conversions to integer types wider than word and pointer.
33383 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
33384 (rtx_equal_for_cselib_p): Pass 0 to it.
33385 * cselib.c (cselib_hasher::equal): Likewise.
33386 (rtx_equal_for_cselib_1): Add depth argument. If depth
33387 is 128, don't look up VALUE locs and punt. Increment
33388 depth in recursive calls when walking VALUE locs.
33390 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
33392 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
33393 (make_gcov_file_name): Use the canonical path name for generating
33395 (read_line): Fix handling of files with ascii null bytes.
33397 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
33399 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
33400 to initialise a vector register instead
33401 of using a const_int.
33403 2017-03-30 Jakub Jelinek <jakub@redhat.com>
33405 PR translation/80189
33406 * gimplify.c (omp_default_clause): Use %qs instead of %s in
33407 diagnostic messages.
33409 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
33412 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
33413 (dfp_diex_<mode>): Update mode of operand 1.
33414 * doc/extend.texi (dxex, dxexq): Document change to return type.
33415 (diex, diexq): Document change to argument type.
33417 2017-03-30 Martin Jambor <mjambor@suse.cz>
33420 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
33421 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
33422 it reflects the signature changes performed at the callee side.
33423 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
33424 to cgraph_build_function_type_skip_args.
33425 (build_function_decl_skip_args): Adjust call to the above function.
33427 2017-03-30 Jakub Jelinek <jakub@redhat.com>
33430 * config/i386/sse.md
33431 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
33432 register as dest whenever it is a MEM not rtx_equal_p to the
33433 corresponding dup operand, and when forcing into reg move the
33434 reg into the memory afterwards.
33435 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
33436 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
33437 for the force_reg mode.
33438 (avx512vl_vextractf128<mode>): Use register as dest either
33439 always when a MEM, or when it is a MEM not rtx_equal_p to the
33440 corresponding dup operand, or even not when it is a CONST_VECTOR
33441 depending on the mode and lo vs. hi.
33442 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
33444 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
33445 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
33446 Likewise. Require that operands[2] is even.
33447 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
33448 Remove extraneous parens. Require that operands[2] is a multiple
33450 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
33451 operands[0] is a MEM if <mask_applied>, the predicates/constraints
33452 disallow memory then.
33454 2017-03-30 Richard Biener <rguenther@suse.de>
33456 PR tree-optimization/77498
33457 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
33458 to non-constants over backedges.
33460 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
33462 PR rtl-optimization/80233
33463 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
33464 as last_combined_insn. Do not test for BARRIER_P separately.
33466 2017-03-29 Andreas Schwab <schwab@suse.de>
33469 * calls.c (prepare_call_address): Convert funexp to Pmode before
33470 copying to temp reg.
33472 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33474 PR tree-optimization/80158
33475 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
33476 Handle possible future case of more than one alternate
33478 (replace_rhs_if_not_dup): Likewise.
33479 (replace_one_candidate): Likewise.
33481 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
33483 PR rtl-optimization/80193
33484 * ira.c (ira): Do not check allocation for LRA.
33486 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
33488 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
33489 (nvptx_output_simt_exit): Declare.
33490 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
33491 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
33492 (init_softstack_frame): Move initialization of crtl->is_leaf to...
33493 (nvptx_declare_function_name): ...here. Emit declaration of local
33494 memory space buffer for omp_simt_enter insn.
33495 (nvptx_output_unisimt_switch): New.
33496 (nvptx_output_softstack_switch): New.
33497 (nvptx_output_simt_enter): New.
33498 (nvptx_output_simt_exit): New.
33499 * config/nvptx/nvptx.h (struct machine_function): New fields
33500 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
33501 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
33502 (UNSPECV_SIMT_EXIT): Ditto.
33503 (omp_simt_enter_insn): New insn.
33504 (omp_simt_enter): New expansion.
33505 (omp_simt_exit): New insn.
33506 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
33508 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
33509 (expand_GOMP_SIMT_ENTER_ALLOC): New.
33510 (expand_GOMP_SIMT_EXIT): New.
33511 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
33512 (GOMP_SIMT_ENTER_ALLOC): Ditto.
33513 (GOMP_SIMT_EXIT): Ditto.
33514 * target-insns.def (omp_simt_enter): New insn.
33515 (omp_simt_exit): Ditto.
33516 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
33518 (lower_rec_simd_input_clauses): Implement SIMT privatization.
33519 (lower_rec_input_clauses): Likewise.
33520 (lower_lastprivate_clauses): Handle SIMT privatization.
33522 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
33523 (ompdevlow_adjust_simt_enter): New.
33524 (find_simtpriv_var_op): New.
33525 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
33526 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
33528 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
33529 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
33530 (copy_decl_for_dup_finish): Ditto.
33532 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
33534 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
33537 * config/i386/i386.c (ix86_option_override_internal): Always
33538 allow -mpreferred-stack-boundary=3 for 64-bit targets.
33540 2017-03-28 Bin Cheng <bin.cheng@arm.com>
33542 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
33544 2017-03-28 Bin Cheng <bin.cheng@arm.com>
33546 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
33547 mark new edge's irreducible flag accordign to it.
33548 (vect_do_peeling): Check loop preheader edge's irreducible flag
33549 and pass it to function slpeel_add_loop_guard.
33551 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
33553 PR tree-optimization/80218
33554 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
33555 Update block frequencies and counts.
33557 2017-03-28 Richard Biener <rguenther@suse.de>
33559 PR tree-optimization/78644
33560 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
33561 of a simplification result we may not use it at all.
33563 2017-03-28 Richard Biener <rguenther@suse.de>
33566 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
33567 without arguments, generate default definition of a SSA name.
33569 2017-03-28 Richard Biener <rguenther@suse.de>
33571 PR middle-end/80222
33572 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
33573 TYPE_REF_CAN_ALIAS_ALL references.
33574 * fold-const.c (fold_indirect_ref_1): Likewise.
33576 2017-03-28 Martin Liska <mliska@suse.cz>
33579 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
33580 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
33582 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
33583 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
33585 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
33586 (EXTRA_SPECS): Define.
33587 (SUBTARGET_EXTRA_SPECS): Likewise.
33588 (SUBTARGET_CPP_SPEC): Likewise.
33589 * config/arc/elf.h (EXTRA_SPECS): Renamed to
33590 SUBTARGET_EXTRA_SPECS.
33591 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
33593 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
33595 * config/arc/simdext.md (vst64_insn): Update pattern.
33596 (vld32wh_insn): Likewise.
33597 (vld32wl_insn): Likewise.
33598 (vld64_insn): Likewise.
33599 (vld32_insn): Likewise.
33601 2017-03-28 Marek Polacek <polacek@redhat.com>
33604 * fold-const.c (fold_comparison): Use protected_set_expr_location
33605 instead of SET_EXPR_LOCATION.
33607 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
33609 * tree.c (add_expr): Avoid name lookup warning.
33611 2017-03-27 Jeff Law <law@redhat.com>
33613 PR tree-optimization/80216
33614 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
33615 function name. Limit recursion depth.
33616 (record_temporary_equivalences): Corresponding changes.
33618 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
33620 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
33623 2017-03-27 Jakub Jelinek <jakub@redhat.com>
33626 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
33628 * cfgcleanup.c (reg_note_cfa_p): New array.
33629 (insns_have_identical_cfa_notes): New function.
33630 (old_insns_match_p): Don't cross-jump in between /f
33631 and non-/f instructions. If both i1 and i2 are frame related,
33632 verify all CFA notes, their order and content.
33634 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
33637 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
33638 HImode and SImode with zero extend to DImode to one insn.
33639 (bswap<mode>2_extenddi): Likewise.
33640 (bswapsi2_extenddi): Likewise.
33641 (bswaphi2_extendsi): Likewise.
33642 (bswaphi2): Combine bswap HImode and SImode into one insn.
33643 Separate memory insns from swapping register.
33644 (bswapsi2): Likewise.
33645 (bswap<mode>2): Likewise.
33646 (bswaphi2_internal): Delete, no longer used.
33647 (bswapsi2_internal): Likewise.
33648 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
33649 store, and gpr<-gpr swap insns.
33650 (bswap<mode>2_store): Likewise.
33651 (bswaphi2_reg): Register only splitter, combine with the splitter.
33652 (bswaphi2 splitter): Likewise.
33653 (bswapsi2_reg): Likewise.
33654 (bswapsi2 splitter): Likewise.
33655 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
33656 the insns into load, store, and register/register insns.
33657 (bswapdi2_ldbrx): Likewise.
33658 (bswapdi2_load): Likewise.
33659 (bswapdi2_store): Likewise.
33660 (bswapdi2_reg): Likewise.
33662 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
33664 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
33665 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
33667 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
33670 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
33672 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
33673 special handling for target option conflicts between dform
33674 options (-mpower9-dform, -mpower9-dform-vector,
33675 -mpower9-dform-scalar) and -mno-direct-move.
33677 2017-03-27 Richard Biener <rguenther@suse.de>
33679 PR tree-optimization/80181
33680 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
33682 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
33684 * config/arc/predicates.md (move_double_src_operand): Replace the
33685 call to move_double_src_operand with a call to address_operand.
33687 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
33689 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
33690 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
33691 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
33693 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
33695 * config/arc/predicates.md (long_immediate_loadstore_operand):
33696 Consider scaled addresses cases.
33698 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
33700 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
33701 restored when in interrupt.
33702 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
33703 doesn't have delay slot.
33705 2017-03-27 Richard Biener <rguenther@suse.de>
33708 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
33709 inlined thunk clones.
33711 2017-03-27 Jakub Jelinek <jakub@redhat.com>
33714 * asan.c (instrument_derefs): Copy over last operand from
33715 original COMPONENT_REF to the new COMPONENT_REF with
33716 DECL_BIT_FIELD_REPRESENTATIVE.
33717 * ubsan.c (instrument_object_size): Likewise.
33719 2017-03-27 Richard Biener <rguenther@suse.de>
33721 PR tree-optimization/80170
33722 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
33723 sure DR/SCEV didnt fold in constants we do not see when looking
33724 at the reference base alignment.
33726 2017-03-27 Richard Biener <rguenther@suse.de>
33728 PR middle-end/80171
33729 * gimple-fold.c (fold_ctor_reference): Properly guard against
33730 NULL return value from canonicalize_constructor_val.
33732 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
33735 * config/i386/i386.c (ix86_expand_builtin)
33736 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
33737 flags reg setting and flags reg using instructions.
33738 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
33739 clobbering instructions to zero extend op2.
33741 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
33743 * doc/install.texi (Configuration) <--with-aix-soname>:
33744 Update link to AIX ld.
33746 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
33748 PR rtl-optimization/80160
33749 PR rtl-optimization/80159
33750 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
33751 reg_alternate_class into account.
33753 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
33756 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
33757 to consider in curr_insn_transform.
33759 2017-03-24 Jakub Jelinek <jakub@redhat.com>
33761 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
33762 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
33763 and emit_mode_inner.
33765 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33767 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
33768 argument to the overloaded builtin variants. Use the new flag to
33769 deprecate certain builtin variants.
33770 * config/s390/s390-builtin-types.def: Add new builtin types.
33771 * config/s390/s390-builtins.h: Support new flags field for
33772 overloaded builtins.
33773 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
33774 (s390_macro_to_expand): Enable vector float data type.
33775 (s390_cpu_cpp_builtins_internal): Indicate support of the new
33776 builtins by incrementing the __VEC__ version number.
33777 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
33779 (s390_resolve_overloaded_builtin): Emit error messages depending
33780 on the builtin flags.
33781 * config/s390/s390.c (s390_expand_builtin): Support additional
33782 flags argument. Change error message to match the messages
33783 emitted in s390-c.c.
33784 * config/s390/s390.md: New UNSPEC_* constants.
33785 (op_type): Add new instruction types.
33786 * config/s390/vecintrin.h: Add new builtins and test data class
33788 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
33789 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
33790 (VEC_INEXACT, VEC_NOINEXACT): New constants.
33791 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
33792 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
33793 ("vec_mergel<mode>"): V_HW -> VEC_HW.
33795 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
33796 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
33797 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
33798 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
33800 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
33801 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
33802 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
33803 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
33805 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
33806 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
33807 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
33808 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
33809 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
33810 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
33811 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
33813 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
33814 ("vec_scatter_element<V_HW_4:mode>_DI")
33815 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
33816 ("vec_fpint<mode>", "vflls")
33817 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
33818 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
33819 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
33820 ("*vec_cmphe<mode>_cc"): ... these.
33822 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
33823 mode constant instead of magic value.
33825 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33827 * config/s390/s390.c (s390_expand_vec_compare): Support other
33828 vector floating point modes than just V2DF.
33829 (s390_expand_vcond): Likewise.
33830 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
33831 (s390_cannot_change_mode_class): Prevent mode changes between TF
33832 and V1TF in vector registers.
33833 * config/s390/s390.md (DF, SF): New mode attributes.
33834 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
33835 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
33836 SFmode support for VRs.
33837 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
33839 (VFT, VF_HW): New mode iterators.
33840 (vw, sdx): New mode attributes.
33841 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
33842 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
33843 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
33844 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
33845 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
33846 also the new vector floating point modes. Renaming to ...
33848 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
33849 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
33850 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
33851 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
33852 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
33853 ("vec_unordered<mode>"): ... these.
33855 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
33856 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
33857 ("*vec_extendv2df"): New insn definitions.
33859 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33861 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
33862 ("mulditi3_2", "*muldi3_sign"): New patterns.
33863 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
33864 rename the pattern definition.
33866 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33868 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
33870 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
33872 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33874 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
33875 instruction if possible.
33876 * config/s390/vector.md (vec_halfnumelts): New mode
33878 ("*vec_vllezlf<mode>"): New pattern.
33880 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33882 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
33883 ("popcountv4si2", "popcountv2di2"): Rename to ...
33884 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
33885 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
33887 ("popcount<mode>2_vxe"): New pattern.
33889 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33891 * common/config/s390/s390-common.c (processor_flags_table): Add
33893 * config.gcc: Add arch12.
33894 * config/s390/driver-native.c (s390_host_detect_local_cpu):
33895 Default to arch12 for unknown CPU model numbers.
33896 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
33897 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
33898 PROCESSOR_max sanity check.
33899 * config/s390/s390-opts.h (enum processor_type): Add
33901 * config/s390/s390.c (processor_table): Add arch12.
33902 (s390_expand_builtin): Add check for B_VXE flag.
33903 (s390_issue_rate): Add PROCESSOR_ARCH12.
33904 (s390_get_sched_attrmask): Likewise.
33905 (s390_get_unit_mask): Likewise.
33906 (s390_sched_score): Enable z13 scheduling for arch12.
33907 (s390_sched_reorder): Likewise.
33908 (s390_sched_variable_issue): Likewise.
33909 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
33911 (s390_tune_attr): Use z13 scheduling also for arch12.
33912 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
33913 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
33914 (TARGET_VXE_P): New macros.
33915 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
33916 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
33917 * config/s390/s390.opt: Add arch12 as processor_type.
33919 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33921 * config/s390/s390.md
33922 ("fixuns_truncdddi2", "fixuns_trunctddi2")
33923 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
33924 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
33926 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
33927 Rename expanders to ...
33929 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
33930 ("fixuns_truncdddi2_emu"): ... these.
33932 ("fixuns_trunc<mode>si2_emu"): New expander.
33934 ("*fixuns_truncdfdi2_z13"): Rename to ...
33935 ("*fixuns_truncdfdi2_vx"): ... this.
33937 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33939 * config/s390/2964.md: Remove the single element vector compare
33940 instructions which are no longer used.
33941 * config/s390/s390.c (s390_select_ccmode): Remove handling of
33943 (s390_canonicalize_comparison): Remove handling of DFmode
33945 (s390_expand_vec_compare_scalar): Remove function.
33946 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
33947 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
33949 ("*cmp<mode>_ccs"): Add wfcdb instruction.
33951 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33953 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
33955 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
33956 will anyway by matched by mov<mode>_64dfp.
33958 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33960 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
33961 vlef/vstef. Add missing operand to vleif.
33963 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33965 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
33966 pair for all vector types with 64 bit elements.
33967 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
33968 * config/s390/vector.md (V_HW_64): ... here.
33969 (V_128_NOSINGLE): New mode iterator.
33970 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
33971 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
33972 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
33973 ("*vec_load_pairv2di"): Change to ...
33974 ("*vec_load_pair<mode>"): ... this one.
33976 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33978 * config/s390/constraints.md: Add comments.
33979 (jKK): Reject element sizes > 8 bytes.
33980 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
33982 * config/s390/s390.md: Add the s_operand checks formerly in
33983 s390_split_ok_p to various splitters where they are still
33985 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
33986 for 128 bit vectors. Plus two splitters.
33988 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33990 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
33993 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33996 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
33997 error if the boundary argument is not constant.
33999 2017-03-24 Jakub Jelinek <jakub@redhat.com>
34001 PR rtl-optimization/80112
34002 * loop-doloop.c (doloop_condition_get): Don't check condition
34003 if cmp isn't SET with IF_THEN_ELSE src.
34005 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34007 PR tree-optimization/80158
34008 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
34009 replacing a candidate statement, also replace it for the
34010 candidate's alternate interpretation.
34011 (replace_rhs_if_not_dup): Likewise.
34012 (replace_one_candidate): Likewise.
34014 2017-03-24 Richard Biener <rguenther@suse.de>
34016 PR tree-optimization/80167
34017 * graphite-isl-ast-to-gimple.c
34018 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
34020 (translate_isl_ast_to_gimple::get_rename): Likewise.
34022 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
34024 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
34025 handling of certain combinations of target options, including the
34026 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
34027 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
34029 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34032 * config/arm/arm.md (*load_multiple): Add reload_completed to
34033 matching condition.
34035 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34036 Richard Biener <rguenth@suse.de>
34038 PR tree-optimization/79908
34039 PR tree-optimization/80136
34040 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
34041 been cast away, gimplify_and_add suffices.
34043 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
34045 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
34047 2017-03-23 Richard Biener <rguenther@suse.de>
34049 PR tree-optimization/80032
34050 * gimplify.c (gimple_push_cleanup): Forced unconditional
34051 cleanups still have to go to the conditional_cleanups
34054 2017-03-22 Jakub Jelinek <jakub@redhat.com>
34056 PR tree-optimization/80072
34057 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
34059 (next_operand_entry_id): Change type to unsigned int.
34060 (sort_by_operand_rank): Make sure to return the right return value
34061 even if unsigned fields are bigger than INT_MAX.
34062 (struct oecount): Change cnt and id type to unsigned int.
34063 (oecount_hasher::equal): Formatting fix.
34064 (oecount_cmp): Make sure to return the right return value
34065 even if unsigned fields are bigger than INT_MAX.
34066 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
34069 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
34070 TREE_READONLY on result if writing it more than once.
34073 * doc/invoke.texi (-fsanitize=thread): Document that with
34074 -fnon-call-exceptions atomics are not able to throw
34078 * tsan.c: Include tree-eh.h.
34079 (instrument_builtin_call): Call maybe_clean_eh_stmt or
34080 maybe_clean_or_replace_eh_stmt where needed.
34081 (instrument_memory_accesses): Add cfg_changed argument.
34082 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
34083 if it returned true.
34084 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
34086 PR rtl-optimization/63191
34087 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
34088 wrapper function, moved the whole old content into ...
34089 (ix86_delegitimize_address_1): ... this. New inline function.
34090 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
34091 true as last argument instead of ix86_delegitimize_address.
34093 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
34095 * config/aarch64/aarch64.c (generic_branch_cost): Copy
34096 cortexa57_branch_cost.
34098 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
34100 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
34102 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34105 * doc/md.texi (Constraints): Document wA constraint.
34106 * config/rs6000/constraints.md (wA): New.
34107 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
34108 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
34109 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
34110 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
34112 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
34115 * gimplify.c (is_oacc_declared): New function.
34116 (oacc_default_clause): Use it to set default flags for acc declared
34117 variables inside parallel regions.
34118 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
34119 declared variables.
34120 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
34121 declare attribute to any decl as necessary.
34123 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
34126 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
34127 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
34128 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
34129 (arm_arch_lpae): This.
34130 * config/arm/arm.c (arm_arch7ve): Rename into ...
34131 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
34132 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
34135 2017-03-22 Martin Liska <mliska@suse.cz>
34138 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
34139 error message instead of an ICE.
34141 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34143 * doc/extend.texi (6.11 Additional Floating Types): Revise.
34145 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
34147 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
34149 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
34152 2017-03-21 Martin Sebor <msebor@redhat.com>
34154 * doc/extend.texi: Use "cannot" instead of "can't."
34155 * doc/hostconfig.texi: Same.
34156 * doc/install.texi: Same.
34157 * doc/invoke.texi: Same.
34158 * doc/loop.texi: Same.
34159 * doc/md.texi: Same.
34160 * doc/objc.texi: Same.
34161 * doc/rtl.texi: Same.
34162 * doc/tm.texi: Same.
34163 * doc/tm.texi.in: Same.
34164 * doc/trouble.texi: Same.
34166 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
34169 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
34170 (collect_checksum_attributes): Set it.
34171 (die_checksum_ordered): Use it.
34173 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34175 PR tree-optimization/79908
34176 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
34177 change: For a VA_ARG whose LHS has been cast away, use
34178 force_gimple_operand to construct the side effects.
34180 2017-03-21 David Malcolm <dmalcolm@redhat.com>
34182 PR translation/80001
34183 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
34184 more amenable to translation.
34185 (oacc_loop_auto_partitions): Likewise.
34187 2017-03-21 Marek Polacek <polacek@redhat.com>
34188 Martin Sebor <msebor@redhat.com>
34190 PR tree-optimization/80109
34191 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
34192 on INTEGRAL_TYPE_P.
34194 2017-03-21 Jakub Jelinek <jakub@redhat.com>
34195 Segher Boessenkool <segher@kernel.crashing.org>
34198 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
34199 check reg_used_between_p between insn and one of succ or succ2
34200 depending on if succ is artificial insn not inserted into insn
34203 2017-03-21 Martin Liska <mliska@suse.cz>
34205 PR gcov-profile/80081
34206 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
34207 * doc/gcc.texi: Include gcov-dump stuff.
34208 * doc/gcov-dump.texi: New file.
34210 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
34212 PR rtl-optimization/79150
34213 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
34214 conditional jump, if the jump is the last insn of the loop.
34216 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34217 Richard Biener <rguenth@suse.de>
34219 PR tree-optimization/79908
34220 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
34221 been cast away, use force_gimple_operand to construct the side
34224 2017-03-21 Martin Liska <mliska@suse.cz>
34226 PR libfortran/79956
34227 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
34230 2017-03-21 Brad Spengler <spender@grsecurity.net>
34233 * plugin.c (htab_hash_plugin): New function.
34234 (add_new_plugin): Use it and adjust.
34235 (parse_plugin_arg_opt): Adjust.
34236 (init_one_plugin): Likewise.
34238 2017-03-21 Richard Biener <rguenther@suse.de>
34240 PR tree-optimization/80032
34241 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
34242 if set force the cleanup to happen unconditionally.
34243 (gimplify_target_expr): Push inserted clobbers with force_uncond
34244 to avoid them being removed by control-dependent DCE.
34246 2017-03-21 Richard Biener <rguenther@suse.de>
34248 PR tree-optimization/80122
34249 * tree-inline.c (copy_bb): Do not expans va-arg packs or
34250 va_arg_pack_len when the inlined call stmt requires pack
34252 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
34254 2017-03-21 Jakub Jelinek <jakub@redhat.com>
34257 * tsan.c (instrument_builtin_call): If the memory model argument
34258 is not a constant, assume it is valid.
34261 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
34264 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
34266 PR rtl-optimization/79910
34267 * combine.c (can_combine_p): Do not allow combining an I0 or I1
34268 if its dest is used by an insn before I2 (other than the combined
34269 insns themselves, which are properly handled already).
34271 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
34274 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
34276 * combine.c (record_used_regs): New static function.
34277 (try_combine): Handle situations where there is an additional
34278 instruction between I2 and I3 which needs to have a LOG_LINK
34282 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
34284 * combine.c (try_combine): Delete redundant i1 test. Call
34285 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
34287 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34290 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
34291 alternatives 13/14.
34293 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34295 PR tree-optimization/80054
34296 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
34297 the optimization if a PHI or any of its arguments is not dominated
34298 by the candidate's basis. Use gphi* rather than gimple* as
34300 (replace_profitable_candidates): Clean up a gimple* variable that
34301 should be a gphi* variable.
34303 2017-03-20 Martin Sebor <msebor@redhat.com>
34306 * doc/extend.texi (attribute constructor): Document present limitation.
34308 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
34311 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
34312 __POWER9_VECTOR__ #ifdef control, change template definition to
34313 use Power9-specific built-in function.
34314 (vec_any_eq): Likewise.
34315 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
34316 to control outcomes from this test.
34317 (vector_ae_<mode>p): For VEC_F modes, likewise.
34319 2017-03-20 Ian Lance Taylor <iant@google.com>
34321 * config/i386/i386.c (ix86_function_regparm): Save an extra
34322 register for -fsplit-stack with DECL_STATIC_CHAIN.
34324 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
34327 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
34328 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
34330 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
34332 * config/riscv/riscv.c (riscv_print_operand): Use "fence
34334 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
34337 2017-03-20 Marek Polacek <polacek@redhat.com>
34340 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
34342 2017-03-20 Richard Biener <rguenther@suse.de>
34344 PR tree-optimization/80113
34345 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
34346 allocate extra SSA name for PHI def.
34347 (add_close_phis_to_outer_loops): Likewise.
34348 (add_close_phis_to_merge_points): Likewise.
34349 (copy_loop_close_phi_args): Likewise.
34350 (copy_cond_phi_nodes): Likewise.
34352 2017-03-20 Martin Liska <mliska@suse.cz>
34354 PR middle-end/79753
34355 * tree-chkp.c (chkp_build_returned_bound): Do not build
34356 returned bounds for a LHS that's not a BOUNDED_P type.
34358 2017-03-20 Martin Liska <mliska@suse.cz>
34362 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
34363 COMPLEX_CST and VECTOR_CST.
34365 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34368 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
34369 target operand. A new splitter adds the clobber statement in case
34370 the target operand is dead anyway.
34372 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
34374 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
34375 to age-old versions of binutils and glibc.
34377 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
34379 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
34381 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
34383 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
34385 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
34387 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
34388 requirement for binutils 2.13.
34390 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
34392 * combine.c (try_combine): Delete redundant i1 test. Call
34393 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
34395 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
34397 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
34398 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
34400 <riscv64-*-elf>: Re-arrange section
34401 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
34402 <riscv32-*-linux>: Likewise.
34403 <riscv64-*-elf>: Likewise
34404 <riscv64-*-linux>: Likewise.
34406 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
34409 * aarch64.opt(verbose-cost-dump): Fix typo.
34411 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
34414 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
34415 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
34417 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
34419 * reload.c (find_reloads): When reloading a nonoffsettable address,
34420 use RELOAD_OTHER for it and its address reloads.
34422 PR rtl-optimization/79910
34423 * combine.c (record_used_regs): New static function.
34424 (try_combine): Handle situations where there is an additional
34425 instruction between I2 and I3 which needs to have a LOG_LINK
34428 2017-03-17 Jeff Law <law@redhat.com>
34430 PR tree-optimization/71437
34431 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
34432 conditional in the hash table first.
34433 (vrp_dom_walker::before_dom_children): Extract condition from
34434 ASSERT_EXPR. Record condition, its inverion and any implied
34435 conditions as well.
34437 2017-03-17 Marek Polacek <polacek@redhat.com>
34438 Markus Trippelsdorf <markus@trippelsdorf.de>
34440 PR tree-optimization/80079
34441 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
34444 2017-03-17 Richard Biener <rguenther@suse.de>
34446 PR middle-end/80075
34447 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
34448 Properly verify the LHS before the RHS possibly claims to be
34450 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
34453 2017-03-17 Martin Jambor <mjambor@suse.cz>
34455 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
34456 (List of -O2 options): Likewise.
34457 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
34460 2017-03-17 Tom de Vries <tom@codesourcery.com>
34462 * gcov-dump.c (print_usage): Print bug_report_url.
34464 2017-03-17 Richard Biener <rguenther@suse.de>
34466 PR middle-end/80050
34467 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
34468 (parser::peek): Likewise.
34470 2017-03-17 Richard Biener <rguenther@suse.de>
34472 PR tree-optimization/80048
34473 * sese.c (free_sese_info): Properly release rename_map and
34474 copied_bb_map elements.
34476 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
34478 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
34479 Add linked-list forward and backlinks. Insert on
34480 construction, remove on destruction.
34481 (class pass_store_merging): Add m_stores_head field.
34482 (pass_store_merging::terminate_and_process_all_chains):
34483 Iterate over m_stores_head list.
34484 (pass_store_merging::terminate_all_aliasing_chains):
34486 (pass_store_merging::execute): Check for debug stmts first.
34487 Push new chains onto the m_stores_head stack.
34489 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
34492 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
34493 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
34494 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
34496 2017-03-16 Jeff Law <law@redhat.com>
34498 PR tree-optimization/71437
34499 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
34500 member function. Implementation moved into after_dom_children
34501 member function and into the threader's thread_outgoing_edges
34503 (dom_opt_dom_walker::after_dom_children): Simplify by moving
34504 some code into new thread_outgoing_edges.
34505 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
34506 definition. Simplify marker handling (do it here). Assume we always
34507 have the available expression and the const/copies tables.
34508 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
34510 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
34511 * tree-vrp.c (equiv_stack): No longer file scoped.
34512 (vrp_dom_walker): New class.
34513 (vrp_dom_walker::before_dom_children): New member function.
34514 (vrp_dom_walker::after_dom_children): Likewise.
34515 (identify_jump_threads): Setup domwalker. Use it rather than
34516 walking edges in a random order by hand. Simplify setup/finalization.
34517 (finalize_jump_threads): Remove.
34518 (vrp_finalize): Do not call identify_jump_threads here.
34519 (execute_vrp): Do it here instead and call thread_through_all_blocks
34522 PR tree-optimization/71437
34523 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
34525 (simplify_stmt_for_jump_threading): Add basic_block argument. All
34527 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
34528 (dom_opt_dom_walker::thread_across_edge): Remove
34529 handle_dominating_asserts argument. All callers changed.
34530 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
34531 changes. Remove calls to lhs_of_dominating_assert. Other
34532 uses of handle_dominating_asserts turn into unconditional code
34533 (simplify_control_stmt_condition_1): Likewise.
34534 (simplify_control_stmt_condition): Likewise.
34535 (thread_through_normal_block, thread_across_edge): Likewise.
34536 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
34537 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
34538 object if it is not an SSA_NAME.
34539 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
34540 before calling into the VRP specific simplifiers.
34541 (identify_jump_threads): Remove handle_dominating_asserts
34544 2017-03-16 Jakub Jelinek <jakub@redhat.com>
34547 * tree-diagnostic.c (default_tree_printer): No longer static.
34548 * tree-diagnostic.h (default_tree_printer): New prototype.
34550 2017-03-16 Tamar Christina <tamar.christina@arm.com>
34552 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
34553 Change ins into fmov.
34555 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34557 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
34558 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
34559 Use h_con constraint for operand 1.
34560 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
34561 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
34563 2017-03-15 Jeff Law <law@redhat.com>
34565 PR tree-optimization/71437
34566 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
34567 (record_temporary_equivalences): Use it.
34569 PR tree-optimization/71437
34570 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
34571 tree-ssa-scopedtables.
34572 (lookup_avail_expr, build_and_record_new_cond): Likewise.
34573 (record_conditions, record_cond, vuse_eq): Likewise.
34574 (record_edge_info): Adjust to API tweak of record_conditions.
34575 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
34576 (record_temporary_equivalences, optimize_stmt): Likewise.
34577 (eliminate_redundant_computations): Likewise.
34578 (record_equivalences_from_stmt): Likewise.
34579 * tree-ssa-scopedtables.c: Include options.h and params.h.
34580 (vuse_eq): New function, moved from tree-ssa-dom.c
34581 (build_and_record_new_cond): Likewise.
34582 (record_conditions): Likewise. Accept vector of conditions rather
34583 than edge_equivalence structure for first argument.
34584 for the first argument.
34585 (avail_exprs_stack::lookup_avail_expr): New member function, moved
34586 from tree-ssa-dom.c.
34587 (avail_exprs_stack::record_cond): Likewise.
34588 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
34589 from tree-ssa-dom.c.
34590 (avail_exprs_stack): Add new member functions lookup_avail_expr
34592 (record_conditions): Declare.
34594 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
34597 * lra-constraints.c (process_alt_operands): Increase reject for
34598 reloading an input/output operand.
34600 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
34603 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
34604 insns to convert from signed/unsigned char/short to IEEE 128-bit
34606 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
34608 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
34611 * config/i386/i386.c (ix86_vector_duplicate_value): Create
34612 subreg of inner mode for values already in registers.
34614 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
34616 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
34617 iteration reg is used after the loop.
34619 2017-03-14 Martin Sebor <msebor@redhat.com>
34621 PR tree-optimization/79800
34622 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
34623 precision in negative-positive range.
34624 (format_floating): Call non-const overload with adjusted precision.
34626 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
34629 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
34632 2017-03-14 Martin Sebor <msebor@redhat.com>
34634 PR middle-end/80020
34635 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
34636 * builtins.def (aligned_alloc): Use it.
34639 * Makefile.in (GTFILES): Add calls.c.
34640 * calls.c: Include "gt-calls.h".
34642 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
34644 PR rtl-optimization/79728
34645 * regs.h (struct target_regs): New field
34646 x_contains_allocatable_regs_of_mode.
34647 (contains_allocatable_regs_of_mode): New macro.
34648 * reginfo.c (init_reg_sets_1): Initialize it, and change
34649 contains_reg_of_mode so it includes global regs as well.
34650 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
34651 rather than contains_regs_of_mode.
34653 2017-03-14 Martin Liska <mliska@suse.cz>
34655 * doc/invoke.texi: Document options that can't be combined with
34656 -fcheck-pointer-bounds.
34658 2017-03-14 Martin Liska <mliska@suse.cz>
34660 PR middle-end/79831
34661 * doc/invoke.texi (-Wchkp): Document the option.
34663 2017-03-14 Martin Liska <mliska@suse.cz>
34665 * Makefile.in: Install gcov-dump.
34667 2017-03-14 Martin Liska <mliska@suse.cz>
34669 * multiple_target.c (expand_target_clones): Bail out for
34670 an invalid attribute.
34672 2017-03-14 Richard Biener <rguenther@suse.de>
34674 * alias.c (struct alias_set_entry): Pack properly.
34675 * cfgloop.h (struct loop): Likewise.
34676 * cse.c (struct set): Likewise.
34677 * ipa-utils.c (struct searchc_env): Likewise.
34678 * loop-invariant.c (struct invariant): Likewise.
34679 * lra-remat.c (struct cand): Likewise.
34680 * recog.c (struct change_t): Likewise.
34681 * rtl.h (struct address_info): Likewise.
34682 * symbol-summary.h (function_summary): Likewise.
34683 * tree-loop-distribution.c (struct partition): Likewise.
34684 * tree-object-size.c (struct object_size_info): Likewise.
34685 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
34686 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
34687 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
34688 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
34689 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
34690 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
34691 (struct _stmt_vec_info): Likewise.
34693 2017-03-14 Martin Liska <mliska@suse.cz>
34696 * multiple_target.c (create_dispatcher_calls): Check that
34697 a target can create a function dispatcher.
34699 2017-03-14 Martin Liska <mliska@suse.cz>
34702 * multiple_target.c (expand_target_clones): Drop local.local
34703 flag for default implementation.
34705 2017-03-14 Richard Biener <rguenther@suse.de>
34707 PR tree-optimization/80030
34708 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
34710 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
34712 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
34713 gcc_fallthrough() instead of __attribute__((fallthrough));
34715 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
34717 * doc/gcc.texi: Remove "up" link to (DIR).
34718 * doc/gccint.texi: Ditto.
34720 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
34722 * doc/install.texi (Specific) <avr>: Remove reference to
34725 2017-03-13 Jeff Law <law@redhat.com>
34727 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
34728 attribute rather than comments.
34730 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
34731 match_scratch operand is highest.
34733 2017-03-13 Martin Liska <mliska@suse.cz>
34735 PR middle-end/78339
34736 * ipa-pure-const.c (warn_function_noreturn): If the declarations
34737 is a CHKP clone, use original declaration.
34739 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
34741 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
34742 (arc_conditional_register_usage): Use a different allocation order
34743 when optimizing for size.
34744 * common/config/arc/arc-common.c (arc_option_optimization_table):
34745 Section anchors default on when optimizing for size.
34747 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
34749 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
34751 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
34753 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
34754 * config/arc/arc.md (cpu_facility): Add cd variant.
34755 (*movqi_insn): Add code density variant.
34756 (*movhi_insn): Likewise.
34757 (*movqi_insn): Likewise.
34758 (*addsi3_mixed): Likewise.
34759 (subsi3_insn): Likewise.
34761 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
34763 * config/arc/arc.md (movsi_cond_exec): Update constraint.
34765 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
34767 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
34768 expressions with MINUS and UNARY ops.
34770 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34773 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
34775 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
34776 between vec_select and vector argument.
34777 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
34778 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
34779 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
34780 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
34781 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
34782 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
34784 2017-03-13 Richard Biener <rguenther@suse.de>
34787 * params.def (vect-max-peeling-for-alignment): Fix typo.
34789 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
34791 * doc/install.texi (Specific) <mips-*-*>: Remove description of
34792 issue that only occurred with binutils below 2.18.
34794 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
34796 * doc/install.texi (Specific) <cris-axis-elf>: No longer
34797 refer to binutils 2.11/2.12 minimum.
34799 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
34801 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
34802 ftp.kernel.org and simplify binutils requirement.
34804 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
34806 * doc/invoke.texi (Warning Options): Fix spelling of link-time
34808 (Optimize Options): Ditto. Also remove redundancy.
34810 2017-03-10 David Malcolm <dmalcolm@redhat.com>
34812 PR translation/79848
34813 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
34815 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
34816 to G_ to avoid double translation.
34818 2017-03-10 David Malcolm <dmalcolm@redhat.com>
34820 PR translation/79923
34821 * auto-profile.c (get_combined_location): Convert leading
34822 character of diagnostics to lower case and remove trailing period.
34823 (read_profile): Likewise for various diagnostics.
34824 * config/arm/arm.c (arm_option_override): Remove trailing period
34825 from various diagnostics.
34826 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
34827 (msp430_expand_delay_cycles): Likewise.
34829 2017-03-10 David Malcolm <dmalcolm@redhat.com>
34832 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
34833 full command-line argument, rather than just "str".
34834 (aarch64_validate_march): Likewise.
34835 (aarch64_validate_mtune): Likewise.
34837 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
34839 PR rtl-optimization/78911
34840 * lra-assigns.c (must_not_spill_p): New function.
34841 (spill_for): Use it.
34843 2017-03-10 Jakub Jelinek <jakub@redhat.com>
34845 PR tree-optimization/79981
34846 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
34847 ATOMIC_COMPARE_EXCHANGE ifn result.
34848 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
34849 IFN_ATOMIC_COMPARE_EXCHANGE.
34851 2017-03-10 David Malcolm <dmalcolm@redhat.com>
34854 * opts.c (parse_sanitizer_options): Add missing question mark to
34855 "did you mean" message.
34857 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34859 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
34861 (VMULEUH_UNS): Likewise.
34862 (VMULOUB_UNS): Likewise.
34863 (VMULOUH_UNS): Likewise.
34864 * config/rs6000/rs6000.c (builtin_function_type): Remove
34865 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
34867 2017-03-10 David Malcolm <dmalcolm@redhat.com>
34870 * read-rtl-function.c (function_reader::read_rtx_operand): Update
34871 x with result of extra_parsing_for_operand_code_0.
34872 (function_reader::extra_parsing_for_operand_code_0): Convert
34873 return type from void to rtx, returning x. When reading
34874 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
34875 larger size containing struct block_symbol.
34877 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
34879 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
34880 -mfloat128-hardware without -m64.
34882 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
34885 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
34886 entries to the case statement that marks unsigned arguments to
34887 overloaded functions.
34889 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
34891 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
34892 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
34894 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
34897 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
34898 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
34900 2017-03-10 Martin Liska <mliska@suse.cz>
34904 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
34905 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
34908 2017-03-10 Olivier Hainque <hainque@adacore.com>
34910 * tree-switch-conversion (array_value_type): Start by resetting
34911 candidate type to it's main variant.
34913 2017-03-10 Jakub Jelinek <jakub@redhat.com>
34915 PR rtl-optimization/79909
34916 * combine.c (try_combine): Use simplify_replace_rtx on individual
34917 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
34918 of the whole CALL_INSN_FUNCTION_USAGE.
34920 PR tree-optimization/79972
34921 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
34922 get_range_info on SSA_NAMEs. Formatting fixes.
34924 2017-03-10 Richard Biener <rguenther@suse.de>
34925 Jakub Jelinek <jakub@redhat.com>
34927 PR tree-optimization/77975
34928 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
34929 edge to be constant.
34930 (get_val_for): For constant x return it. Formatting fix.
34931 (loop_niter_by_eval): Avoid pointless looping if the next iteration
34932 would use the same bases as the current one.
34934 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34936 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
34937 instead of vec_select for V1TImode.
34938 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
34940 (VSX_LE_128): Add V1TI to this mode iterator.
34941 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
34942 (*vsx_le_perm_store_<mode>): Likewise.
34943 (pre-reload splitter for VSX stores): Likewise.
34944 (post-reload splitter for VSX stores): Likewise.
34945 (*vsx_xxpermdi2_le_<mode>): Likewise.
34946 (*vsx_lxvd2x2_le_<mode>): Likewise.
34947 (*vsx_stxvd2x2_le_<mode>): Likewise.
34949 2017-03-09 Michael Eager <eager@eagercon.com>
34951 Correct failures with --enable-checking=yes,rtl.
34953 * config/microblaze/microblaze.c (microblaze_expand_shift):
34954 Replace GET_CODE test with CONST_INT_P and INTVAL test with
34955 test for const0_rtx.
34956 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
34957 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
34959 2017-03-09 Richard Biener <rguenther@suse.de>
34961 PR tree-optimization/79977
34962 * graphite-scop-detection.c (scop_detection::merge_sese):
34963 Handle the case of extra exits to blocks dominating the entry.
34965 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
34967 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
34970 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
34972 PR rtl-optimization/79949
34973 * lra-constraints.c (process_alt_operands): Check memory when
34974 trying to predict a cycle. Print about the overall increase.
34976 2017-03-09 Richard Biener <rguenther@suse.de>
34978 PR middle-end/79971
34979 * gimple-expr.c (useless_type_conversion_p): Preserve
34980 TYPE_SATURATING for fixed-point types.
34982 2017-03-09 Richard Biener <rguenther@suse.de>
34985 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
34986 alignment of BLKmode params.
34988 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34991 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
34992 (VALL_NO_V2Q): Likewise.
34994 * config/aarch64/aarch64-simd.md
34995 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
34997 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
34998 VALL_NO_V2Q mode iterator.
34999 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
35001 2017-03-09 Martin Liska <mliska@suse.cz>
35003 PR tree-optimization/79631
35004 * tree-chkp-opt.c (chkp_is_constant_addr): Call
35005 tree_int_cst_sign_bit just for INTEGER constants.
35007 2017-03-09 Martin Liska <mliska@suse.cz>
35011 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
35014 2017-03-09 Marek Polacek <polacek@redhat.com>
35017 * tree.c (inchash::add_expr): Handle TREE_VEC.
35019 2017-03-09 Martin Liska <mliska@suse.cz>
35022 (chkp_narrow_size_and_offset): New function.
35023 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
35024 (void chkp_parse_bit_field_ref): New function.
35025 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
35026 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
35028 2017-03-09 Martin Liska <mliska@suse.cz>
35031 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
35032 (chkp_find_bounds_1): Remove gcc_unreachable.
35034 2017-03-09 Jakub Jelinek <jakub@redhat.com>
35037 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
35038 BUILT_IN_SYNC*, determine the access type from the size suffix and
35039 always build a MEM_REF with that type. Handle forgotten
35040 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
35043 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
35044 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
35045 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
35046 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
35047 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
35048 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
35049 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
35050 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
35051 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
35052 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
35053 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
35054 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
35055 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
35056 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
35057 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
35058 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
35059 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
35060 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
35061 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
35062 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
35063 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
35064 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
35065 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
35066 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
35067 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
35068 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
35069 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
35070 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
35071 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
35072 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
35073 definitions outside of __OPTIMIZE__ guarded section.
35076 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
35077 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
35078 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
35079 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
35082 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35084 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
35085 ("vfenez<mode>"): Add missing constraints.
35087 2017-03-08 Martin Sebor <msebor@redhat.com>
35090 * config/nds32/nds32.c (nds32_option_override):
35091 Fix misspelled diagnostic.
35093 2017-03-08 Jakub Jelinek <jakub@redhat.com>
35096 * gimplify.c (gimplify_omp_for): Replace index var in outer
35097 taskloop statement with an artificial variable and add
35098 OMP_CLAUSE_PRIVATE clause for it.
35100 2017-03-08 Richard Biener <rguenther@suse.de>
35102 PR tree-optimization/79955
35103 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
35104 for accesses that are completely outside of the variable.
35106 2017-03-08 Andrew Haley <aph@redhat.com>
35108 PR tree-optimization/79943
35109 * tree-ssa-loop-split.c (compute_new_first_bound): When
35110 calculating the new upper bound, (END-BEG) should be added, not
35113 2017-03-08 Jakub Jelinek <jakub@redhat.com>
35115 * config/avr/avr.md (setmemhi): Make sure match_dup
35116 operand number comes before match_scratch.
35118 2017-03-08 Richard Biener <rguenther@suse.de>
35120 PR tree-optimization/79920
35121 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
35122 with ncopies == 1 to ...
35123 (vect_transform_slp_perm_load): ... here. Properly compute
35124 all element loads by iterating VF times over the group. Do
35125 not handle ncopies (computed in a broken way) in
35126 vect_create_mask_and_perm.
35128 2017-03-08 Jakub Jelinek <jakub@redhat.com>
35131 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
35132 is a uniform vector, use uniform_vector_p return value instead of
35133 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
35135 2017-03-07 Marek Polacek <polacek@redhat.com>
35137 PR middle-end/79809
35138 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
35139 (alloca_call_type): Likewise.
35141 2017-03-07 Martin Liska <mliska@suse.cz>
35143 * gcov.c (process_args): Put comment to correct location.
35145 2017-03-07 Martin Liska <mliska@suse.cz>
35147 PR middle-end/68270
35148 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
35149 Use array_at_struct_end_p instead of DECL_CHAIN (field).
35150 (chkp_narrow_bounds_for_field): Likewise.
35151 (chkp_parse_array_and_component_ref): Pass one more argument to
35154 2017-03-07 Richard Biener <rguenther@suse.de>
35156 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
35159 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
35161 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
35162 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
35164 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35167 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
35168 to end of description.
35169 (PARAM_MAX_STORES_TO_MERGE): Likewise.
35171 2017-03-07 Jakub Jelinek <jakub@redhat.com>
35173 PR rtl-optimization/79901
35174 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
35176 (*avx512f_<code><mode>3<mask_name>): ... this.
35177 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
35178 iterator instead of VI8_AVX2_AVX512BW.
35180 PR rtl-optimization/79901
35181 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
35182 min/max expander, expand it using expand_vec_cond_expr.
35185 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
35188 2017-03-06 Jakub Jelinek <jakub@redhat.com>
35191 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
35192 to void * for PCH reasons.
35193 * dwarf2out.c (output_loc_operands, output_die): Cast
35194 v.val_vec.array to unsigned char *.
35196 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
35199 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
35202 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
35204 PR rtl-optimization/79571
35205 * lra-constraints.c (process_alt_operands): Calculate static
35206 reject and subtract it from overall when only addresses will be
35209 2017-03-06 Julia Koval <julia.koval@intel.com>
35212 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
35213 incoming stack boundary to 128 for 64-bit targets.
35215 2017-03-06 Richard Biener <rguenther@suse.de>
35217 PR tree-optimization/79894
35218 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
35219 to NULL after folding it.
35221 2017-03-06 Richard Biener <rguenther@suse.de>
35223 PR tree-optimization/79824
35224 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
35225 check disabling peeling for gaps.
35227 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
35229 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
35230 attributes): Document gettimeofday.
35232 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
35234 * config/s390/s390.c (s390_option_override_internal): Set
35235 PARAM_MIN_VECT_LOOP_BOUND
35237 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
35239 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
35240 * config/s390/s390.md: Likewise.
35242 2017-03-06 Jakub Jelinek <jakub@redhat.com>
35245 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
35246 (<avx2_avx512>_perm<mode>): Rename to ...
35247 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
35249 (<avx512>_perm<mode>_mask): Rename to ...
35250 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
35252 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
35253 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
35254 instead of VI8F_256_512.
35255 (avx512f_perm<mode>): New define_expand.
35256 (avx512f_perm<mode>_mask): Likewise.
35257 (avx512f_perm<mode>_1<mask_name>): New define_insn.
35258 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
35260 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
35262 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
35263 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
35265 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
35267 2017-03-06 Martin Liska <mliska@suse.cz>
35270 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
35271 when having a SSA NAME w/o VAR_DECL assigned to it.
35273 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
35275 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
35276 msa_dpsub_<su>_d): Fix MODE for vec_select.
35278 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
35280 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
35282 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
35284 2017-03-06 Richard Biener <rguenther@suse.de>
35286 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
35287 * plugin.c (register_plugin_info): Likewise.
35288 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
35290 2017-03-05 Jakub Jelinek <jakub@redhat.com>
35292 * config/i386/sse.md (sse_storehps, sse_storelps,
35293 avx_<castmode><avxsizesuffix>_<castmode>,
35294 avx512f_<castmode><avxsizesuffix>_<castmode>,
35295 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
35296 in condition that at least one operand is not a MEM.
35298 2017-03-03 Jakub Jelinek <jakub@redhat.com>
35300 PR middle-end/79805
35301 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
35302 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
35304 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
35305 gimple_call_nothrow_p flag based on whether original builtin can throw.
35306 If it can, emit following stmts on the fallthrough edge.
35307 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
35308 don't create new bb if inserting just debug stmts on the edge, try to
35309 insert them on the fallthru bb or just reset debug stmts.
35311 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
35314 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
35315 restore recog_data (including the operand rtxes inside it) around
35316 the call to get_insn_template.
35318 2017-03-03 Martin Sebor <msebor@redhat.com>
35320 PR tree-optimization/79699
35321 * context.c (context::~context): Free MPFR caches to avoid
35322 a memory leak on program exit.
35324 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35326 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
35327 Use wide_int::ulow () instead of .elt (0).
35329 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
35331 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
35332 (*pushxf): Limit oF constraint to 32bit targets and add oC
35333 constraint for 64bit targets.
35334 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
35335 (*pushdf): Change rmF constraint to rmC.
35337 2017-03-03 Martin Liska <mliska@suse.cz>
35339 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
35340 Remove unused variable.
35342 2017-03-03 Jakub Jelinek <jakub@redhat.com>
35345 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
35346 is a memory operand, increase num_memory.
35347 (ix86_expand_args_builtin): Likewise.
35349 2017-03-03 Jan Hubicka <jh@suse.cz>
35352 * ipa-devirt.c (maybe_record_node): Properly handle
35353 __cxa_pure_virtual visibility.
35355 2017-03-03 Martin Liska <mliska@suse.cz>
35357 PR tree-optimization/79803
35358 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
35360 (pass_loop_prefetch::execute): Disabled optimization if an
35361 assumption about L1 cache size is not met.
35363 2017-03-03 Martin Liska <mliska@suse.cz>
35365 PR rtl-optimization/79574
35366 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
35367 (hash_scan_set): Likewise.
35368 (dump_hash_table): Likewise.
35369 (hoist_code): Likewise.
35371 2017-03-03 Richard Biener <rguenther@suse.de>
35373 * fixed-value.c (fixed_from_string): Restore use of elt (1)
35374 in place of uhigh ().
35375 (fixed_convert_from_real): Likewise.
35377 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
35380 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
35382 2017-03-03 Richard Biener <rguenther@suse.de>
35384 PR middle-end/79818
35385 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
35386 TYPE_OVERFLOW_UNDEFINED check.
35388 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35390 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
35392 (vector_ae_<mode>_p): Likewise.
35393 (vector_nez_<mode>_p): Likewise.
35394 (vector_ne_v2di_p): Likewise.
35395 (vector_ae_v2di_p): Likewise.
35396 (vector_ne_<mode>_p): Likewise.
35397 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
35399 (vsx_tsqrt<mode>2_fe): Likewise.
35401 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
35404 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
35406 2017-03-02 Jakub Jelinek <jakub@redhat.com>
35408 PR rtl-optimization/79780
35409 * cprop.c (one_cprop_pass): When second and further conditional trap
35410 in a single basic block is turned into an unconditional trap, turn it
35411 into a deleted note to avoid RTL verification failures.
35413 2017-03-02 Richard Biener <rguenther@suse.de>
35415 * fold-const.c (const_binop): Use ulow () instead of elt (0).
35417 2017-03-02 Richard Biener <rguenther@suse.de>
35419 PR tree-optimization/79345
35421 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
35422 param and abort the walk, returning -1 if it is hit.
35423 (walk_aliased_vdefs): Take a limit param and pass it on.
35424 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
35425 defaulting to 0 and return a signed int.
35426 * tree-ssa-uninit.c (struct check_defs_data): New struct.
35427 (check_defs): New helper.
35428 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
35429 about uninitialized memory.
35430 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
35431 bogus uninitialized warning.
35432 (fixed_convert_from_real): Likewise.
35434 2017-03-02 Bin Cheng <bin.cheng@arm.com>
35436 PR tree-optimization/66768
35437 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
35438 iv_use if base object can't be determined.
35440 2017-03-02 Jakub Jelinek <jakub@redhat.com>
35442 PR tree-optimization/79345
35443 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
35444 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
35445 (get_pattern_stats): Initialize it.
35446 * genemit.c (gen_expand): Verify match_scratch numbers come after
35447 match_operand/match_dup numbers.
35448 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
35449 match_scratch numbers.
35450 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
35452 * config/s390/s390.md (trunctdsd2): Likewise.
35454 2017-03-02 Richard Biener <rguenther@suse.de>
35456 * wide-int.h (wide_int_storage::operator=): Implement in terms
35459 2017-03-02 Richard Biener <rguenther@suse.de>
35461 PR tree-optimization/79777
35462 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
35463 the to insert expression to sth existing.
35465 2017-03-01 Martin Sebor <msebor@redhat.com>
35467 PR middle-end/79692
35468 * gimple-ssa-sprintf.c
35469 (directive::known_width_and_precision): New function.
35470 (format_integer): Use it.
35471 (get_mpfr_format_length): Consider the full range of precision
35472 when computing %g output with the # flag. Set the likely byte
35473 count to 3 rather than 1 when precision is indeterminate.
35474 (format_floating): Correct the lower bound of precision.
35476 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35478 * doc/invoke.texi: Document default code model for 64-bit Linux.
35480 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35483 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
35484 udiv rather than div since input pattern is unsigned.
35486 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
35488 * config/i386/i386.c (print_reg): Warn for values of
35489 unsupported size in integer register.
35491 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
35494 * config/rs6000/predicates.md (current_file_function_operand): Do
35495 not allow self calls to be local if the function is replaceable.
35497 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
35500 * config/rs6000/altivec.h (vec_ctz and others): Change the
35501 preprocessor macro that controls conditional compilation from
35502 _ARCH_PWR9 to __POWER9_VECTOR__.
35503 (vec_all_ne): Change parameterization of __altivec_scalar_pred
35504 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
35505 control (instead of _ARCH_PWR9 control) so that template
35506 definition uses power9-specific function.
35507 (vec_any_eq): Likewise.
35508 (vec_all_ne): Change macro definition to use a power9-specific
35509 expansion under #ifdef __POWER9_VECTOR__ control (instead of
35510 _ARCH_PWR9 control).
35511 (vec_any_eq) Likewise.
35512 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
35513 expansion for CMPNEF to remove support for xvcmpnesp instruction.
35514 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
35515 support for xvcmpnedp instruction.
35516 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
35517 macro expansion so that Power9 implementation of vec_all_ne does
35518 not use the AltiVec predicate framework.
35519 (VCMPNEH_P): Likewise.
35520 (VCMPNEW_P): Likewise.
35521 (VCMPNED_P): Likewise.
35522 (VCMPNEFP_P): Likewise.
35523 (VCMPNEDP_P): Likewise.
35524 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
35525 implementation of vec_any_eq to not use AltiVec predicate
35527 (VCMPAEH_P): Likewise.
35528 (VCMPAEW_P): Likewise.
35529 (VCMPAED_P): Likewise.
35530 (VCMPAEFP_P): Likewise.
35531 (VCMPAEDP_P): Likewise.
35532 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
35533 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
35534 not use the AltiVec predicate framework.
35535 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
35536 of vec_any_eq to not use AltiVec predicate framework.
35537 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
35538 support for predefined __POWER9_VECTOR__ macro to indicate that
35539 Power9 instruction selection is enabled.
35540 (altivec_overloaded_builtins): Remove extraneous
35541 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
35542 function argument types RS6000_BTI_bool_V16QI and
35543 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
35544 entry for overloaded function argument types RS6000_BTI_bool_V4SI
35545 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
35546 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
35547 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
35548 Power9 for implementations of vec_cmpne. Change the signature for
35549 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
35550 (representing vec_all_ne) to remove the previously described first
35551 argument of type RS6000_BTI_INTSI, as this was an artifact of
35552 reliance on the AltiVec predicate framework, which is no longer
35553 used in the implementation of these functions. Add
35554 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
35555 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
35556 since, unlike the AltiVec predicate framework implementation, we
35557 do not share function descriptors between vec_alle and vec_anyeq.
35558 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
35559 set of modes that receive special treatment even when
35560 TARGET_P9_VECTOR is true. The special treatment emits code that
35561 does not depend on Power9 instructions.
35562 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
35563 define_expand to not rely on AltiVec predicate framework.
35564 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
35566 (vector_ne_v2di_p): Change this define_expand to not rely on
35567 AltiVec predicate framework.
35568 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
35570 (vector_ne_<mode>_p): Change this define_expand to not rely on
35571 AltiVec predicate framework.
35572 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
35574 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
35575 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
35576 define_insn pattern.
35577 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
35578 define_insn pattern because the xvcmpne<VSs>. instruction is not
35580 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
35581 instruction is not supported.
35583 2017-03-01 Jakub Jelinek <jakub@redhat.com>
35585 * config/nvptx/nvptx.c: Include intl.h.
35587 2017-03-01 Martin Jambor <mjambor@suse.cz>
35590 * ipa-prop.h (ipa_bits): Removed field known.
35591 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
35592 to pointers. Adjusted their comments to warn about their sharing.
35593 (ipcp_transformation_summary): Change bits to a vector of pointers.
35594 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
35595 (ipa_get_ipa_bits_for_value): Declare.
35596 * tree-vrp.h (value_range): Mark as GTY((for_user)).
35597 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
35598 (ipa_bits_hash_table): Likewise.
35599 (ipa_vr_ggc_hash_traits): Likewise.
35600 (ipa_vr_hash_table): Likewise.
35601 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
35602 being pointers and vr_known being removed.
35603 (ipa_set_jf_unknown): Likewise.
35604 (ipa_get_ipa_bits_for_value): New function.
35605 (ipa_set_jfunc_bits): Likewise.
35606 (ipa_get_value_range): New overloaded functions.
35607 (ipa_set_jfunc_vr): Likewise.
35608 (ipa_compute_jump_functions_for_edge): Use the above functions to
35609 construct bits and vr parts of jump functions.
35610 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
35611 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
35613 (ipcp_grow_transformations_if_necessary): Also allocate
35614 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
35616 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
35617 them. Fix too long lines.
35618 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
35619 vr_known being removed.
35620 (ipa_read_jump_function): Use new setter functions to construct bits
35621 and vr parts of jump functions or set them to NULL.
35622 (write_ipcp_transformation_info): Adjust for bits being pointers.
35623 (read_ipcp_transformation_info): Likewise.
35624 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
35626 Include gt-ipa-prop.h.
35627 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
35629 (ipcp_store_bits_results): Likewise.
35630 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
35631 Do not write to existing jump functions but use a temporary instead.
35633 2017-03-01 Jakub Jelinek <jakub@redhat.com>
35636 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
35637 attempt to use its first operand as BIT_FIELD_REF base.
35639 2017-03-01 Richard Biener <rguenther@suse.de>
35641 PR middle-end/79721
35642 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
35643 interpolating formula in wrapping arithmetic.
35644 (chrec_apply): Convert chrec_evaluate return value to wanted type.
35646 2017-03-01 Jakub Jelinek <jakub@redhat.com>
35648 PR tree-optimization/79734
35649 * tree-vect-generic.c (expand_vector_condition): Optimize
35650 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
35651 Handle VEC_COND_EXPR where comparison has different inner width from
35652 type's inner width.
35654 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
35656 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
35657 markup, and similar issues. Remove @opindex entries for things
35658 that aren't options. Add missing -mmpy-option entries.
35660 2017-02-28 Jakub Jelinek <jakub@redhat.com>
35662 PR tree-optimization/79737
35663 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
35664 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
35665 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
35666 instead of byte_size. Formatting fix.
35667 (shift_bytes_in_array_right): Formatting fix.
35669 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
35672 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
35673 condition on optimize for the leaf function test.
35675 2017-02-28 Martin Liska <mliska@suse.cz>
35678 * read-rtl-function.c (function_reader::handle_unknown_directive):
35679 Bail out when one uses -flto.
35681 2017-02-28 Martin Liska <mliska@suse.cz>
35683 * common.opt: Replace space with tabular for options of <number>
35685 * config/i386/i386.opt: Show <number> value for
35686 -mlarge-data-threshold.
35687 * opts.c (print_filtered_help): Do not display number in hexadecimal
35690 2017-02-28 Martin Liska <mliska@suse.cz>
35692 * common.opt: Fix --help=option -Q for options which are of
35695 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
35697 * config/i386/i386.c (print_reg): Error out for values
35698 of 8-bit size in invalid integer register.
35700 2017-02-28 Martin Sebor <msebor@redhat.com>
35702 PR tree-optimization/79691
35703 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
35705 2017-02-28 Jakub Jelinek <jakub@redhat.com>
35708 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
35709 gcc_unreachable with output_operand_lossage.
35711 2017-02-28 Richard Biener <rguenther@suse.de>
35713 PR tree-optimization/79740
35714 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
35716 (visit_nary_op): Insert the nary into the hashtable if we
35717 pattern-matched sth.
35718 * tree-ssa-pre.c (eliminate_insert): Robustify.
35720 2017-02-28 Richard Biener <rguenther@suse.de>
35722 PR middle-end/79731
35723 * fold-const.c (decode_field_reference): Reject out-of-bound
35726 2017-02-28 Jakub Jelinek <jakub@redhat.com>
35728 * config/i386/i386.c: Include intl.h.
35729 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
35730 instead of just cond ? "..." : "...".
35731 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
35732 * coverage.c (read_counts_file): Likewise.
35733 * omp-offload.c: Include intl.h.
35734 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
35735 of just cond ? "..." : "...".
35736 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
35737 of just cond ? "..." : "...".
35739 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
35742 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
35744 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
35745 'tune for' CPU name.
35746 * config/arm/arm-cpu-data.h: Regenerated.
35748 2017-02-28 Richard Biener <rguenther@suse.de>
35750 PR tree-optimization/79732
35751 * tree-inline.c (expand_call_inline): Do not shadow var.
35753 2017-02-28 Richard Biener <rguenther@suse.de>
35755 PR tree-optimization/79723
35756 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
35757 address-space properly.
35759 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
35761 * doc/optinfo.texi (Optimization groups): Fix option used for
35763 * doc/invoke.texi (-fopt-info): Document "omp".
35764 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
35765 (OPTGROUP_ALL): Add OPTGROUP_OMP.
35766 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
35767 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
35768 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
35770 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
35772 * dumpfile.c (optgroup_options): Instead of "openmp", associate
35773 OPTGROUP_OMP with "omp".
35775 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
35778 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
35779 for arithmetic shift of unsigned V2DI.
35781 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
35783 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
35784 arc/linux.h headers.
35785 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
35786 (LINK_SPEC): Likewise.
35787 (ARC_TLS_EXTRA_START_SPEC): Likewise.
35788 (EXTRA_SPECS): Likewise.
35789 (STARTFILE_SPEC): Likewise.
35790 (ENDFILE_SPEC): Likewise.
35791 (LIB_SPEC): Likewise.
35792 (TARGET_SDATA_DEFAULT): Likewise.
35793 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
35794 (MULTILIB_DEFAULTS): Likewise.
35795 (DWARF2_UNWIND_INFO): Likewise.
35796 * config/arc/big.h: New file.
35797 * config/arc/elf.h: Likewise.
35798 * config/arc/linux.h: Likewise.
35799 * config/arc/t-uClibc: Remove.
35801 2017-02-27 Bin Cheng <bin.cheng@arm.com>
35803 PR tree-optimization/77536
35804 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
35805 (tree_transform_and_unroll_loop): Use above function to compute the
35806 estimated niter of unrolled loop and use it when scaling profile.
35807 Also use count info rather than frequency if it's non-zero.
35808 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
35809 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
35810 (vect_transform_loop): Call above function.
35812 2017-02-27 Richard Biener <rguenther@suse.de>
35814 PR tree-optimization/45397
35815 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
35816 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
35817 (visit_nary_op): Add pattern matching for CSEing sign-changed
35818 or truncated operations with wider ones.
35820 2017-02-27 Richard Biener <rguenther@suse.de>
35822 PR tree-optimization/79690
35823 * tree-vect-stmts.c (vectorizable_store): Use vector type
35824 built from the DR with address-space.
35826 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
35828 * doc/invoke.texi (Optimize Options): Refine the description
35829 of asan-use-after-return.
35831 2017-02-25 Alan Modra <amodra@gmail.com>
35833 PR rtl-optimization/79584
35834 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
35835 base, not ad->base_term, the reg within base. Remove assertion
35836 that ad->base == ad->base_term. Replace gen_int_mode using
35837 bogus mode with const0_rtx.
35839 2017-02-25 Jakub Jelinek <jakub@redhat.com>
35841 PR middle-end/79396
35842 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
35843 FMA_EXPR like tcc_binary or tcc_unary.
35845 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
35848 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
35850 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
35851 (output_loc_operands): Handle DW_OP_call_ref and
35852 DW_OP_GNU_variable_value.
35853 (struct variable_value_struct): New type.
35854 (struct variable_value_hasher): Likewise.
35855 (variable_value_hash): New variable.
35856 (string_types): Remove.
35857 (copy_loc_descr): New function.
35858 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
35859 (prepend_loc_descr_to_each): New function.
35860 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
35861 instead of add_loc_descr_to_each if the first argument is single
35862 location list and the second has multiple.
35863 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
35864 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
35865 when looking for variable value which doesn't have other location info.
35866 (loc_list_from_tree): Formatting fix.
35867 (gen_array_type_die): Simplify DW_AT_string_length handling.
35868 (adjust_string_types): Remove.
35869 (gen_subprogram_die): Don't call adjust_string_types nor test/set
35870 string_types. Call resolve_variable_values.
35871 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
35872 (resolve_addr_in_expr): Likewise. Add A argument.
35873 (copy_deref_exprloc): Remove deref argument. Adjust for the
35874 original expression being DW_OP_GNU_variable_value with optionally
35875 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
35876 optionally after it.
35877 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
35878 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
35879 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
35880 (variable_value_hasher::hash, variable_value_hasher::equal): New
35882 (resolve_variable_value_in_expr, resolve_variable_value,
35883 resolve_variable_values, note_variable_value_in_expr,
35884 note_variable_value): New functions.
35885 (dwarf2out_early_finish): Call note_variable_value on all toplevel
35888 2017-02-24 Jakub Jelinek <jakub@redhat.com>
35891 * opts.h (handle_generated_option): Add GENERATED_P argument.
35892 * opts-common.c (handle_option): Adjust function comment.
35893 (handle_generated_option): Add GENERATED_P argument, pass it to
35895 (control_warning_option): Pass false to handle_generated_option
35897 * opts.c (maybe_default_option): Pass true to handle_generated_option
35899 * optc-gen.awk: Likewise.
35901 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
35903 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
35904 a REG, look at the REG it is a SUBREG of.
35905 (splitter for cmpeqsi_t): Ditto.
35907 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
35909 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
35910 the special USEs with the pattern of the insn, not the insn itself.
35912 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
35915 * doc/invoke.texi: Document -mload-store-pairs.
35917 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
35918 Sandra Loosemore <sandra@codesourcery.com>
35920 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
35921 argument isn't a CONST_INT.
35922 (nios2_alternate_compare_const): Assert op is a CONST_INT.
35923 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
35924 (nios2_validate_compare): Bypass alternate compare logic if *op2
35925 is not a CONST_INT.
35926 (ldstwm_operation_p): Return false if first_base is not a REG or
35927 if first_offset is not a CONST_INT.
35929 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
35931 * config/cris/cris.md: Use correct operand in a define_peephole2.
35933 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
35935 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
35937 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
35939 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
35940 this_insn if it is an INSN or JUMP_INSN.
35941 (force_offsettable): Look at base, not at addr.
35942 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
35943 on things that aren't necessarily CONST_INTs.
35945 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
35947 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
35948 -mfpmath=sse is the default also for x86-32 targets with SSE2
35949 instruction set when @option{-ffast-math} is enabled
35951 2017-02-24 Jeff Law <law@redhat.com>
35953 PR rtl-optimizatoin/79286
35954 * ira.c (update_equiv_regs): Drop may_trap_p exception to
35957 2017-02-24 Richard Biener <rguenther@suse.de>
35959 PR tree-optimization/79389
35960 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
35963 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
35965 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
35966 function comment to reflect reality.
35967 (loop_exits_before_overflow): Fix typo in function description.
35969 2017-02-24 Richard Biener <rguenther@suse.de>
35971 PR tree-optimization/79389
35972 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
35973 properly that a threading opportunity exists. Detect conditional
35974 copy/constant propagation opportunities.
35976 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
35978 * config/visium/visium.md (type): Add trap.
35979 (b): New mode attribute.
35980 (*btst): Rename into...
35981 (*btst<mode>): ...this and adjust.
35982 (*cbranchsi4_btst_insn): Rename into...
35983 (*cbranch<mode>4_btst_insn): ...this and adjust.
35984 (trap): New define_insn.
35986 2017-02-23 Jakub Jelinek <jakub@redhat.com>
35988 PR tree-optimization/79389
35989 * ifcvt.c (struct noce_if_info): Add rev_cond field.
35990 (noce_reversed_cond_code): New function.
35991 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
35992 reversed_comparison_code. Formatting fix.
35993 (noce_try_store_flag): Test rev_cond != NULL in addition to
35994 reversed_comparison_code.
35995 (noce_try_store_flag_constants): Likewise.
35996 (noce_try_store_flag_mask): Likewise.
35997 (noce_try_addcc): Use rev_cond if non-NULL instead of
35998 reversed_comparison_code.
35999 (noce_try_cmove_arith): Likewise. Formatting fixes.
36000 (noce_try_minmax, noce_try_abs): Clear rev_cond.
36001 (noce_find_if_block): Initialize rev_cond.
36002 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
36003 instead of false as last argument never attempt to reverse it
36006 2017-02-23 Bin Cheng <bin.cheng@arm.com>
36008 PR tree-optimization/79663
36009 * tree-predcom.c (combine_chains): Process refs in reverse order
36010 only for ZERO length chains, and add explaining comment.
36012 2017-02-23 Jeff Law <law@redhat.com>
36014 PR tree-optimization/79578
36015 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
36016 in call to operand_equal_p.
36018 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
36021 * config/i386/cpuid.h: Fix another undefined behavior.
36023 2017-02-23 Richard Biener <rguenther@suse.de>
36025 PR tree-optimization/79683
36026 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
36027 vector types for data-refs.
36029 2017-02-23 Martin Liska <mliska@suse.cz>
36031 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
36033 2017-02-23 Jakub Jelinek <jakub@redhat.com>
36035 PR middle-end/79665
36036 * internal-fn.c (get_range_pos_neg): Moved to ...
36037 * tree.c (get_range_pos_neg): ... here. No longer static.
36038 * tree.h (get_range_pos_neg): New prototype.
36039 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
36040 are known to be in between 0 and signed maximum inclusive, try to
36041 expand both unsigned and signed divmod and use the cheaper one from
36044 2017-02-22 Jeff Law <law@redhat.com>
36046 PR tree-optimization/79578
36047 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
36048 to compare base operands.
36050 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
36053 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
36054 gpc_reg_operand instead of fpr_reg_operand.
36056 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
36058 * config/mips/mips.c (mips_return_in_memory): Force FP
36059 vector types to be returned in memory for o32 ABI.
36061 2017-02-22 Jakub Jelinek <jakub@redhat.com>
36063 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
36064 instead of DW_TAG_member for static data member declarations and don't
36065 set no_linkage_name for static inline data members.
36066 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
36069 2017-02-22 Martin Liska <mliska@suse.cz>
36071 * doc/invoke.texi: Replace inequality signs with square brackets
36074 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
36077 * lra-constraints.c (simplify_operand_subreg): Handle
36078 WORD_REGISTER_OPERATIONS targets.
36080 2017-02-22 Jakub Jelinek <jakub@redhat.com>
36083 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
36084 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
36085 elimination by swapping fld*.
36087 2017-02-22 Richard Biener <rguenther@suse.de>
36089 PR tree-optimization/79673
36090 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
36091 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
36092 irrelevant address-space qualifiers and avoiding a
36093 ADDR_SPACE_CONVERT_EXPR from fold_convert.
36095 2017-02-22 Richard Biener <rguenther@suse.de>
36097 PR tree-optimization/79666
36098 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
36099 to not symbolically negate if that may introduce undefined
36102 2017-02-22 Martin Liska <mliska@suse.cz>
36105 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
36106 * data-streamer-out.c (streamer_write_gcov_count_stream):
36108 * value-prof.c (stream_out_histogram_value): Make assert more
36109 precise based on type of counter.
36111 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
36114 * config/i386/i386.md (standard_x87sse_constant_load splitter):
36115 Use nonimmediate_operand instead of memory_operand for operand 1.
36116 (float-extend standard_x87sse_constant_load splitter): Ditto.
36118 2017-02-21 Jeff Law <law@redhat.com>
36120 PR tree-optimization/79621
36121 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
36122 blocks with edges to themselves.
36124 2017-02-21 Jakub Jelinek <jakub@redhat.com>
36127 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
36128 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
36129 Use gimple_call_builtin_p.
36132 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
36133 on temporarily removed DEBUG_INSNs.
36135 PR tree-optimization/79649
36136 * tree-loop-distribution.c (classify_partition): Give up on
36137 non-generic address space loads/stores.
36139 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
36141 * doc/loop.texi (Loop manipulation): Remove nonexistent
36142 tree_ssa_loop_version from the documentation.
36143 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
36145 2017-02-21 Jakub Jelinek <jakub@redhat.com>
36148 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
36149 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
36150 * config/rs6000/rs6000.c: Include except.h.
36151 (rs6000_expand_split_stack_prologue): Call
36152 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
36154 2017-02-21 Martin Jambor <mjambor@suse.cz>
36157 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
36158 have been analyzed.
36160 2017-02-21 Martin Jambor <mjambor@suse.cz>
36162 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
36163 for backward compatibility only.
36164 * doc/invoke.texi (Option Summary): Remove all references to
36165 -fipa-cp-alignment.
36167 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
36171 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36173 * lra-constraints.c (curr_insn_transform): Handle
36174 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
36176 2017-02-21 Martin Liska <mliska@suse.cz>
36178 * config/i386/i386.opt: Replace -masm-dialect with -masm.
36180 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
36182 PR translation/79638
36183 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
36185 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
36188 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
36189 (arm_function_ok_for_sibcall): Return false for an indirect call by
36190 descriptor if all the argument registers are used.
36191 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
36192 alignment of the function.
36194 2017-02-21 Jakub Jelinek <jakub@redhat.com>
36196 PR tree-optimization/61441
36197 * simplify-rtx.c (simplify_const_unary_operation): For
36198 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
36199 the sNaN unmodified.
36201 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
36203 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
36204 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
36205 instead of SYSTEM_HEADER_DIR.
36207 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
36208 Martin Liška <mliska@suse.cz>
36210 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
36211 Fix typos and grammar, use active voice, and clarify.
36213 2017-02-20 Marek Polacek <polacek@redhat.com>
36215 PR middle-end/79537
36216 * gimplify.c (gimplify_expr): Handle unused *&&L;.
36219 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
36221 2017-02-20 Jakub Jelinek <jakub@redhat.com>
36224 * config/i386/i386.c (ix86_expand_builtin): Handle
36225 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
36226 ix86_builtins_isa[fcode].isa as a requirement of those
36227 flags and any other flag in the bitmask.
36228 (ix86_init_mmx_sse_builtins): Use 0 instead of
36229 ~OPTION_MASK_ISA_64BIT as mask.
36230 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
36231 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
36232 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
36233 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
36235 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36238 * lra-constraints.c (split_reg): Check requested split mode
36239 is supported by the register.
36241 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36243 * lra-constraints.c (simplify_operand_subreg): Remove early
36246 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36249 * lra-constraints.c (curr_insn_transform): Tighten condition
36250 for converting SUBREG reloads from OP_OUT to OP_INOUT.
36252 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36255 * lra-constraints.c (curr_insn_transform): Handle
36256 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
36258 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
36261 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
36263 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
36265 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
36268 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
36271 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
36273 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
36276 2017-02-18 Jakub Jelinek <jakub@redhat.com>
36278 * final.c (last_columnnum, override_columnnum): New variables.
36279 (final_start_function): Set last_columnnum, pass it to begin_prologue
36280 hook and pass 0 to dwarf2out_begin_prologue.
36281 (final_scan_insn): Update override_columnnum. Pass last_columnnum
36282 to source_line debug hook.
36283 (notice_source_line): Compute last_columnnum and for debug_column_info
36284 return true on column changes.
36285 * debug.h (struct gcc_debug_hooks): Add column argument to
36286 source_line and begin_prologue hooks.
36287 (debug_nothing_int_charstar_int_bool): Remove prototype.
36288 (debug_nothing_int_int_charstar,
36289 debug_nothing_int_int_charstar_int_bool): New prototypes.
36290 (dwarf2out_begin_prologue): Add column argument.
36291 * debug.c (do_nothing_debug_hooks): Adjust source_line and
36292 begin_prologue hooks.
36293 (debug_nothing_int_charstar_int_bool): Remove.
36294 (debug_nothing_int_int_charstar,
36295 debug_nothing_int_int_charstar_int_bool): New functions.
36296 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
36297 through to dwarf2out_source_line.
36298 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
36299 (dwarf2out_source_line): Add column argument, emit it if requested.
36300 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
36302 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
36303 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
36304 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
36305 through to dwarf2out_begin_prologue.
36306 (vmsdbgout_source_line): Add column argument, pass it through to
36307 dwarf2out_source_line.
36308 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
36309 dbxout_source_line caller.
36310 (dbxout_source_line): Add column argument.
36312 * common.opt (gno-column-info, gcolumn-info): New options.
36313 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
36314 (check_die): Also test for multiple DW_AT_decl_column attributes.
36315 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
36316 DW_AT_decl_column if requested.
36317 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
36319 (gen_variable_die): Likewise.
36320 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
36321 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
36324 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
36325 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
36326 (ix86_handle_option): Handle OPT_m3dnowa.
36327 * doc/invoke.texi (-m3dnowa): Document.
36328 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
36329 -m3dnowa instead of -m3dnow -march=athlon.
36332 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
36333 instead of gcc_assert for K, r and R code checks. Formatting fixes.
36335 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36338 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
36339 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
36340 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
36341 generator for vsx_xxpermdi_<mode>_be.
36342 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
36343 force big-endian semantics.
36344 (vsx_xxpermdi_<mode>_be): New define_expand with same
36345 implementation as previous version of vsx_xxpermdi_<mode>.
36347 2017-02-17 Jakub Jelinek <jakub@redhat.com>
36349 PR tree-optimization/79327
36350 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
36351 variable, its initialization and use.
36353 2017-02-17 Julia Koval <julia.koval@intel.com>
36355 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
36356 (OPTION_MASK_ISA_PKU_UNSET): New.
36357 (ix86_handle_option): Handle -mrdpid.
36358 * config/i386/cpuid.h (bit_RDPID): New.
36359 * config/i386/driver-i386.c (host_detect_local_cpu):
36360 Detect RDPID feature.
36361 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
36362 * config/i386/i386-c.c (ix86_target_macros_internal):
36364 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
36365 (ix86_valid_target_attribute_inner_p): Add "rdpid".
36366 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
36367 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
36368 * config/i386/i386.md (define_insn "rdpid"): New.
36369 * config/i386/i386.opt Add -mrdpid.
36370 * config/i386/immintrin.h (_rdpid_u32): New.
36372 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
36374 PR rtl-optimization/79541
36375 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
36376 instead of transforming it into USE.
36378 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
36380 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
36381 If HONOR_SNANS (SFmode) force the input to a register.
36382 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
36383 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
36384 an frsp or similar insn.
36386 2017-02-17 Martin Liska <mliska@suse.cz>
36388 PR rtl-optimization/79577
36389 * params.def (selsched-max-sched-times): Increase minimum to 1.
36391 2017-02-17 Martin Liska <mliska@suse.cz>
36393 PR rtl-optimization/79574
36394 * gcse.c (want_to_gcse_p): Prevent integer overflow.
36396 2017-02-17 Martin Liska <mliska@suse.cz>
36398 PR tree-optimization/79529
36399 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
36400 ssa_defined_default_def_p to handle cases which are implicitly
36402 * tree-ssa.c (ssa_defined_default_def_p): New function.
36403 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
36404 which are implicitly defined.
36405 * tree-ssa.h (ssa_defined_default_def_p): Declare.
36407 2017-02-17 Richard Biener <rguenther@suse.de>
36409 PR middle-end/79576
36410 * params.def (max-ssa-name-query-depth): Limit to 10.
36412 2017-02-17 Richard Biener <rguenther@suse.de>
36414 PR tree-optimization/79552
36415 * tree-ssa-structalias.c (visit_loadstore): Properly verify
36418 2017-02-17 Richard Biener <rguenther@suse.de>
36421 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
36423 2017-02-17 Marek Polacek <polacek@redhat.com>
36425 PR middle-end/79536
36426 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
36427 (fold_negate_expr): New wrapper.
36429 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
36431 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
36432 Correct terminology and de-emphasize pre-standard behavior.
36434 2017-02-16 Alan Modra <amodra@gmail.com>
36436 PR rtl-optimization/79286
36437 * ira.c (def_dominates_uses): New function.
36438 (update_equiv_regs): Don't create an equivalence for insns that
36439 may trap where the register def does not dominate the use.
36441 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
36443 PR rtl-optimization/78127
36444 * lra.c (lra): Call lra_eliminate before finish the loop after
36447 2017-02-16 Richard Biener <rguenther@suse.de>
36449 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
36451 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
36452 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
36453 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
36454 (isl_val_int_from_wi): New function.
36455 (extract_affine_gmp): Rename to ...
36456 (extract_affine_wi): ... this, take a widest_int.
36457 (extract_affine_int): Just wrap extract_affine_wi.
36458 (add_param_constraints): Use isl_val_int_from_wi.
36459 (add_loop_constraints): Likewise, and extract_affine_wi.
36461 2017-02-15 Jeff Law <law@redhat.com>
36463 PR middle-end/79521
36464 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
36465 ira_init_register_move_cost_if_necessary.
36467 2017-02-15 Martin Sebor <msebor@redhat.com>
36469 PR middle-end/32003
36470 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
36471 removed in a prior commit.
36473 2017-02-15 Bin Cheng <bin.cheng@arm.com>
36475 PR tree-optimization/79347
36476 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
36477 counters during peeling.
36479 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
36481 * Makefile.in (site.exp): Remove "set ISLVER".
36483 2017-02-15 Jakub Jelinek <jakub@redhat.com>
36486 * real.c (real_from_integer): Call real_convert even for decimal.
36488 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
36491 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
36493 2017-02-14 Andrew Pinski <apinski@cavium.com>
36495 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
36496 cores and change the partno/implementer to be correct.
36497 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
36498 the 'B" as the implementer.
36499 * config/aarch64/aarch64-tune.md: Regenerate.
36501 2017-02-14 Carl Love <cel@us.ibm.com>
36503 * config/rs6000/rs6000.c: Add case statement entry to make the
36504 xvcvuxdsp built-in argument unsigned.
36505 * config/rs6000/vsx.md: Fix the source and return operand types so they
36506 match the instruction definitions from the ISA document. Fix typo
36507 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
36510 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
36513 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
36514 member early_clobber_alts.
36515 * lra-lives.c (reg_early_clobber_p): New.
36516 (process_bb_lives): Use it.
36517 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
36518 (debug_operand_data): Initialize early_clobber_alts.
36519 (setup_operand_alternative): Set up early_clobber_alts.
36520 (collect_non_operand_hard_regs): Ditto. Pass early clobber
36521 alternatives to new_insn_reg.
36522 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
36524 (lra_update_insn_regno_info): Pass the new arg.
36526 2017-02-14 Jakub Jelinek <jakub@redhat.com>
36528 PR middle-end/79505
36529 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
36530 (new_oacc_loop_raw): Don't clear already cleared fields.
36533 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
36534 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
36535 _mm512_prefetch_i64gather_ps): New inline functions and macros.
36537 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
36540 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
36542 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
36545 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
36546 the extra instruction to the right place to store 128-bit constant
36549 2017-02-14 Martin Sebor <msebor@redhat.com>
36551 PR middle-end/79448
36552 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
36553 warning for strings of unknown length.
36555 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
36557 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
36559 2017-02-14 Jeff Law <law@redhat.com>
36562 * ira-costs.c (scan_one_insn): Initialize register move costs
36563 for pseudos seen in USE/CLOBBER insns.
36565 PR tree-optimization/79095
36566 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
36567 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
36568 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
36569 if the operands are known to be not equal, then the resulting range
36571 (intersect_ranges): If the new range is ~[0,0] and the old range is
36572 wide, then prefer ~[0,0].
36573 * tree-vrp.c (overflow_comparison_p_1): New function.
36574 (overflow_comparison_p): New function.
36575 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
36576 if NAME is used in an overflow test.
36577 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
36578 overflow check that can be expressed as an equality test, then adjust
36579 ops to be that equality test.
36581 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36583 * config/s390/s390-builtin-types.def: Remove flags argument.
36584 * config/s390/s390.c (s390_init_builtins): Likewise.
36586 2017-02-14 Martin Liska <mliska@suse.cz>
36588 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
36589 vector. Fix trailing white spaces.
36591 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
36593 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
36596 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36598 PR rtl-optimization/68664
36599 * config/arm/arm.c (arm_sched_can_speculate_insn):
36600 New function. Declare prototype.
36601 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
36603 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36605 PR rtl-optimization/68664
36606 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
36608 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
36610 2017-02-14 Amit Pawar <amit.pawar@amd.com>
36612 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
36613 max skip bytes for function, loop and jump.
36615 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
36617 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
36618 ABS_EXPR for gimple dump.
36620 2017-02-14 Jakub Jelinek <jakub@redhat.com>
36623 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
36625 PR tree-optimization/79408
36626 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
36627 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
36628 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
36629 also if rhs1 is INTEGER_CST.
36631 2017-02-14 Richard Biener <rguenther@suse.de>
36633 PR middle-end/79432
36634 * tree-into-ssa.c (insert_phi_nodes): When the function can
36635 have abnormal edges rewrite SSA names with broken use-def
36636 dominance out of SSA and register them for PHI insertion.
36638 2017-02-13 Martin Sebor <msebor@redhat.com>
36640 PR middle-end/79496
36641 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
36642 clearing info.nowrite flag when snprintf size argument is a range.
36644 2017-02-13 Jakub Jelinek <jakub@redhat.com>
36646 * cprop.c (cprop_jump): Add missing space in string literal.
36647 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
36648 (get_constraint_for_component_ref): Likewise.
36649 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
36650 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
36651 * lra-constraints.c (process_alt_operands): Likewise.
36652 * ipa-inline.c (inline_small_functions): Likewise.
36653 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
36654 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
36655 * trans-mem.c (diagnose_tm_1_op): Likewise.
36656 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
36657 (grid_parallel_clauses_gridifiable): Likewise.
36659 * config/nvptx/mkoffload.c (process): Add space in between
36662 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
36663 "MOD4_SSE_REGS" and "ALL_REGS".
36665 * spellcheck.c (test_data): Add , in between "foo" and "food".
36667 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36670 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
36671 boundary crossing check and subsequent code generation agree.
36673 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36675 * config/aarch64/aarch64.c (has_memory_op): Delete.
36676 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
36679 2017-02-13 Jakub Jelinek <jakub@redhat.com>
36681 PR rtl-optimization/79388
36682 PR rtl-optimization/79450
36683 * combine.c (distribute_notes): When removing TEM_INSN for which
36684 corresponding dest has last value recorded, invalidate that last
36687 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36689 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
36690 of explicit '@'. Add missing assembly comment marker on branch costs
36693 2017-02-13 Nathan Sidwell <nathan@acm.org>
36695 * gengtype-lex.l (<in_struct>): Add '/'.
36697 2017-02-13 Martin Liska <mliska@suse.cz>
36700 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
36702 2017-02-13 Richard Biener <rguenther@suse.de>
36704 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
36706 * configure: Re-generate.
36707 * config.in: Likewise.
36708 * graphite-dependences.c: Simplify as if
36709 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
36710 * graphite-isl-ast-to-gimple.c: Likewise.
36711 * graphite-optimize-isl.c: Likewise.
36712 * graphite-poly.c: Likewise.
36713 * graphite-sese-to-poly.c: Likewise.
36714 * graphite.h: Likewise.
36715 * toplev.c: Include isl/version.h and use isl_version () for
36716 printing the ISL version.
36717 * doc/install.texi: Update ISL requirement.
36719 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
36721 * doc/standards.texi (Standards): Update reference to
36724 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
36726 * doc/extend.texi (Named Address Spaces): sourceware.org now
36728 * doc/install.texi (Binaries): Ditto.
36731 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
36733 * doc/cpp.texi: Replace "stringify"/"stringification" with C
36734 standard terminology "stringize"/"stringizing" throughout.
36735 * doc/cppinternals.texi: Likewise.
36737 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
36739 * doc/extend.texi: Fix some spelling mistakes and typos.
36740 * doc/invoke.texi: Likewise.
36742 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
36745 * params.def (inline-min-speedup) Change from 10 to 8.
36747 2017-02-11 Jakub Jelinek <jakub@redhat.com>
36749 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
36752 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
36755 * ipa-inline-analysis.c (get_minimal_bb): New function.
36756 (record_modified): Use it.
36757 (remap_edge_change_prob): Handle also ancestor functions.
36759 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
36761 * doc/contrib.texi (Contributors): Remove broken link into
36762 the Mauve CVS repository.
36764 2017-02-11 Jakub Jelinek <jakub@redhat.com>
36766 PR middle-end/79454
36767 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
36768 result computation whenever lhs doesn't have vector mode, not
36769 just when it has BLKmode.
36771 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
36773 * doc/makefile.texi (profiledbootstrap): Refer to the
36774 installation instructions only in textual form.
36776 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36779 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
36781 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
36783 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
36784 (Specific): Update mingw-w64 reference.
36786 (Specific): Remove broken link to Renesas RX processor.
36788 2017-02-10 Richard Biener <rguenther@suse.de>
36790 * toplev.c (process_options): Do not mention obsolete graphite
36791 options when printing sorry message about missing graphite support.
36792 Mention -floop-nest-optimize.
36794 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
36796 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
36797 (vtst_p16): Likewise.
36798 (vtstq_p8): Likewise.
36799 (vtstq_p16): Likewise.
36801 (vtstq_p64): Likewise.
36802 * config/arm/arm_neon.h (vgetq_lane_p64): New.
36803 (vset_lane_p64): New.
36804 (vsetq_lane_p64): New.
36806 2017-02-10 Jakub Jelinek <jakub@redhat.com>
36808 PR tree-optimization/79411
36809 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
36810 stmt operands are SSA_NAMEs used in abnormal phis.
36811 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
36814 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
36817 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
36820 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
36822 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
36824 2017-02-09 Jakub Jelinek <jakub@redhat.com>
36826 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
36830 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
36831 not arbitrary TREE_CONSTANT.
36834 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
36835 "omp declare target link" attribute unless is_global_var.
36836 * omp-offload.c (find_link_var_op): Likewise.
36838 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
36839 Chung-Lin Tang <cltang@codesourcery.com>
36841 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
36843 (gimplify_adjust_omp_clauses): Don't delete TILE.
36844 (gimplify_omp_for): Deal with TILE.
36845 * internal-fn.c (expand_GOACC_TILE): New function.
36846 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
36848 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
36849 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
36851 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
36852 avoid DIV for outermost collapse var.
36853 (expand_oacc_for): Insert tile element loop as needed. Adjust.
36854 Remove out of date comments, fix whitespace.
36855 * omp-general.c (omp_extract_for_data): Deal with tiling.
36856 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
36857 adjust OLF_DIM_BASE value.
36858 (struct omp_for_data): Add tiling field.
36859 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
36860 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
36861 for auto loops. Remove default auto determining, moved to
36862 oacc_loop_fixed_partitions.
36863 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
36864 stmts, add e_mask field.
36865 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
36866 (oacc_thread_numbers): Use oacc_dim_call.
36867 (oacc_xform_tile): New.
36868 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
36869 (finish_oacc_loop): Adjust for ifns vector.
36870 (oacc_loop_discover_walk): Append loop abstraction sites to list,
36871 add case for GOACC_TILE fns.
36872 (oacc_loop_xform_loop): Delete.
36873 (oacc_loop_process): Iterate over call list directly, and add
36874 handling for GOACC_TILE fns.
36875 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
36877 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
36878 vector partitioning to outer loops. Assign 2 partitions to loops
36879 when available. Add TILE handling.
36880 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
36881 (execite_oacc_device_lower): Process GOACC_TILE fns,
36882 ignore unknown specs.
36883 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
36884 * tree.c (omp_clause_num_ops): Adjust TILE ops.
36885 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
36887 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
36889 * configure.ac (ACX_BUGURL): Update.
36890 * configure: Regenerate.
36892 2017-02-09 Richard Biener <rguenther@suse.de>
36894 PR tree-optimization/69823
36895 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
36896 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
36898 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
36900 * config/arc/arc-c.def: Add __NPS400__ definition.
36901 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
36902 (TARGET_NPS400): Define.
36904 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
36906 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
36908 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
36909 pointer, arch_info.
36910 (arc_cpu_types): Fill the arch_info field with a pointer into the
36911 arc_arch_types table.
36912 (arc_selected_cpu): Declare.
36913 * config/arc/arc.c (arc_selected_cpu): Make global.
36914 (arc_selected_arch): Delete.
36915 (arc_base_cpu): Delete.
36916 (arc_override_options): Remove references to deleted variables,
36917 update access to arch information.
36918 (ARC_OPT): Update access to arch information.
36919 (ARC_OPTX): Likewise.
36920 * config/arc/arc.h (arc_base_cpu): Remove declaration.
36921 (TARGET_ARC600): Update access to arch information.
36922 (TARGET_ARC601): Likewise.
36923 (TARGET_ARC700): Likewise.
36924 (TARGET_EM): Likewise.
36925 (TARGET_HS): Likewise.
36926 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
36929 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
36932 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
36933 condition/operands for integer GE/LE/GEU/LEU operations.
36935 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
36937 PR translation/79397
36938 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
36941 2017-02-08 Martin Jambor <mjambor@suse.cz>
36944 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
36945 whether allocation happened.
36946 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
36947 nothing was allocated.
36949 2017-02-08 Jakub Jelinek <jakub@redhat.com>
36951 PR tree-optimization/79408
36952 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
36953 constant, but SSA_NAME with a known integer range, use the minimum
36954 of that range instead of op1 to determine if modulo can be replaced
36955 with its first operand.
36957 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36959 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
36961 2017-02-08 Richard Biener <rguenther@suse.de>
36963 PR tree-optimization/71824
36964 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
36965 Check all loops contained in the merged region.
36967 2017-02-07 Andrew Pinski <apinski@cavium.com>
36969 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
36971 2017-02-07 Andrew Pinski <apinski@cavium.com>
36973 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
36974 (thunderxt88): Likewise.
36975 (thunderxt81): Disable LSE and change v8.1 to v8.
36976 (thunderxt83): Likewise.
36978 2017-02-07 Jakub Jelinek <jakub@redhat.com>
36979 Richard Biener <rguenther@suse.de>
36981 PR middle-end/79399
36982 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
36983 type from int to size_t.
36984 * ira-costs.c (struct_costs_size): Change type from int to size_t.
36986 2017-02-07 Jakub Jelinek <jakub@redhat.com>
36988 PR rtl-optimization/79386
36989 * cprop.c (bypass_conditional_jumps): Initialize
36990 bypass_last_basic_block already before splitting bbs after
36991 unconditional traps...
36992 (bypass_conditional_jumps): ... rather than here.
36995 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
36996 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
36997 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
36998 fix -masm=intel patterns.
37000 2017-02-07 Richard Biener <rguenther@suse.de>
37002 PR tree-optimization/79256
37003 PR middle-end/79278
37004 * builtins.c (get_object_alignment_2): Use min_align_of_type
37005 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
37006 and ADJUST_FIELD_ALIGN.
37008 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
37010 * doc/tm.texi: Regenerate.
37011 * stor-layout.c (layout_decl): Adjust.
37012 (update_alignment_for_field): Likewise.
37013 (place_field): Likewise.
37014 (min_align_of_type): Likewise.
37015 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
37016 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
37017 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
37018 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
37019 * config/frv/frv.c (frv_adjust_field_align): Likewise.
37020 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
37021 * config/i386/i386.c (x86_field_alignment): Likewise.
37022 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
37023 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
37024 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
37025 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
37026 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
37027 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
37031 2017-01-30 Richard Biener <rguenther@suse.de>
37033 PR tree-optimization/79256
37034 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
37035 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
37038 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
37040 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
37041 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
37042 builtins to SImode and emit a zero-extend, if necessary.
37044 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
37046 * docs/invoke.texi (RISC-V Options): Alphabetize.
37048 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
37050 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
37053 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
37055 * config/riscv/riscv.c: New file.
37056 * common/config/riscv/riscv-common.c: Likewise.
37057 * config.gcc: Likewise.
37058 * config/riscv/constraints.md: Likewise.
37059 * config/riscv/elf.h: Likewise.
37060 * config/riscv/generic.md: Likewise.
37061 * config/riscv/linux.h: Likewise.
37062 * config/riscv/multilib-generator: Likewise.
37063 * config/riscv/peephole.md: Likewise.
37064 * config/riscv/pic.md: Likewise.
37065 * config/riscv/predicates.md: Likewise.
37066 * config/riscv/riscv-builtins.c: Likewise.
37067 * config/riscv/riscv-c.c: Likewise.
37068 * config/riscv/riscv-ftypes.def: Likewise.
37069 * config/riscv/riscv-modes.def: Likewise.
37070 * config/riscv/riscv-opts.h: Likewise.
37071 * config/riscv/riscv-protos.h: Likewise.
37072 * config/riscv/riscv.h: Likewise.
37073 * config/riscv/riscv.md: Likewise.
37074 * config/riscv/riscv.opt: Likewise.
37075 * config/riscv/sync.md: Likewise.
37076 * config/riscv/t-elf-multilib: Likewise.
37077 * config/riscv/t-linux: Likewise.
37078 * config/riscv/t-linux-multilib: Likewise.
37079 * config/riscv/t-riscv: Likewise.
37080 * configure.ac: Likewise.
37081 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
37082 Waterman as RISC-V maintainers.
37083 * doc/install.texi: Add RISC-V entries.
37084 * doc/invoke.texi: Add RISC-V options section.
37085 * doc/md.texi: Add RISC-V constraints section.
37086 * configure: Regenerated.
37088 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
37091 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
37092 false values to be constant vectors with all 0 or all 1 bits set.
37093 (vcondu<mode><mode>): Likewise.
37094 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
37096 (fpmask_comparison_operator): Update comment.
37097 (vecint_comparison_operator): New predicate.
37098 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
37099 vector conditionals when the true and false values are constant
37100 vectors with all 0 bits or all 1 bits set.
37102 2017-02-06 Martin Sebor <msebor@redhat.com>
37104 PR tree-optimization/79376
37105 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
37107 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
37109 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
37110 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
37111 to simplify split condition.
37113 2017-02-06 Jakub Jelinek <jakub@redhat.com>
37115 * omp-expand.c (oxpand_omp_atomic_fetch_op,
37116 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
37119 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
37121 PR rtl-optimization/68664
37122 * target.def (can_speculate_insn): New hook.
37123 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
37124 * doc/tm.texi: Regenerate.
37125 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
37126 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
37127 (rs6000_sched_can_speculate_insn): New function.
37129 2017-02-06 Jakub Jelinek <jakub@redhat.com>
37131 PR tree-optimization/79284
37132 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
37133 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
37134 vectorizable_mask_load_store, vectorizable_operation,
37135 vect_is_simple_cond, get_same_sized_vectype): Use it instead
37136 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
37137 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
37138 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
37139 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
37140 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
37141 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
37142 is_gimple_assign (stmt). Replace another such test with
37143 is_gimple_assign (stmt).
37145 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
37148 * config/avr/avr.c (rtl-iter.h): Include it.
37149 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
37150 (avr_legitimate_combined_insn): ...and implementation.
37152 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
37154 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
37155 * config/s390/s390.c (s390_const_operand_ok)
37156 (s390_canonicalize_comparison, s390_extract_part)
37157 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
37158 (s390_contiguous_bitmask_p, s390_rtx_costs)
37159 (legitimize_pic_address): Likewise.
37160 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
37161 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
37162 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
37163 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
37164 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
37166 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
37168 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
37169 REGNO($0) == REGNO($1).
37171 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37173 * config/s390/linux.h(SIZE_TYPE): Add comment.
37175 2017-02-06 Julian Brown <julian@codesourcery.com>
37176 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
37177 Virendra Pathak <virendra.pathak@broadcom.com>
37179 * config/aarch64/aarch64-cores.def: Change the scheduler
37181 * config/aarch64/aarch64.md: Include thunderx2t99.md.
37182 * config/aarch64/thunderx2t99.md: New file.
37184 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
37186 * doc/standards.texi (Go Language): Update link to language
37189 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
37191 * tree-eh.c (lower_resx): Sanitize profile.
37192 (cleanup_empty_eh_move_lp): Likewise.
37194 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
37197 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
37199 * cfgloopmanip.h (loop_version): Update prototype.
37200 * modulo-sched.c (sms_schedule): Update call of loop_version.
37201 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
37202 * tree-parloops.c (gen_parallel_loop): Likewise.
37203 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
37204 * tree-ssa-loop-split.c (split_loop): Likewise.
37205 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
37206 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
37208 2017-02-05 Martin Liska <mliska@suse.cz>
37211 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
37213 (print_operand_address): Initialize a struct to zero.
37215 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
37217 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
37218 garbage collector only in textual form.
37220 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
37222 * doc/extend.texi (x86 specific memory model extensions for
37223 transactional memory): Simplify a phrase.
37225 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
37228 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
37229 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
37230 (atomic_storedi_1): Likewise.
37232 2017-02-04 Jakub Jelinek <jakub@redhat.com>
37234 PR tree-optimization/79338
37235 * tree-parloops.c (gather_scalar_reductions): Don't call
37236 vect_analyze_loop_form for loop->inner before destroying loop's
37239 2017-02-03 Martin Sebor <msebor@redhat.com>
37241 PR tree-optimization/79327
37242 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
37243 when precision has resulted in leading zeros.
37244 (format_integer): Adjust the likely counter to assume an unknown
37245 argument that may be zero is non-zero.
37247 2017-02-03 Jason Merrill <jason@redhat.com>
37250 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
37251 avoid copying non-taken branch.
37253 2017-02-03 Jakub Jelinek <jakub@redhat.com>
37255 PR tree-optimization/79340
37256 * tree-vect-loop.c (vectorizable_reduction): Release
37257 vec_defs elements after safe_splicing them into other vectors.
37260 PR tree-optimization/79327
37261 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
37262 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
37264 (format_integer): Use wide_int_to_tree instead of build_int_cst
37265 + to_?hwi. If argmin is NULL, just set argmin and argmax to
37266 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
37267 of shortest and longest sequence.
37269 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
37271 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
37272 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
37274 2017-02-03 Walter Lee <walt@tilera.com>
37277 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
37278 after initial stackframe link reg save.
37279 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
37281 2017-02-03 Jakub Jelinek <jakub@redhat.com>
37284 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
37285 wu for stxssp alternative.
37287 2017-02-03 Martin Sebor <msebor@redhat.com>
37289 PR tree-optimization/79352
37290 * gimple-fold.c (get_range_strlen): Add argument.
37291 (get_range_strlen): Change return type to bool.
37292 (get_maxval_strlen): Pass in a dummy argument.
37293 * gimple-fold.h (get_range_strlen): Change return type to bool.
37294 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
37295 * tree.h (array_at_struct_end_p): Add argument.
37296 * tree.c (array_at_struct_end_p): Handle it.
37298 2017-02-03 Martin Liska <mliska@suse.cz>
37301 * multiple_target.c (create_dispatcher_calls): Redirect edge
37302 from a caller of a dispatcher.
37303 (expand_target_clones): Make the clones local.
37304 (ipa_target_clone): Do both target clones and resolvers.
37305 (ipa_dispatcher_calls): Remove the pass.
37306 (pass_dispatcher_calls::gate): Likewise.
37307 (make_pass_dispatcher_calls): Likewise.
37308 * passes.def (pass_target_clone): Put as very first IPA early
37311 2017-02-03 Martin Liska <mliska@suse.cz>
37313 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
37314 in case of a function with ifunc attribute.
37316 2017-02-03 Martin Liska <mliska@suse.cz>
37318 * cgraph.c (cgraph_node::dump): Dump function version info.
37319 * symtab.c (symtab_node::dump_base): Add missing new line.
37321 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
37323 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
37324 (ifcombine_ifandif): Use it.
37326 2017-02-03 Martin Liska <mliska@suse.cz>
37328 * doc/invoke.texi: Document default value for
37329 use-after-scope-direct-emission-threshold.
37331 2017-02-03 Martin Liska <mliska@suse.cz>
37333 PR tree-optimization/79339
37334 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
37335 (format_floating): Likewise.
37337 2017-02-03 Martin Liska <mliska@suse.cz>
37340 * ipa-prop.c (ipa_node_params_t::insert): Remove current
37342 (ipa_node_params_t::remove): Likewise.
37343 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
37344 initialization from removed ipa_node_params_t::insert.
37345 (ipa_node_params::~ipa_node_params): Move from removed
37346 ipa_node_params_t::release.
37347 * symbol-summary.h (symbol_summary::m_released): New member.
37348 Do not release a summary twice. Do not allow to call finalizer
37349 for types of a summary that live in GGC memory.
37351 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
37353 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
37356 2017-02-02 Martin Sebor <msebor@redhat.com>
37358 PR middle-end/79275
37359 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
37360 (format_string): Tighten up the range of output for non-constant
37361 strings and correct the expected range for wide non-constant strings.
37363 2017-02-02 Martin Sebor <msebor@redhat.com>
37365 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
37367 PR middle-end/32003
37368 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
37370 (-fdump-tree-@var): Add to index and document how to come up
37371 with pass-specific option and dump file names.
37372 (-fdump-passes): Clarify where to look for output.
37374 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
37376 PR middle-end/77445
37377 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
37378 statistics of the analyzed path; allow threading for speed when
37379 any of BBs along the path are optimized for speed.
37381 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
37383 PR middle-end/78468
37384 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
37385 settings of the virtual registers.
37388 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
37390 * explow.c (get_dynamic_stack_size): Take known alignment of stack
37391 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
37394 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37396 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
37397 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
37399 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37401 * config/s390/s390.md: Add missing comments with the expanded
37403 * config/s390/vector.md: Likewise.
37404 * config/s390/vx-builtins.md: Likewise.
37406 2017-02-02 Jakub Jelinek <jakub@redhat.com>
37409 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
37410 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
37411 conditions on a single line.
37413 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37415 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
37416 __S390_VX__ to __VX__.
37418 2017-02-01 Andrew Pinski <apinski@cavium.com>
37420 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
37421 stmt_info to record_stmt_cost.
37422 (vect_get_known_peeling_cost): Pass stmt_info if known to
37424 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
37425 cpu_vector_cost field into
37426 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
37427 field into vec_int_stmt_cost and vec_fp_stmt_cost.
37428 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
37429 splitting of scalar_stmt_cost and vec_stmt_cost.
37430 (thunderx_vector_cost): Likewise.
37431 (cortexa57_vector_cost): LIkewise.
37432 (exynosm1_vector_cost): Likewise.
37433 (xgene1_vector_cost): Likewise.
37434 (thunderx2t99_vector_cost): Improve after the splitting of the two
37436 (aarch64_builtin_vectorization_cost): Update for the splitting of
37437 scalar_stmt_cost and vec_stmt_cost.
37439 2017-02-01 Torvald Riegel <triegel@redhat.com>
37440 Richard Henderson <rth@redhat.com>
37442 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
37443 conditional on existance of a fast atomic load.
37444 * optabs-query.c (can_atomic_load_p): New function.
37445 * optabs-query.h (can_atomic_load_p): Declare it.
37446 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
37447 no fast atomic load is available for the particular size of access.
37448 (expand_atomic_compare_and_swap): Likewise.
37449 (expand_atomic_load): Likewise.
37450 (expand_atomic_store): Likewise.
37451 (expand_atomic_fetch_op): Likewise.
37452 * testsuite/lib/target-supports.exp
37453 (check_effective_target_sync_int_128): Remove x86 because it provides
37454 no fast atomic load.
37455 (check_effective_target_sync_int_128_runtime): Likewise.
37457 2017-02-01 Richard Biener <rguenther@suse.de>
37459 * graphite.c: Include tree-vectorizer.h for find_loop_location.
37460 (graphite_transform_loops): Provide opt-info for optimized nests.
37461 * tree-parloop.c (parallelize_loops): Provide opt-info for
37462 parallelized loops.
37464 2017-02-01 Richard Biener <rguenther@suse.de>
37466 PR middle-end/79315
37467 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
37468 was not set before.
37470 2017-02-01 Richard Biener <rguenther@suse.de>
37472 PR tree-optimization/71824
37473 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
37474 Verify the loops are valid in the merged SESE region.
37475 (scop_detection::can_represent_loop_1): Check analyzing the
37476 evolution of the number of iterations in the region succeeds.
37478 2017-01-31 Ian Lance Taylor <iant@golang.org>
37480 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
37481 REG_ARGS_SIZE note to 32-bit push insns and call insn.
37483 2017-01-31 David Malcolm <dmalcolm@redhat.com>
37485 PR preprocessor/79210
37486 * input.c (get_substring_ranges_for_loc): Replace line_width
37487 assertion with error-handling.
37489 2017-01-31 Richard Biener <rguenther@suse.de>
37491 PR tree-optimization/77318
37492 * graphite-sese-to-poly.c (extract_affine): Fix assert.
37493 (create_pw_aff_from_tree): Take loop parameter.
37494 (add_condition_to_pbb): Pass loop of the condition to
37495 create_pw_aff_from_tree.
37497 2017-01-31 Jakub Jelinek <jakub@redhat.com>
37499 * config/s390/s390.c (s390_asan_shadow_offset): New function.
37500 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
37502 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
37506 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
37508 (convert_int_to_float128): Likewise.
37509 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
37510 (convert_int_to_float128): Likewise.
37511 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
37512 (UNSPEC_IEEE128_CONVERT): Likewise.
37513 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
37514 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
37515 Use local variables for IBM extended format.
37516 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
37517 (fix_trunc<mode>si2_fprs): Likewise.
37518 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
37519 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
37520 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
37521 to know that we can now have integers of all sizes in vector
37523 (fix<uns>_<mode>di2_hw): Likewise.
37524 (float<uns>_<mode>si2_hw): Likewise.
37525 (fix_<mode>si2_hw): Likewise.
37526 (fixuns_<mode>si2_hw): Likewise.
37527 (float<uns>_<mode>di2_hw): Likewise.
37528 (float_<mode>di2_hw): Likewise.
37529 (float_<mode>si2_hw): Likewise.
37530 (floatuns_<mode>di2_hw): Likewise.
37531 (floatuns_<mode>si2_hw): Likewise.
37532 (xscvqp<su>wz_<mode>): Delete, no longer used.
37533 (xscvqp<su>dz_<mode>): Likewise.
37534 (xscv<su>dqp_<mode>): Likewise.
37535 (ieee128_mfvsrd_64bit): Likewise.
37536 (ieee128_mfvsrd_32bit): Likewise.
37537 (ieee128_mfvsrwz): Likewise.
37538 (ieee128_mtvsrw): Likewise.
37539 (ieee128_mtvsrd_64bit): Likewise.
37540 (ieee128_mtvsrd_32bit): Likewise.
37542 2017-01-31 Martin Liska <mliska@suse.cz>
37545 * ipa-prop.c (ipa_free_all_node_params): Call release method
37546 instead of ~sumbol_summary to not to trigger double times
37549 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
37551 PR tree-optimization/71691
37552 * bitmap.h (class auto_bitmap): New.
37553 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
37554 is_maybe_undefined instead of ssa_undefined_value_p.
37556 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37558 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
37559 __S390_ARCH_LEVEL__ to __ARCH__.
37561 2017-01-31 Jakub Jelinek <jakub@redhat.com>
37563 PR tree-optimization/79267
37564 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
37565 if should_remove_lhs_p is true.
37567 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
37570 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
37571 (add_alignment_attribute): New.
37572 (base_type_die): Add alignment attribute.
37573 (subrange_type_die): Likewise.
37574 (modified_type_die): Likewise.
37575 (gen_array_type_die): Likewise.
37576 (gen_descr_array_type_die: Likewise.
37577 (gen_enumeration_type_die): Likewise.
37578 (gen_subprogram_die): Likewise.
37579 (gen_variable_die): Likewise.
37580 (gen_field_die): Likewise.
37581 (gen_ptr_to_mbr_type_die): Likewise.
37582 (gen_struct_or_union_type_die): Likewise.
37583 (gen_subroutine_type_die): Likewise.
37584 (gen_typedef_die): Likewise.
37585 (base_type_cmp): Compare alignment attribute.
37587 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37590 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
37591 (setb_unsigned) New pattern for setb with CCUNS.
37592 * config/rs6000/rs6000.c (expand_block_compare): Use a different
37593 subfc./subfe sequence to avoid overflow problems. Generate a
37594 shorter sequence with cmpld/setb for power9.
37595 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
37596 for generating subfc. instruction.
37597 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
37598 now uses this instruction.
37600 2017-01-30 Ian Lance Taylor <iant@google.com>
37603 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
37604 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
37606 2017-01-30 Martin Sebor <msebor@redhat.com>
37608 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
37609 Move constant to the right of a relational operator.
37610 (get_mpfr_format_length, format_character, format_string): Ditto.
37611 (should_warn_p, maybe_warn): Same.
37613 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
37615 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
37618 * asan.c (get_translation_unit_decl): Remove function.
37619 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
37621 2017-01-30 Martin Liska <mliska@suse.cz>
37623 PR gcov-profile/79259
37624 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
37625 -fprofile-generate.
37627 2017-01-30 Martin Liska <mliska@suse.cz>
37630 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
37631 Initialize variables with NULL value.
37633 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
37636 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
37638 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
37640 2017-01-30 Richard Biener <rguenther@suse.de>
37642 PR tree-optimization/79276
37643 * tree-vrp.c (process_assert_insertions): Properly adjust common
37644 when removing a duplicate.
37646 2017-01-30 Richard Biener <rguenther@suse.de>
37648 PR tree-optimization/79256
37649 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
37650 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
37652 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
37654 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
37657 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
37658 ("*r<noxa>sbg_<mode>_sll_bitmask")
37659 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
37660 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
37661 Use contiguous_bitmask_nowrap_operand.
37663 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37666 * config/rs6000/altivec.h (vec_xl): Revise #define.
37667 (vec_xst): Likewise.
37669 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
37671 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
37673 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
37675 PR rtl-optimization/79194
37676 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
37677 traps before call to bypass_conditional_jumps.
37679 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
37681 PR tree-optimization/71374
37682 * lra-constraints.c (check_conflict_input_operands): New.
37683 (match_reload): Use it.
37685 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
37688 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
37689 account to calculate conflict_set.
37691 2017-01-27 Bin Cheng <bin.cheng@arm.com>
37693 PR rtl-optimization/78559
37694 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
37695 other_insn in combine.
37697 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
37699 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
37700 uint16_type_node for BT_UINT16.
37702 2017-01-27 David Malcolm <dmalcolm@redhat.com>
37704 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
37705 "RTL Tests" to menu.
37706 (GIMPLE Tests): New node.
37707 (RTL Tests): New node.
37709 2017-01-27 Richard Biener <rguenther@suse.de>
37711 PR tree-optimization/79245
37712 * tree-loop-distribution.c (distribute_loop): Apply cost
37713 modeling also to detected patterns.
37715 2017-01-27 Richard Biener <rguenther@suse.de>
37717 PR tree-optimization/71433
37718 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
37719 (compare_assert_loc): New function.
37720 (process_assert_insertions): Sort and optimize assert locations
37721 to remove duplicates and push down identical assertions on
37722 edges to their destination block.
37724 2017-01-27 Richard Biener <rguenther@suse.de>
37726 PR tree-optimization/79244
37727 * tree-vrp.c (remove_range_assertions): Forcefully propagate
37728 out SSA names even if abnormal.
37730 2017-01-27 Jakub Jelinek <jakub@redhat.com>
37732 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
37733 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
37734 instead of MPFR_RNDN.
37736 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
37739 * arm.c (arm_option_override): Don't call build_target_option_node
37740 until after doing all option overrides.
37741 (arm_valid_target_attribute_tree): Likewise.
37743 2017-01-27 Martin Liska <mliska@suse.cz>
37745 * doc/invoke.texi (-fprofile-arcs): Document profiling support
37746 for {cd}tors and C++ {cd}tors.
37748 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
37750 * config/s390/s390.md ("*setmem_long_and")
37751 ("*setmem_long_and_31z"): Use zero_extend instead of and.
37753 2017-01-26 Martin Sebor <msebor@redhat.com>
37755 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
37758 2017-01-26 Martin Sebor <msebor@redhat.com>
37760 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
37761 HAVE_DFmode before using XFmode or DFmode.
37762 (parse_directive): Avoid using the z length modifier to avoid
37763 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
37765 PR middle-end/78703
37766 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
37767 to accept adjustment as an array.
37768 (get_int_range): New function.
37769 (struct directive): Make width and prec arrays.
37770 (directive::set_width, directive::set_precision): Call get_int_range.
37771 (format_integer, format_floating): Handle width and precision ranges.
37772 (format_string, parse_directive): Same.
37774 2017-01-26 Jakub Jelinek <jakub@redhat.com>
37777 * dwarf2out.c (generate_skeleton_bottom_up): For children with
37778 comdat_type_p set, just clone them, but keep the children in the
37782 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
37783 which have direct callers with -fvar-tracking-assignments enabled
37785 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
37786 inside of type units.
37788 2017-01-26 Martin Sebor <msebor@redhat.com>
37790 PR middle-end/78703
37791 * gimple-ssa-sprintf.c (struct result_range): Add likely and
37793 (struct format_result): Replace number_chars, number_chars_min,
37794 and number_chars_max with a single member of struct result_range.
37796 (format_result::operator+=): Adjust.
37797 (struct fmtresult): Remove bounded. Handle likely and unlikely
37799 (fmtresult::adjust_for_width_or_precision): New function.
37800 (fmtresult:type_max_digits): New function.
37801 (bytes_remaining): Handle likely and unlikely counters.
37802 (min_bytes_remaining): Remove.
37803 (format_percent): Simplify.
37804 (format_integer, format_floating): Set likely and unlikely counters.
37805 (get_string_length, format_character, format_string): Same.
37806 (format_plain, should_warn_p): New function.
37807 (maybe_warn): Call should_warn_p. Update diagnostic messages
37808 and handle those for all directives, including plain strings.
37809 (format_directive): Handle likely and unlikely counters.
37810 Remove unnecessary quoting from diagnostics. Add an informational
37812 (add_bytes): Remove.
37813 (pass_sprintf_length::compute_format_length): Simplify.
37814 (try_substitute_return_value): Handle likely and unlikely counters.
37816 2017-01-26 Carl Love <cel@us.ibm.com>
37818 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
37819 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
37821 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
37824 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
37825 endianess for subregs into account.
37826 * lra-constraints.c (lra_constraints): Do risky transformations
37827 always on the first iteration.
37828 * lra-lives.c (check_pseudos_live_through_calls): Add arg
37829 last_call_used_reg_set.
37830 (process_bb_lives): Define and use last_call_used_reg_set.
37831 * lra.c (lra): Always continue after lra_constraints on the first
37834 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
37836 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
37838 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
37840 2017-01-26 Jakub Jelinek <jakub@redhat.com>
37842 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
37843 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
37844 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
37845 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
37846 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
37847 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
37848 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
37849 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
37850 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
37852 2017-01-26 Marek Polacek <polacek@redhat.com>
37855 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
37856 for the third operand.
37858 2017-01-26 Jakub Jelinek <jakub@redhat.com>
37860 PR middle-end/79236
37861 * omp-low.c (struct omp_context): Add simt_stmt field.
37862 (scan_omp_for): Return omp_context *.
37863 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
37864 context to the _simt_ SIMD stmt.
37865 (lower_omp_for): For combined SIMD with sibling _simt_
37866 SIMD, make sure to use the same decls in _looptemp_
37867 clauses as in the sibling.
37869 2017-01-26 David Sherwood <david.sherwood@arm.com>
37871 PR middle-end/79212
37872 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
37875 2017-01-26 Jakub Jelinek <jakub@redhat.com>
37878 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
37879 emit fld b; fld a; if possible.
37881 * brig-builtins.def: Update copyright years.
37882 * config/arm/arm_acle_builtins.def: Update copyright years.
37884 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
37887 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
37888 constraint instead of o for the stxsd instruction.
37890 2017-01-25 Carl Love <cel@us.ibm.com>
37892 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
37893 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
37895 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
37897 * doc/invoke.texi (C++ Dialect Options): Fix typo.
37899 2017-01-25 Richard Biener <rguenther@suse.de>
37901 PR tree-optimization/69264
37902 * target.def (vector_alignment_reachable): Improve documentation.
37903 * doc/tm.texi: Regenerate.
37904 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
37906 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
37907 earlier changes with respect to TYPE_USER_ALIGN.
37908 (vector_alignment_reachable_p): Likewise. Improve dumping.
37910 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37913 * config/arm/arm.md (xordi3): Force constant operand into a register
37916 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37918 * doc/invoke.texi (-fstore-merging): Correct default optimization
37919 levels at which it is enabled.
37920 (-O): Move -fstore-merging from list to...
37923 2017-01-25 Richard Biener <rguenther@suse.de>
37926 * omp-expand.c: Include debug.h.
37927 (expand_omp_taskreg): Make sure to generate early debug before
37928 outlining anything from a function.
37929 (expand_omp_target): Likewise.
37930 (grid_expand_target_grid_body): Likewise.
37932 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
37935 * asan.c (get_translation_unit_decl): New function.
37936 (asan_add_global): Extract modules file name from globals
37937 TRANSLATION_UNIT_DECL name.
37939 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
37942 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
37943 for long calls with APCS frame and VFP.
37945 2017-01-24 David Malcolm <dmalcolm@redhat.com>
37947 * cfg.c (original_copy_tables_initialized_p): New function.
37948 * cfg.h (original_copy_tables_initialized_p): New decl.
37949 * cfgrtl.c (relink_block_chain): Guard the call to
37950 free_original_copy_tables with a call to
37951 original_copy_tables_initialized_p.
37952 * cgraph.h (symtab_node::native_rtl_p): New decl.
37953 * cgraphunit.c (symtab_node::native_rtl_p): New function.
37954 (symtab_node::needed_p): Don't assert for early assembly output
37955 for __RTL functions.
37956 (cgraph_node::finalize_function): Set "force_output" for __RTL
37958 (cgraph_node::analyze): Bail out early for __RTL functions.
37959 (analyze_functions): Update assertion to support __RTL functions.
37960 (cgraph_node::expand): Bail out early for __RTL functions.
37961 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
37963 * function.h (struct function): Update comment for field
37965 * gimple-expr.c: Include "tree-pass.h".
37966 (gimple_has_body_p): Return false for __RTL functions.
37967 * Makefile.in (OBJS): Add run-rtl-passes.o.
37968 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
37970 (gcc::pass_manager::get_clean_slate): New accessor.
37971 * passes.c: Include "insn-addr.h".
37972 (should_skip_pass_p): Add logging. Update logic for running
37973 "expand" to be compatible with both __GIMPLE and __RTL. Guard
37974 property-provider override so it is only done for gimple passes.
37976 (skip_pass): New function.
37977 (execute_one_pass): Call skip_pass when skipping passes.
37978 * read-md.c (md_reader::read_char): Support filtering
37979 the input to a subset of line numbers.
37980 (md_reader::md_reader): Initialize fields
37981 m_first_line and m_last_line.
37982 (md_reader::read_file_fragment): New function.
37983 * read-md.h (md_reader::read_file_fragment): New decl.
37984 (md_reader::m_first_line): New field.
37985 (md_reader::m_last_line): New field.
37986 * read-rtl-function.c (function_reader::create_function): Only
37987 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
37988 curr_properties. Set DECL_INITIAL to a dummy block.
37989 (read_rtl_function_body_from_file_range): New function.
37990 * read-rtl-function.h (read_rtl_function_body_from_file_range):
37992 * run-rtl-passes.c: New file.
37993 * run-rtl-passes.h: New file.
37995 2017-01-24 Jeff Law <law@redhat.com>
37997 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
38000 2017-01-24 Bin Cheng <bin.cheng@arm.com>
38002 PR tree-optimization/79159
38003 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
38004 (record_nonwrapping_iv): Improve boundary using above function if no
38005 value range information.
38007 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
38008 Martin Jambor <mjambor@suse.cz>
38010 * brig-builtins.def: New file.
38011 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
38012 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
38013 (DEF_HSAIL_SAT_BUILTIN): Likewise.
38014 (DEF_HSAIL_INTR_BUILTIN): Likewise.
38015 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
38016 * builtin-types.def (BT_INT8): New.
38017 (BT_INT16): Likewise.
38018 (BT_UINT8): Likewise.
38019 (BT_UINT16): Likewise.
38020 (BT_FN_ULONG): Likewise.
38021 (BT_FN_UINT_INT): Likewise.
38022 (BT_FN_UINT_ULONG): Likewise.
38023 (BT_FN_UINT_LONG): Likewise.
38024 (BT_FN_UINT_PTR): Likewise.
38025 (BT_FN_ULONG_PTR): Likewise.
38026 (BT_FN_INT8_FLOAT): Likewise.
38027 (BT_FN_INT16_FLOAT): Likewise.
38028 (BT_FN_UINT32_FLOAT): Likewise.
38029 (BT_FN_UINT16_FLOAT): Likewise.
38030 (BT_FN_UINT8_FLOAT): Likewise.
38031 (BT_FN_UINT64_FLOAT): Likewise.
38032 (BT_FN_UINT16_UINT32): Likewise.
38033 (BT_FN_UINT32_UINT16): Likewise.
38034 (BT_FN_UINT16_UINT16_UINT16): Likewise.
38035 (BT_FN_INT_PTR_INT): Likewise.
38036 (BT_FN_UINT_PTR_UINT): Likewise.
38037 (BT_FN_LONG_PTR_LONG): Likewise.
38038 (BT_FN_ULONG_PTR_ULONG): Likewise.
38039 (BT_FN_VOID_UINT64_UINT64): Likewise.
38040 (BT_FN_UINT8_UINT8_UINT8): Likewise.
38041 (BT_FN_INT8_INT8_INT8): Likewise.
38042 (BT_FN_INT16_INT16_INT16): Likewise.
38043 (BT_FN_INT_INT_INT): Likewise.
38044 (BT_FN_UINT_FLOAT_UINT): Likewise.
38045 (BT_FN_FLOAT_UINT_UINT): Likewise.
38046 (BT_FN_ULONG_UINT_UINT): Likewise.
38047 (BT_FN_ULONG_UINT_PTR): Likewise.
38048 (BT_FN_ULONG_ULONG_ULONG): Likewise.
38049 (BT_FN_UINT_UINT_UINT): Likewise.
38050 (BT_FN_VOID_UINT_PTR): Likewise.
38051 (BT_FN_UINT_UINT_PTR: Likewise.
38052 (BT_FN_UINT32_UINT64_PTR): Likewise.
38053 (BT_FN_INT_INT_UINT_UINT): Likewise.
38054 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
38055 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
38056 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
38057 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
38058 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
38059 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
38060 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
38061 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
38062 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
38063 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
38064 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
38065 * doc/frontends.texi: List BRIG FE.
38066 * doc/install.texi (Testing): Add BRIG tesring requirements.
38067 * doc/invoke.texi (Overall Options): Mention BRIG.
38068 * doc/standards.texi (Standards): Doucment BRIG HSA version.
38070 2017-01-24 Richard Biener <rguenther@suse.de>
38072 PR translation/79208
38073 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
38075 2017-01-24 Martin Jambor <mjambor@suse.cz>
38078 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
38079 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
38080 and known_contexts.
38082 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
38084 PR middle-end/79123
38085 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
38086 casts from signed to unsigned really don't have a range.
38088 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
38090 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
38091 GMP_RNDx for compatiblity.
38093 2017-01-24 Martin Liska <mliska@suse.cz>
38096 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
38097 that would prevent us to call alloca with -1 as argument.
38099 2017-01-24 Jakub Jelinek <jakub@redhat.com>
38101 * dwarf2out.c (output_compilation_unit_header, output_file_names):
38102 Avoid -Wformat-security warning.
38104 2017-01-23 Andrew Pinski <apinski@cavium.com>
38106 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
38109 2017-01-23 Martin Sebor <msebor@redhat.com>
38111 PR middle-end/78703
38112 * gimple-ssa-sprintf.c (warn_level): New global.
38113 (format_integer): Use it here and throughout the rest of the file.
38114 Use the same switch to compute sign as base.
38115 (maybe_warn): New function.
38116 (format_directive): Factor out warnings into maybe_warn.
38117 Add debugging output. Use warn_level.
38118 (add_bytes): Use warn_level.
38119 (pass_sprintf_length::compute_format_length): Add debugging output.
38120 (try_substitute_return_value): Same.
38121 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
38123 PR middle-end/78703
38124 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
38125 (struct fmtresult, format_integer, format_floating): Adjust.
38126 (fmtresult::fmtresult): Set max correctly in two argument ctor.
38127 (get_string_length, format_string,format_directive): Same.
38128 (pass_sprintf_length::compute_format_length): Same.
38129 (try_substitute_return_value): Simplify slightly.
38131 PR middle-end/78703
38132 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
38133 (fmtresult::operator+=): Outlined.
38134 (struct fmtresult): Add ctors.
38135 (struct conversion_spec): Rename...
38136 (struct directive): ...to this. Add and remove data members.
38137 (directive::set_width, directive::set_precision): New functions.
38138 (format_percent): Use fmtresult ctor.
38139 (get_width_and_precision): Remove.
38140 (format_integer): Make naming changes. Avoid computing width and
38142 (format_floating): Same. Adjust indentation.
38143 (format_character, format_none): New functions.
38144 (format_string): Moved character handling to format_character.
38145 (format_directive): Remove arguments, change return type.
38146 (parse_directive): New function.
38147 (pass_sprintf_length::compute_format_length): Move directive
38148 parsing to parse_directive.
38150 2017-01-23 Jakub Jelinek <jakub@redhat.com>
38152 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
38153 (assign_assembler_name_if_needed): ... this.
38154 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
38155 (assign_assembler_name_if_needed): ... this.
38156 (free_lang_data_in_cgraph): Adjust callers.
38157 * cgraphunit.c (cgraph_node::analyze): Likewise.
38158 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
38161 2017-01-23 Richard Biener <rguenther@suse.de>
38163 PR tree-optimization/79088
38164 PR tree-optimization/79188
38165 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
38166 resetting loop bounds after last path deletion. Reset loop
38167 bounds of the target loop, make code match the comments.
38168 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
38169 Make sure loops need no fixups.
38171 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
38173 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
38174 exponent support with double type for first argument.
38175 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
38176 type returned by __builtin_vec_extract_sig,
38177 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
38178 functions from "vector int" to "vector unsigned int" or from
38179 "vector long long int" to "vector unsigned long long int".
38180 Changed type returned by __builtin_vec_extract_exp,
38181 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
38182 functions from "vector int" to "vector unsigned int" or from
38183 "vector long long int" to "vector unsigned long long int".
38184 Changed return type of __builtin_vec_test_data_class,
38185 __builtin_vec_test_data_class_sp, and
38186 __builtin_vec_test_data_class_dp from "vector int" to
38187 "vector bool int" or from "vector long long int" to "vector bool
38188 long long int" and changed second argument type from "unsigned
38189 int" to "int". Added new overloaded function forms "vector float
38190 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
38191 "vector float __builtin_vec_insert_exp_sp (vector float, vector
38192 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
38193 double, vector unsigned long long int)" and "vector double
38194 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
38195 long int)". Changed return type of
38196 __builtin_scalar_test_data_class and
38197 __builtin_scalar_test_data_class_sp and
38198 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
38199 int" and changed second argument from "unsigned int" to "int".
38200 Changed type returned by __builtin_scalar_test_neg,
38201 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
38202 from "int" to "bool int". Added new overloaded function form
38203 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
38204 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
38205 exponent double-precision with floating point first argument.
38206 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
38207 documentation of scalar_test_data_class, scalar_test_neg,
38208 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
38209 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
38210 vec_test_data_class built-in functions to reflect refinements in
38211 their type signatures.
38213 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
38215 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
38217 (aarch64_elf_asm_destructor): Likewise.
38219 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
38221 PR rtl-optimization/78634
38222 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
38223 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
38224 * ifcvt.c (noce_try_cmove): Add missing cost check.
38226 PR rtl-optimization/71724
38227 * combine.c (if_then_else_cond): Look for situations where it is
38228 beneficial to undo the work of one of the recursive calls.
38230 2017-01-23 Bin Cheng <bin.cheng@arm.com>
38232 PR tree-optimization/70754
38233 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
38234 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
38235 combined stmt before it if not NULL.
38236 (combine_chains): Process refs reversely and compute dominance point
38239 2017-01-23 Martin Liska <mliska@suse.cz>
38241 PR tree-optimization/79196
38242 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
38243 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
38245 (strlen_optimize_stmt): Call the renamed function.
38247 2017-01-23 Michael Matz <matz@suse.de>
38249 PR tree-optimization/78384
38250 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
38252 2017-01-23 Richard Biener <rguenther@suse.de>
38254 PR tree-optimization/79186
38255 * tree-vrp.c (register_new_assert_for): Make sure we've seen
38256 both incoming edges before moving an assert.
38258 2017-01-23 Martin Jambor <mjambor@suse.cz>
38260 * ipa-prop.c (load_from_param_1): Removed.
38261 (load_from_unmodified_param): Bits from load_from_param_1 put back
38263 (load_from_param): Removed.
38264 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
38265 with stmt. Reverted back to use of load_from_unmodified_param.
38267 2017-01-23 Martin Jambor <mjambor@suse.cz>
38270 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
38271 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
38272 field a pointer to garbage collected vector, mark lattices and
38273 ipcp_orig_node with GTY((skip)).
38274 (ipa_get_param_count): Adjust to descriptors being a pointer.
38275 (ipa_get_param): Likewise.
38276 (ipa_get_type): Likewise.
38277 (ipa_get_param_move_cost): Likewise.
38278 (ipa_set_param_used): Likewise.
38279 (ipa_get_controlled_uses): Likewise.
38280 (ipa_set_controlled_uses): Likewise.
38281 (ipa_is_param_used): Likewise.
38282 (ipa_node_params_t): Move into garbage collector. New methods insert
38284 (ipa_node_params_sum): Annotate wth GTY(()).
38285 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
38287 (ipa_load_from_parm_agg): Adjust declaration.
38288 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
38289 * ipa-profile.c (ipa_profile): Likewise.
38290 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
38291 (ipa_populate_param_decls): Make descriptors parameter garbage
38293 (ipa_dump_param): Adjust to descriptors being a pointer.
38294 (ipa_alloc_node_params): Likewise.
38295 (ipa_initialize_node_params): Likewise.
38296 (load_from_param_1): Make descriptors parameter garbage collected.
38297 (load_from_unmodified_param): Likewise.
38298 (load_from_param): Likewise.
38299 (ipa_load_from_parm_agg): Likewise.
38300 (ipa_node_params::~ipa_node_params): Removed.
38301 (ipa_free_all_node_params): Remove call to delete operator.
38302 (ipa_node_params_t::insert): New.
38303 (ipa_node_params_t::remove): Likewise.
38304 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
38305 copy known_csts and known_contexts vectors.
38306 (ipa_read_node_info): Adjust to descriptors being a pointer.
38307 (ipcp_modif_dom_walker): Make m_descriptors field garbage
38309 (ipcp_transform_function): Make descriptors variable garbage
38312 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
38314 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
38315 * config/i386/avx512dqintrin.h: Ditto.
38316 * config/i386/avx512fintrin.h: Ditto.
38317 * config/i386/i386.c: Handle new builtins.
38318 * config/i386/i386-builtin.def: Add new builtins.
38319 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
38320 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
38322 2017-01-23 Jakub Jelinek <jakub@redhat.com>
38323 Martin Liska <mliska@suse.cz>
38325 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
38326 * asan.c (asan_expand_poison_ifn): Support stores and use
38327 appropriate ASAN report function.
38328 * internal-fn.c (expand_ASAN_POISON_USE): New function.
38329 * internal-fn.def (ASAN_POISON_USE): Declare.
38330 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
38331 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
38332 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
38333 ASAN_POISON calls w/o LHS.
38334 * tree-ssa.c (execute_update_addresses_taken): Create clobber
38335 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
38336 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
38337 * gimplify.c (asan_poison_variables): Add attribute
38338 use_after_scope_memory to variables that really needs to live
38340 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
38341 having the attribute.
38343 2017-01-23 Martin Liska <mliska@suse.cz>
38345 * asan.c (create_asan_shadow_var): New function.
38346 (asan_expand_poison_ifn): Likewise.
38347 * asan.h (asan_expand_poison_ifn): New declaration.
38348 * internal-fn.c (expand_ASAN_POISON): Likewise.
38349 * internal-fn.def (ASAN_POISON): New builtin.
38350 * sanopt.c (pass_sanopt::execute): Expand
38351 asan_expand_poison_ifn.
38352 * tree-inline.c (copy_decl_for_dup_finish): Make function
38354 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
38355 * tree-ssa.c (is_asan_mark_p): New function.
38356 (execute_update_addresses_taken): Rewrite local variables
38357 (identified just by use-after-scope as addressable) into SSA.
38359 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
38361 * doc/install.texi (Specific): opensource.apple.com uses https
38362 now. Remove trailing slash.
38364 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
38366 * README.Portability: Remove note on an Irix compatibility issue.
38368 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
38370 * gcov.c (INCLUDE_ALGORITHM): Define.
38371 (INCLUDE_VECTOR): Define.
38372 No longer include <vector> and <algorithm> directly.
38374 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
38376 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
38378 * doc/invoke.texi (Code Gen Options): Ditto.
38380 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
38383 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
38385 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
38387 rtl-optimization/79125
38388 * cprop.c (local_cprop_pass): Handle cases where we make an
38389 unconditional trap.
38391 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
38395 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
38396 read from, for big endian.
38398 2017-01-20 Jiong Wang <jiong.wang@arm.com>
38400 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
38401 register pauth builtins for LP64 only.
38403 2017-01-20 Marek Polacek <polacek@redhat.com>
38406 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
38409 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
38411 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
38413 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
38414 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
38415 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
38417 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38420 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
38421 in big-endian mode when they are not a single duplicated value.
38423 2017-01-20 Richard Biener <rguenther@suse.de>
38425 * BASE-VER: Bump to 7.0.1.
38427 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
38429 * omp-low.c (omplow_simd_context): New struct. Use it...
38430 (lower_rec_simd_input_clauses): ...here and...
38431 (lower_rec_input_clauses): ...here to hold common data. Adjust all
38432 references to idx, lane, max_vf, is_simt.
38434 2017-01-20 Graham Markall <graham.markall@embecosm.com>
38436 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
38439 2017-01-20 Martin Jambor <mjambor@suse.cz>
38441 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
38442 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
38444 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
38445 (GTFILES): Rename hsa.c to hsa-common.c.
38446 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
38447 * hsa-dump.c: Likewise.
38448 * hsa-gen.c: Likewise.
38449 * hsa-regalloc.c: Likewise.
38450 * ipa-hsa.c: Likewise.
38451 * omp-expand.c: Likewise.
38452 * omp-low.c: Likewise.
38453 * toplev.c: Likewise.
38455 2017-01-20 Marek Polacek <polacek@redhat.com>
38458 * doc/invoke.texi: Document -Wduplicated-branches.
38459 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
38460 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
38461 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
38462 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
38463 return 0 only when not OEP_LEXICOGRAPHIC.
38464 (fold_build_cleanup_point_expr): Use the expression
38465 location when building CLEANUP_POINT_EXPR.
38466 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
38467 * tree.c (add_expr): Handle error_mark_node.
38469 2017-01-20 Martin Liska <mliska@suse.cz>
38472 * tree-profile.c (init_ic_make_global_vars): Do not call
38474 (gimple_init_gcov_profiler): Likewise.
38476 2017-01-20 Martin Liska <mliska@suse.cz>
38479 * cgraph.h (maybe_create_reference): Remove argument and
38481 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
38483 * ipa-cp.c (create_specialized_node): Likewise.
38484 * symtab.c (symtab_node::maybe_create_reference): Handle
38485 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
38487 2017-01-20 Martin Liska <mliska@suse.cz>
38489 * read-rtl-function.c (function_reader::create_function): Use
38490 build_decl instread of build_decl_stat.
38492 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
38494 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
38495 * config/i386/avx512dqintrin.h: Ditto.
38496 * config/i386/avx512fintrin.h: Ditto.
38497 * config/i386/i386-builtin-types.def: Add new types.
38498 * config/i386/i386.c: Handle new types.
38499 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
38500 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
38501 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
38502 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
38503 (__builtin_ia32_kshiftridi): New.
38504 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
38506 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
38510 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
38511 define to rs6000_init_stack_protect_guard.
38512 (rs6000_init_stack_protect_guard): New function.
38514 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
38515 Yunqiang Su <yunqiang.su@imgtec.com>
38517 * config.gcc (supported_defaults): Add madd4.
38518 (with_madd4): Add validation.
38519 (all_defaults): Add madd4.
38520 * config/mips/mips.opt (mmadd4): New option.
38521 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
38523 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
38525 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
38526 (ISA_HAS_FUSED_MADD4): Likewise.
38527 * doc/invoke.texi (-mmadd4): Document the new option.
38528 * doc/install.texi (--with-madd4): Document the new option.
38530 2017-01-19 Jiong Wang <jiong.wang@arm.com>
38532 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
38533 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
38534 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
38535 (aarch64_init_pauth_hint_builtins): New.
38536 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
38537 (aarch64_expand_builtin): Expand new builtins.
38539 2017-01-19 Jiong Wang <jiong.wang@arm.com>
38541 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
38542 * combine-stack-adj.c (no_unhandled_cfa): Handle
38543 REG_CFA_TOGGLE_RA_MANGLE.
38544 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
38545 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
38546 info for return address signing.
38547 (aarch64_expand_epilogue): Likewise.
38549 2017-01-19 Jiong Wang <jiong.wang@arm.com>
38551 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
38552 * config/aarch64/aarch64-protos.h
38553 (aarch64_return_address_signing_enabled): New declaration.
38554 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
38556 (aarch64_expand_prologue): Sign return address before it's pushed onto
38558 (aarch64_expand_epilogue): Authenticate return address fetched from
38560 (aarch64_override_options): Sanity check for ILP32 and ISA level.
38561 (aarch64_attributes): New function attributes for "sign-return-address".
38562 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
38563 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
38564 ("*do_return"): Generate combined instructions according to key index.
38565 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
38566 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
38568 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
38569 * config/aarch64/aarch64.opt (msign-return-address=): New.
38570 * doc/extend.texi (AArch64 Function Attributes): Documents
38571 "sign-return-address=".
38572 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
38574 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
38576 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
38577 overall option summary.
38579 2017-01-19 Jiong Wang <jiong.wang@arm.com>
38581 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
38582 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
38583 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
38584 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
38586 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
38588 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
38589 -mpower9-minmax by default for -mcpu=power9.
38590 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
38591 128-bit floating point.
38593 2017-01-20 Alan Modra <amodra@gmail.com>
38595 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
38596 optimizing for size.
38598 2017-01-20 Alan Modra <amodra@gmail.com>
38601 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
38602 for strcmp and strncmp from corresponding builtin decl.
38604 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
38606 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
38607 instead of i386/rtems-64.h.
38608 * config/i386/rtems-64.h: Remove.
38610 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
38614 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
38616 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
38618 2017-01-19 Tamar Christina <tamar.christina@arm.com>
38620 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
38621 Change int to HOST_WIDE_INT.
38622 * config/aarch64/aarch64-protos.h
38623 (aarch64_simd_gen_const_vector_dup): Likewise.
38624 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
38626 2017-01-19 David Malcolm <dmalcolm@redhat.com>
38628 * langhooks-def.h (lhd_type_for_size): New decl.
38629 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
38630 * langhooks.c (lhd_type_for_size): New function, taken from
38633 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
38635 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
38636 define_bypass for CR latency.
38637 (power9-cracked-alu): Update bypass latency and remove power9-branch.
38638 (power9-alu2): Add define_bypass for CR latency.
38640 (power9-mul): Update insn latency.
38641 (power9-mul-compare): Update insn latency, bypass latency and remove
38644 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38646 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
38648 * config/aarch64/aarch64.md
38649 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
38650 aarch64_nopcrelative_literal_loads.
38651 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
38653 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
38655 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
38656 TARGET_LOONGSON_3A.
38657 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
38659 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
38662 * config.gcc (supported_defaults): Add lxc1-sxc1.
38663 (with_lxc1_sxc1): Add validation.
38664 (all_defaults): Add lxc1-sxc1.
38665 * config/mips/mips.opt (mlxc1-sxc1): New option.
38666 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
38668 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
38669 __mips_no_lxc1_sxc1.
38670 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
38671 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
38672 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
38674 2017-01-19 Richard Biener <rguenther@suse.de>
38676 PR tree-optimization/72488
38677 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
38678 sure to restore SSA info.
38679 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
38681 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
38683 PR rtl-optimization/79121
38684 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
38685 of the inner type when shifting an extended value.
38687 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
38690 * symtab.c (symtab_node::equal_address_to): Fix comparing of
38691 interposable aliases.
38693 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
38696 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
38697 Use the evmergelohi instruction.
38698 (mov_si<mode>_e500_subreg4_2_le): Likewise.
38699 (mov_sitf_e500_subreg8_2_be): Likewise.
38700 (mov_sitf_e500_subreg12_2_le): Likewise.
38701 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
38702 (mov_si<mode>_e500_subreg4_2_be): Likewise.
38703 (mov_sitf_e500_subreg8_2_le): Likewise.
38704 (mov_sitf_e500_subreg12_2_be): Likewise.
38706 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38708 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
38709 attribute from vecsimple to vecperm.
38710 (altivec_vbpermq2): Likewise.
38712 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38715 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
38717 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38718 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
38719 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
38720 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
38721 case where N arg is SIZE_MAX.
38722 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
38723 (cmpstrsi): Add pattern.
38725 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
38727 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
38728 __builtin_vec_revb builtins.
38729 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
38730 built-in functions to support generation of the ISA 3.0 XXBR<x>
38731 vector byte reverse instructions.
38732 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
38733 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
38734 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
38735 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
38736 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
38737 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
38738 (P9V_BUILTIN_VEC_REVB): Likewise.
38739 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
38740 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
38741 (p9_xxbrq_v16qi): Likewise.
38742 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
38743 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
38744 (p9_xxbrh_v8hi): Likewise.
38745 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
38746 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
38747 vec_revb built-in functions.
38749 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
38751 PR rtl-optimization/78952
38752 * config/i386/i386.md (any_extract): New code iterator.
38753 (*insvqi_2): Use any_extract for source operand.
38754 (*insvqi_3): Use any_shiftrt for source operand.
38756 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
38758 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
38760 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
38762 2017-01-18 Matthias Klose <doko@ubuntu.com>
38764 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
38766 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38768 * config/rs6000/altivec.h (vec_bperm): Change #define.
38769 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
38770 (altivec_vbpermq2): New define_insn.
38771 (altivec_vbpermd): Likewise.
38772 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
38773 function interface.
38774 (VBPERMD): Likewise.
38775 (VBPERM): New polymorphic function interface.
38776 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
38777 Add entries for P9V_BUILTIN_VEC_VBPERM.
38778 * doc/extend.texi: Add interfaces for vec_bperm.
38780 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38782 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
38783 first letter of error messages.
38784 (s390_resolve_overloaded_builtin): Likewise.
38785 * config/s390/s390.c (s390_expand_builtin): Likewise.
38786 (s390_invalid_arg_for_unprototyped_fn): Likewise.
38787 (s390_valid_target_attribute_inner_p): Likewise.
38788 * config/s390/s390.md ("tabort"): Likewise.
38790 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
38792 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
38793 (ISA_AVOID_DIV_HILO): New macro.
38794 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
38795 (ISA_HAS_DDIV): Likewise.
38797 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
38799 * doc/invoke.texi (fabi-version): Correct number of occurrences.
38801 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
38803 * doc/invoke.texi (fabi-version): Spelling fix.
38805 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
38808 * doc/invoke.texi (fabi-version): Mention mangling fix for
38811 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
38814 * doc/invoke.texi (fabi-version): Document discriminator mangling.
38816 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
38819 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
38820 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
38822 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
38824 (stack_protect_test): Ditto.
38825 * config/rs6000/rs6000.opt (mstack-protector-guard=,
38826 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
38828 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
38829 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
38830 -mstack-protector-guard-offset=.
38831 (RS/6000 and PowerPC Options): Ditto.
38833 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
38835 * config/i386/i386.h (MASK_CLASS_P): New define.
38836 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
38837 there are no registers from different register sets also when
38838 mask registers are used. Update function comment.
38839 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
38840 to (*k/*r) and (*k/*km) alternatives.
38842 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
38844 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
38845 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
38846 (EH_RETURN_HANDLER_RTX): New define.
38847 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
38848 Force frame pointer in EH return functions.
38849 (aarch64_expand_epilogue): Add barrier for eh_return.
38850 (aarch64_final_eh_return_addr): Remove.
38851 (aarch64_eh_return_handler_rtx): New function.
38852 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
38854 (aarch64_eh_return_handler_rtx): New prototype.
38856 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38858 * config/rs6000/altivec.h (vec_rlmi): New #define.
38859 (vec_vrlnm): Likewise.
38860 (vec_rlnm): Likewise.
38861 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
38862 (UNSPEC_VRLNM): Likewise.
38863 (VIlong): New mode iterator.
38864 (altivec_vrl<VI_char>mi): New define_insn.
38865 (altivec_vrl<VI_char>nm): Likewise.
38866 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
38868 (VRLDNM): Likewise.
38869 (RLNM): New polymorphic function entry.
38870 (VRLWMI): New monomorphic function entry.
38871 (VRLDMI): Likewise.
38872 (RLMI): New polymorphic function entry.
38873 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
38874 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
38875 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
38878 2017-01-17 Jakub Jelinek <jakub@redhat.com>
38881 * dwarf2out.c (field_byte_offset): Restore the
38882 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
38883 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
38884 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
38887 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
38890 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
38892 2017-01-17 Jakub Jelinek <jakub@redhat.com>
38895 * dwarf2out.c (add_data_member_location_attribute): For constant
38896 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
38897 instead of DW_AT_data_member_location, DW_AT_bit_offset and
38898 DW_AT_byte_size attributes.
38900 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
38902 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
38903 after forcing to constant memory when the code model is medium.
38905 2017-01-17 Julia Koval <julia.koval@intel.com>
38908 * config/i386/avx512fintrin.h
38909 (_mm512_i32gather_ps): Change __addr type to void const*.
38910 (_mm512_mask_i32gather_ps): Ditto.
38911 (_mm512_i32gather_pd): Ditto.
38912 (_mm512_mask_i32gather_pd): Ditto.
38913 (_mm512_i64gather_ps): Ditto.
38914 (_mm512_mask_i64gather_ps): Ditto.
38915 (_mm512_i64gather_pd): Ditto.
38916 (_mm512_mask_i64gather_pd): Ditto.
38917 (_mm512_i32gather_epi32): Ditto.
38918 (_mm512_mask_i32gather_epi32): Ditto.
38919 (_mm512_i32gather_epi64): Ditto.
38920 (_mm512_mask_i32gather_epi64): Ditto.
38921 (_mm512_i64gather_epi32): Ditto.
38922 (_mm512_mask_i64gather_epi32): Ditto.
38923 (_mm512_i64gather_epi64): Ditto.
38924 (_mm512_mask_i64gather_epi64): Ditto.
38925 (_mm512_i32scatter_ps): Change __addr type to void*.
38926 (_mm512_mask_i32scatter_ps): Ditto.
38927 (_mm512_i32scatter_pd): Ditto.
38928 (_mm512_mask_i32scatter_pd): Ditto.
38929 (_mm512_i64scatter_ps): Ditto.
38930 (_mm512_mask_i64scatter_ps): Ditto.
38931 (_mm512_i64scatter_pd): Ditto.
38932 (_mm512_mask_i64scatter_pd): Ditto.
38933 (_mm512_i32scatter_epi32): Ditto.
38934 (_mm512_mask_i32scatter_epi32): Ditto.
38935 (_mm512_i32scatter_epi64): Ditto.
38936 (_mm512_mask_i32scatter_epi64): Ditto.
38937 (_mm512_i64scatter_epi32): Ditto.
38938 (_mm512_mask_i64scatter_epi32): Ditto.
38939 (_mm512_i64scatter_epi64): Ditto.
38940 (_mm512_mask_i64scatter_epi64): Ditto.
38941 * config/i386/avx512pfintrin.h
38942 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
38943 (_mm512_mask_prefetch_i32gather_ps): Ditto.
38944 (_mm512_mask_prefetch_i64gather_pd): Ditto.
38945 (_mm512_mask_prefetch_i64gather_ps): Ditto.
38946 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
38947 (_mm512_prefetch_i32scatter_ps): Ditto.
38948 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
38949 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
38950 (_mm512_prefetch_i64scatter_pd): Ditto.
38951 (_mm512_prefetch_i64scatter_ps): Ditto.
38952 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
38953 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
38954 * config/i386/avx512vlintrin.h
38955 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
38956 (_mm_mmask_i32gather_ps): Ditto.
38957 (_mm256_mmask_i32gather_pd): Ditto.
38958 (_mm_mmask_i32gather_pd): Ditto.
38959 (_mm256_mmask_i64gather_ps): Ditto.
38960 (_mm_mmask_i64gather_ps): Ditto.
38961 (_mm256_mmask_i64gather_pd): Ditto.
38962 (_mm_mmask_i64gather_pd): Ditto.
38963 (_mm256_mmask_i32gather_epi32): Ditto.
38964 (_mm_mmask_i32gather_epi32): Ditto.
38965 (_mm256_mmask_i32gather_epi64): Ditto.
38966 (_mm_mmask_i32gather_epi64): Ditto.
38967 (_mm256_mmask_i64gather_epi32): Ditto.
38968 (_mm_mmask_i64gather_epi32): Ditto.
38969 (_mm256_mmask_i64gather_epi64): Ditto.
38970 (_mm_mmask_i64gather_epi64): Ditto.
38971 (_mm256_i32scatter_ps): Change __addr type to void*.
38972 (_mm256_mask_i32scatter_ps): Ditto.
38973 (_mm_i32scatter_ps): Ditto.
38974 (_mm_mask_i32scatter_ps): Ditto.
38975 (_mm256_i32scatter_pd): Ditto.
38976 (_mm256_mask_i32scatter_pd): Ditto.
38977 (_mm_i32scatter_pd): Ditto.
38978 (_mm_mask_i32scatter_pd): Ditto.
38979 (_mm256_i64scatter_ps): Ditto.
38980 (_mm256_mask_i64scatter_ps): Ditto.
38981 (_mm_i64scatter_ps): Ditto.
38982 (_mm_mask_i64scatter_ps): Ditto.
38983 (_mm256_i64scatter_pd): Ditto.
38984 (_mm256_mask_i64scatter_pd): Ditto.
38985 (_mm_i64scatter_pd): Ditto.
38986 (_mm_mask_i64scatter_pd): Ditto.
38987 (_mm256_i32scatter_epi32): Ditto.
38988 (_mm256_mask_i32scatter_epi32): Ditto.
38989 (_mm_i32scatter_epi32): Ditto.
38990 (_mm_mask_i32scatter_epi32): Ditto.
38991 (_mm256_i32scatter_epi64): Ditto.
38992 (_mm256_mask_i32scatter_epi64): Ditto.
38993 (_mm_i32scatter_epi64): Ditto.
38994 (_mm_mask_i32scatter_epi64): Ditto.
38995 (_mm256_i64scatter_epi32): Ditto.
38996 (_mm256_mask_i64scatter_epi32): Ditto.
38997 (_mm_i64scatter_epi32): Ditto.
38998 (_mm_mask_i64scatter_epi32): Ditto.
38999 (_mm256_i64scatter_epi64): Ditto.
39000 (_mm256_mask_i64scatter_epi64): Ditto.
39001 (_mm_i64scatter_epi64): Ditto.
39002 (_mm_mask_i64scatter_epi64): Ditto.
39003 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
39004 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
39005 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
39006 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
39007 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
39008 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
39009 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
39010 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
39011 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
39012 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
39013 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
39014 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
39015 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
39016 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
39017 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
39018 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
39019 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
39020 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
39021 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
39022 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
39023 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
39024 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
39025 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
39026 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
39027 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
39028 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
39029 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
39030 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
39031 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
39032 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
39033 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
39034 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
39035 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
39036 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
39037 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
39038 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
39039 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
39040 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
39041 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
39042 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
39043 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
39044 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
39045 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
39046 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
39047 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
39048 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
39049 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
39050 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
39051 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
39052 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
39053 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
39054 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
39055 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
39056 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
39057 definitions accordingly.
39059 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
39060 Kuan-Lin Chen <kuanlinchentw@gmail.com>
39063 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
39066 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
39069 * ira-conflicts.c (ira_build_conflicts): Update total conflict
39070 hard regs for inner regno.
39072 2017-01-17 Martin Liska <mliska@suse.cz>
39075 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
39076 assumption and add comment.
39078 2017-01-17 Nathan Sidwell <nathan@acm.org>
39080 * ipa-visibility.c (localize_node): New function, broken out of ...
39081 (function_and_variable_visibility): ... here. Call it.
39083 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
39085 PR middle-end/77445
39086 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
39087 correctly set frequency of oudgoing edge.
39088 (duplicate_thread_path): Fix profile updating.
39090 2017-01-17 Jakub Jelinek <jakub@redhat.com>
39093 * configure.ac: Add GCC_BASE_VER.
39094 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
39095 version from BASE-VER file.
39096 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
39097 (gcc.o): Depend on $(BASEVER).
39098 * common.opt (dumpfullversion): New option.
39099 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
39100 * doc/invoke.texi: Document -dumpfullversion.
39101 * doc/install.texi: Document --with-gcc-major-version-only.
39102 * configure: Regenerated.
39104 2017-01-17 Richard Biener <rguenther@suse.de>
39106 PR tree-optimization/71433
39107 * tree-vrp.c (register_new_assert_for): Merge same asserts
39108 on all incoming edges.
39109 (process_assert_insertions_for): Handle insertions at the
39112 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
39114 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
39115 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
39117 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
39120 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
39123 2017-01-17 Alan Modra <amodra@gmail.com>
39126 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
39127 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
39128 symbolic stack limit when pic.
39130 2017-01-16 Martin Sebor <msebor@redhat.com>
39132 PR tree-optimization/78608
39133 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
39135 2017-01-16 Jeff Law <law@redhat.com>
39138 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
39139 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
39140 for several include directories that may be relative to sysroot.
39141 * config/i386/x-mingw32 (gplus_includedir): Define.
39142 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
39143 (native_system_includedir): Likewise.
39144 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
39145 override if TARGET_SYSTEM_ROOT is defined.
39146 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
39148 PR tree-optimization/79090
39149 PR tree-optimization/33562
39150 PR tree-optimization/61912
39151 PR tree-optimization/77485
39152 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
39153 and computed trims into the dump file.
39155 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
39157 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
39159 2017-01-16 Jakub Jelinek <jakub@redhat.com>
39162 * gimplify.c (gimplify_init_constructor): If want_value and
39163 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
39167 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
39168 sequence. Formatting fixes.
39169 (doloop_optimize): Formatting fixes.
39172 * gcc.c (debug_level_greater_than_spec_func): New function.
39173 (static_spec_functions): Add debug-level-gt spec function.
39174 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
39176 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
39177 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
39178 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
39179 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
39180 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
39181 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
39183 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
39185 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
39186 QImode fixups to general and mask registers only.
39188 2017-01-16 Carl Love <cel@us.ibm.com>
39190 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
39191 for built-in functions
39192 vector signed char vec_nabs (vector signed char)
39193 vector signed short vec_nabs (vector signed short)
39194 vector signed int vec_nabs (vector signed int)
39195 vector signed long long vec_nabs (vector signed long long)
39196 vector float vec_nabs (vector float)
39197 vector double vec_nabs (vector double)
39198 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
39200 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
39201 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
39202 * doc/extend.texi: Update the documentation file for the new built-in
39205 2017-01-16 Martin Sebor <msebor@redhat.com>
39207 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
39210 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39212 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
39213 UNSPEC_VSX__XXSPLTD to require special splat handling.
39215 2017-01-16 David Malcolm <dmalcolm@redhat.com>
39218 * system.h: Poison strndup.
39220 2017-01-16 Alan Modra <amodra@gmail.com>
39223 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
39226 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
39228 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
39230 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
39232 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
39233 call recog here. Assert that INSN_CODE (insn) is non-negative.
39235 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
39238 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
39240 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
39241 in the currently scheduled RTL fragment.
39243 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
39245 PR rtl-optimization/78751
39246 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
39249 2017-01-14 Jeff Law <law@redhat.com>
39251 PR tree-optimization/79090
39252 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
39253 variable length stores.
39254 (compute_trims): Delete dead assignment to *trim_tail.
39255 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
39258 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
39260 PR rtl-optimization/78626
39261 PR rtl-optimization/78727
39262 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
39263 of a block, and split such blocks after everything else is finished.
39265 2017-01-14 Alan Modra <amodra@gmail.com>
39268 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
39269 target legitimate_combined_insn.
39270 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
39271 (rs6000_legitimate_combined_insn): New function.
39272 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
39274 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
39275 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
39276 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
39278 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
39280 * doc/frontends.texi (G++ and GCC): Remove references to Java.
39282 2017-01-13 Jeff Law <law@redhat.com>
39284 PR tree-optimization/33562
39285 PR tree-optimization/61912
39286 PR tree-optimization/77485
39287 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
39289 (delete_dead_assignment): Likewise.
39290 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
39291 statement to delete_dead_call and delete_dead_assignment.
39293 2017-01-13 David Malcolm <dmalcolm@redhat.com>
39296 * substring-locations.c (format_warning_va): Strengthen case 1 so
39297 that both endpoints of the substring must be within the format
39298 range for just the substring to be printed.
39300 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
39302 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
39303 * config/i386/i386.c (ix86_target_string): Add missing options
39304 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
39305 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
39306 flags_other and ix86_target_other to flags2_other. Display unknown
39308 (ix86_valid_target_attribute_inner_p): Add missing options and
39309 reorder options by implied ISAs, as in ix86_target_string.
39311 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
39313 * hash-table.h (hash_table::too_empty_p): New function.
39314 (hash_table::expand): Use it.
39315 (hash_table::traverse): Likewise.
39316 (hash_table::empty_slot): Use sizeof (value_type) instead of
39317 sizeof (PTR) to convert bytes to elements. Shrink the table
39318 if the current size is excessive for the current number of
39321 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
39323 * ira-costs.c (record_reg_classes): Break from the inner loop
39324 early once alt_fail is known to be true. Update outer loop
39325 handling accordingly.
39327 2017-01-13 Jeff Law <law@redhat.com>
39329 * tree-ssa-dse.c (decrement_count): New function.
39330 (increment_start_addr, maybe_trim_memstar_call): Likewise.
39331 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
39332 when we know the partially dead statement is a mem* function.
39334 PR tree-optimization/61912
39335 PR tree-optimization/77485
39336 * tree-ssa-dse.c: Include expr.h.
39337 (maybe_trim_constructor_store): New function.
39338 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
39340 PR tree-optimization/33562
39341 PR tree-optimization/61912
39342 PR tree-optimization/77485
39343 * doc/invoke.texi: Document new dse-max-object-size param.
39344 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
39345 * tree-ssa-dse.c: Include params.h.
39346 (dse_store_status): New enum.
39347 (initialize_ao_ref_for_dse): New, partially extracted from
39349 (valid_ao_ref_for_dse, normalize_ref): New.
39350 (setup_live_bytes_from_ref, compute_trims): Likewise.
39351 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
39352 (maybe_trim_partially_dead_store): Likewise.
39353 (maybe_trim_complex_store): Likewise.
39354 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
39355 Track what bytes live from the original store. Return tri-state
39356 for dead, partially dead or live.
39357 (dse_dom_walker): Add constructor, destructor and new private members.
39358 (delete_dead_call, delete_dead_assignment): New extracted from
39360 (dse_optimize_stmt): Make a member of dse_dom_walker.
39361 Use initialize_ao_ref_for_dse.
39363 PR tree-optimization/33562
39364 PR tree-optimization/61912
39365 PR tree-optimization/77485
39366 * sbitmap.h (bitmap_count_bits): Prototype.
39367 (bitmap_clear_range, bitmap_set_range): Likewise.
39368 * sbitmap.c (bitmap_clear_range): New function.
39369 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
39371 2017-01-13 Martin Liska <mliska@suse.cz>
39374 * function.c (set_cfun): Add new argument force.
39375 * function.h (set_cfun): Likewise.
39376 * ipa-inline-transform.c (inline_call): Use the function when
39377 strict alising from is dropped for function we inline to.
39379 2017-01-13 Richard Biener <rguenther@suse.de>
39381 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
39382 for dumping GIMPLE INTEGER_CSTs.
39384 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
39386 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
39387 to 201112L since C++17.
39389 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
39392 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
39393 if -fsanitize=kernel-address is present.
39395 2017-01-13 Richard Biener <rguenther@suse.de>
39397 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
39398 as _Literal ( type ) number in case usual suffixes do not
39399 preserve all information.
39401 2017-01-13 Richard Biener <rguenther@suse.de>
39403 PR tree-optimization/77283
39404 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
39405 and ssa-iterators.h.
39406 (is_feasible_trace): Implement a cost model based on joiner
39409 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
39412 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
39413 char or short to __float128/_Float128 directly.
39415 2017-01-12 Martin Sebor <msebor@redhat.com>
39417 to -Wformat-overflow.
39418 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
39419 (min_bytes_remaining): Same.
39420 (get_string_length): Same.
39421 (format_string): Same.
39422 (format_directive): Same.
39424 (pass_sprintf_length::handle_gimple_call): Same.
39426 2017-01-12 Jakub Jelinek <jakub@redhat.com>
39428 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
39429 info.nowrite calls with no lhs that can't throw. Return bool
39430 whether gsi_remove has been called or not.
39431 (pass_sprintf_length::handle_gimple_call): Return bool whether
39432 try_substitute_return_value called gsi_remove. Formatting fix.
39433 (pass_sprintf_length::execute): Don't use gsi_remove if
39434 handle_gimple_call returned true.
39437 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
39438 be removed due to side-effects, don't remove following barrier nor
39439 turn the successor edge into fallthru edge.
39441 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39444 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
39445 element-reversing loads and stores as not swappable.
39447 2017-01-12 Nathan Sidwell <nathan@acm.org>
39448 Nicolai Stange <nicstange@gmail.com>
39450 * combine.c (try_combine): Don't ignore result of overlap checking
39451 loop. Combine overlap & asm check into single loop.
39453 2017-01-12 Richard Biener <rguenther@suse.de>
39455 * tree-pretty-print.c (dump_generic_node): Provide -gimple
39456 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
39458 2017-01-12 Richard Biener <rguenther@suse.de>
39460 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
39461 and TS_TARGET_OPTION directly derive from TS_BASE.
39462 * tree-core.h (tree_optimization_option): Derive from tree_base.
39463 (tree_target_option): Likewise.
39465 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
39467 * config/i386/i386.c (memory_address_length): Increase len
39468 only when rip_relative_addr_p returns false.
39470 2017-01-11 Julia Koval <julia.koval@intel.com>
39472 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
39473 (OPTION_MASK_ISA_SGX_SET): New.
39474 (ix86_handle_option): Handle OPT_msgx.
39475 * config.gcc: Added sgxintrin.h.
39476 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
39477 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
39478 * config/i386/i386.c (ix86_target_string): Add -msgx.
39480 (ix86_option_override_internal): Handle new options.
39481 (ix86_valid_target_attribute_inner_p): Add sgx.
39482 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
39483 * config/i386/i386.opt: Add msgx.
39484 * config/i386/sgxintrin.h: New file.
39485 * config/i386/x86intrin.h: Add sgxintrin.h.
39487 2017-01-11 Jakub Jelinek <jakub@redhat.com>
39490 * fold-const.c (maybe_nonzero_address): Return 1 for function
39492 (tree_single_nonzero_warnv_p): Don't handle function local objects
39496 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
39499 2017-01-11 David Malcolm <dmalcolm@redhat.com>
39502 * opts.c: Include "spellcheck.h"
39503 (struct string_fragment): New struct.
39504 (struct edit_distance_traits<const string_fragment &>): New
39506 (get_closest_sanitizer_option): New function.
39507 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
39509 2017-01-11 Jakub Jelinek <jakub@redhat.com>
39511 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
39513 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
39514 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
39515 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
39516 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
39517 for initial die_offset if dwarf_split_debug_info.
39518 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
39519 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
39521 (output_skeleton_debug_sections): Formatting fix. Use
39522 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
39523 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
39525 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
39527 * config/arm/cortex-a53.md: Add bypasses for
39528 cortex_a53_r2f_cvt.
39529 (cortex_a53_r2f): Only use for transfers.
39530 (cortex_a53_f2r): Likewise.
39531 (cortex_a53_r2f_cvt): Add reservation for conversions.
39532 (cortex_a53_f2r_cvt): Likewise.
39534 2017-01-11 Tamar Christina <tamar.christina@arm.com>
39536 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
39537 to all inlined functions, change static to extern.
39539 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
39542 * config/arm/arm.c (legitimize_pic_address): Handle reference to
39544 (arm_assemble_integer): Likewise.
39546 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
39548 * config.gcc: Use new awk script to check CPU, FPU and architecture
39549 parameters for --with-... options.
39550 * config/arm/parsecpu.awk: New file
39551 * config/arm/arm-cpus.in: New file.
39552 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
39554 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
39556 * config/arm/t-arm: Update dependency rules.
39557 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
39558 of processing .def files.
39559 * config/arm/genopt.sh: Deleted.
39560 * config/arm/gentune.sh: Deleted.
39561 * config/arm/arm-cores.def: Deleted.
39562 * config/arm/arm-arches.def: Deleted.
39563 * config/arm/arm-fpus.def: Deleted.
39564 * config/arm/arm-tune.md: Regenerated.
39565 * config/arm/arm-tables.opt: Regenerated.
39566 * config/arm/arm-cpu.h: New generated file.
39567 * config/arm/arm-cpu-data.h: New generated file.
39568 * config/arm/arm-cpu-cdata.h: New generated file.
39570 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
39573 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
39575 (input_varpool_node): Unpack dynamically_initialized bit.
39577 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
39579 PR rtl-optimization/79032
39580 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
39581 the alignment of the adjusted memory reference against that of MODE,
39582 instead of the alignment of the original memory reference.
39584 2017-01-11 Martin Jambor <mjambor@suse.cz>
39586 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
39588 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
39589 decorated functions.
39591 2017-01-11 Richard Biener <rguenther@suse.de>
39593 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
39594 set range/nonnull info for PHI results. Do not set it on
39595 stmts marked for removal.
39597 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
39599 * expr.c (store_field): In the bitfield case, fetch the return value
39600 from the registers before applying a single big-endian adjustment.
39601 Always do a final load for a BLKmode value not larger than a word.
39603 2017-01-10 David Malcolm <dmalcolm@redhat.com>
39606 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
39607 that we correctly handle column numbers greater than
39608 LINE_MAP_MAX_COLUMN_NUMBER.
39610 2017-01-10 Martin Sebor <msebor@redhat.com>
39612 PR middle-end/78245
39613 * gimple-ssa-sprintf.c (get_destination_size): Call
39614 {init,fini}object_sizes.
39615 * tree-object-size.c (addr_object_size): Adjust.
39616 (pass_through_call): Adjust.
39617 (pass_object_sizes::execute): Adjust.
39618 * tree-object-size.h (fini_object_sizes): Declare.
39620 2017-01-10 Martin Sebor <msebor@redhat.com>
39622 PR tree-optimization/78775
39623 * builtins.c (get_size_range): Move...
39624 * calls.c: ...to here.
39625 (alloc_max_size): Accept zero argument.
39626 (operand_signed_p): Remove.
39627 (maybe_warn_alloc_args_overflow): Call get_size_range.
39628 * calls.h (get_size_range): Declare.
39630 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
39632 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
39633 from TI's devices.csv file as of September 2016.
39634 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
39636 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
39638 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
39639 * doc/invoke.texi: Likewise.
39640 * doc/md.texi: Likewise.
39641 * doc/objc.texi: Likewise.
39643 2017-01-10 Joshua Conner <joshconner@google.com>
39645 * config/arm/fuchsia-elf.h: New file.
39646 * config/fuchsia.h: New file.
39647 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
39648 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
39650 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
39652 2016-01-10 Richard Biener <rguenther@suse.de>
39654 PR tree-optimization/79034
39655 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
39656 Propagate out degenerate PHIs in the joiner.
39658 2017-01-10 Martin Liska <mliska@suse.cz>
39660 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
39661 (sort_congruence_classes_by_decl_uid): Likewise.
39662 (sort_congruence_class_groups_by_decl_uid): Likewise.
39663 (sem_item_optimizer::merge_classes): Sort class, groups in these
39664 classes and members in the groups by DECL_UID of declarations.
39665 This would make merge operations stable.
39667 2017-01-10 Martin Liska <mliska@suse.cz>
39669 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
39670 usage of m_classes_vec.
39671 (sem_item_optimizer::~sem_item_optimizer): Likewise.
39672 (sem_item_optimizer::get_group_by_hash): Likewise.
39673 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
39674 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
39675 (sem_item_optimizer::verify_classes): Likewise.
39676 (sem_item_optimizer::process_cong_reduction): Likewise.
39677 (sem_item_optimizer::dump_cong_classes): Likewise.
39678 (sem_item_optimizer::merge_classes): Likewise.
39679 * ipa-icf.h (congruence_class_hash): Rename from
39680 congruence_class_group_hash. Remove declaration of m_classes_vec.
39682 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
39684 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
39685 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
39686 * config.gcc: Add avx512vpopcntdqintrin.h.
39687 * config/i386/avx512vpopcntdqintrin.h: New.
39688 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
39689 * config/i386/i386-builtin-types.def: Add new types.
39690 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
39691 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
39692 __builtin_ia32_vpopcountq_v8di_mask): New.
39693 * config/i386/i386-c.c (ix86_target_macros_internal): Define
39694 __AVX512VPOPCNTDQ__.
39695 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
39696 (PTA_AVX512VPOPCNTDQ): Define.
39697 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
39698 TARGET_AVX512VPOPCNTDQ_P): Define.
39699 * config/i386/i386.opt: Add mavx512vpopcntdq.
39700 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
39701 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
39703 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
39705 PR middle-end/77484
39706 * predict.def (PRED_CALL): Set to 67.
39708 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
39710 * expr.c (store_field): In the bitfield case, if the value comes from
39711 a function call and is of an aggregate type returned in registers, do
39712 not modify the field mode; extract the value in all cases if the mode
39713 is BLKmode and the size is not larger than a word.
39715 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
39718 * config/i386/cpuid.h: Fix undefined behavior.
39720 2017-01-04 Jeff Law <law@redhat.com>
39722 PR tree-optimization/79007
39723 PR tree-optimization/67955
39724 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
39725 conservative for pt.null when flag_non_call_exceptions is on.
39727 2017-01-09 Jakub Jelinek <jakub@redhat.com>
39729 PR translation/79019
39730 PR translation/79020
39731 * params.def (PARAM_INLINE_MIN_SPEEDUP,
39732 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
39733 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
39735 * config/avr/avr.opt (maccumulate-args): Likewise.
39736 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
39737 * common.opt (freport-bug): Likewise.
39738 * cif-code.def (CIF_FINAL_ERROR): Likewise.
39739 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
39740 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
39741 translatable string.
39742 * config/i386/i386.c (function_value_32): Likewise.
39743 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
39744 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
39746 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
39747 * common/config/msp430/msp430-common.c (msp430_handle_option):
39749 * symtab.c (symtab_node::verify_base): Likewise.
39750 * opts.c (set_debug_level): Likewise.
39751 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
39752 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
39753 missing whitespace to translatable strings.
39754 * config/avr/avr.md (bswapsi2): Fix typo in comment.
39755 * config/sh/superh.h: Likewise.
39756 * config/i386/xopintrin.h: Likewise.
39757 * config/i386/znver1.md: Likewise.
39758 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
39759 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
39760 * double-int.h (struct double_int): Likewise.
39761 * double-int.c (div_and_round_double): Likewise.
39762 * wide-int.cc: Likewise.
39763 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
39764 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
39765 * cfgcleanup.c (crossjumps_occured): Renamed to ...
39766 (crossjumps_occurred): ... this.
39767 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
39770 PR tree-optimization/78899
39771 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
39772 returning bool return struct loop *, NULL for failure and the new
39774 (versionable_outer_loop_p): Don't version outer loop if it has
39775 dont_vectorized bit set.
39776 (tree_if_conversion): When versioning outer loop, ensure
39777 tree_if_conversion is performed also on the inner loop of the
39778 non-vectorizable outer loop copy.
39779 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
39780 LOOP_VECTORIZED in inner loop of the scalar outer loop and
39781 prevent vectorization of it.
39782 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
39783 the outer loop vectorization of the non-scalar version is attempted
39784 before vectorization of the inner loop in scalar version. If
39785 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
39786 vectorization of its inner loop.
39787 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
39788 has 2 inner loops, rename also on edges from bb whose single pred
39789 is outer_loop->header. Fix typo in function comment.
39791 2017-01-09 Martin Sebor <msebor@redhat.com>
39794 * asan.c (asan_emit_stack_protection): Increase local buffer size
39795 to avoid snprintf truncation warning.
39797 2017-01-09 Andrew Pinski <apinski@cavium.com>
39799 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
39800 to reference thunderx2t99 for the tuning structure
39801 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
39803 (thunderx2t99_extra_costs): This.
39804 * config/aarch64/aarch64-tune.md: Regenerate.
39805 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
39806 (vulcan_addrcost_table): This.
39807 (vulcan_regmove_cost): Rename to ...
39808 (thunderx2t99_regmove_cost): This.
39809 (vulcan_vector_cost): Rename to ...
39810 (thunderx2t99_vector_cost): this.
39811 (vulcan_branch_cost): Rename to ...
39812 (thunderx2t99_branch_cost): This.
39813 (vulcan_tunings): Rename to ...
39814 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
39815 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
39817 2017-01-09 Martin Jambor <mjambor@suse.cz>
39821 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
39822 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
39823 (propagate_vr_accross_jump_function): Use the above function for all
39824 value range computations for pass-through jump functions and type
39825 converasion from explicit value range values.
39826 (ipcp_propagate_stage): Do not attempt to deduce types of formal
39827 parameters from TYPE_ARG_TYPES.
39828 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
39829 (ipa_write_node_info): Stream type of the actual argument.
39830 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
39832 2017-01-09 Martin Liska <mliska@suse.cz>
39835 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
39836 (lookup_compiler): Do not show error message with have_E.
39838 2017-01-09 Jakub Jelinek <jakub@redhat.com>
39840 PR tree-optimization/78938
39841 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
39842 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
39843 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
39844 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
39847 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39849 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
39852 2017-01-09 Richard Biener <rguenther@suse.de>
39854 PR tree-optimization/78997
39855 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
39856 name condition properly.
39858 2017-01-09 Richard Biener <rguenther@suse.de>
39861 * dwarf2out.c (is_cxx): New overload with context.
39862 (is_naming_typedef_decl): Use it.
39864 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
39866 * invoke.texi (Option Summary): Correct spacing in option lists
39867 and add line breaks to fix over-long lines.
39869 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
39871 PR middle-end/17660
39873 * extend.texi (Common Variable Attributes): Add xref to GCC
39874 Internals manual to explain mode attribute keywords.
39876 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
39879 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
39880 and Preprocessor Options.
39881 (Options for Linking): Document -pthread here....
39882 (RS/6000 and PowerPC Options): ...not here.
39883 (Solaris 2 Options): ...or here.
39884 * doc/cppopts.texi: Document -pthread.
39886 2017-01-08 Martin Sebor <msebor@redhat.com>
39888 PR middle-end/77708
39889 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
39890 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
39891 New member functions.
39892 (format_directive): Used them.
39894 (pass_sprintf_length::handle_gimple_call): Same.
39895 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
39896 to avoid truncation for any argument.
39897 (extract_affine_mul): Same.
39898 * tree.c (get_file_function_name): Same.
39900 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
39902 PR middle-end/77484
39903 * predict.def (PRED_INDIR_CALL): Set to 86.
39905 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
39907 PR preprocessor/54124
39908 * doc/cppopts.texi: Reformat -d subtable to list the full name
39909 of the options. Add cross-reference to the docs for the general
39910 compiler -d options.
39911 * doc/invoke.texi (Developer Options): Add cross-reference to the
39912 preprocessor-specific -d option documentation.
39914 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
39916 PR preprocessor/13498
39917 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
39918 redudant material, and reflect new command-line options.
39919 (System Headers): Likewise.
39921 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
39923 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
39924 -isystem, and -idirafter. Copy-edit.
39925 * doc/cppopts.texi: Copy-edit. Remove contradiction about
39926 default for -ftrack-macro-expansion. Delete obsolete and
39927 badly-formatted implementation details about -fdebug-cpp output.
39928 * doc/cppwarnopts.texi: Copy-edit.
39930 2017-01-07 David Malcolm <dmalcolm@redhat.com>
39933 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
39934 that the transition from a max line width >= 1<<10 to narrower
39935 lines works correctly.
39937 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
39939 * doc/options.texi (PerFunction): New.
39940 * opt-functions.awk (switch_flags): Map both Optimization and
39941 PerFunction to CL_OPTIMIZATION.
39942 * opth-gen.awk: Test for PerFunction flag along with
39944 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
39945 it only when the latter is present. Skip those that don't in
39946 the hash function generator.
39947 * common.opt (fvar-tracking): Mark as PerFunction instead of
39949 (fvar-tracking-assignments): Likewise.
39950 (fvar-tracking-assignments-toggle): Likewise.
39951 (fvar-tracking-uninit): Likewise.
39953 2017-01-07 Jakub Jelinek <jakub@redhat.com>
39955 PR translation/79018
39956 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
39959 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
39962 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
39963 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
39964 TARGET_LONG_JUMP_TABLE_OFFSETS.
39965 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
39966 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
39967 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
39968 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
39969 * config/m68k/m68k.md (tablejump expander): Likewise.
39970 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
39971 TARGET_LONG_JUMP_TABLE_OFFSETS.
39972 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
39973 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
39975 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
39976 David Holsgrove <david.holsgrove@xilinx.com>
39978 * common/config/microblaze/microblaze-common.c
39979 (TARGET_EXCEPT_UNWIND_INFO): Remove.
39980 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
39982 * config/microblaze/microblaze.c (microblaze_must_save_register)
39983 (microblaze_expand_epilogue, microblaze_return_addr): Handle
39985 (microblaze_eh_return): New function.
39986 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
39987 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
39988 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
39989 * config/microblaze/microblaze.md (eh_return): New pattern.
39991 2017-01-06 Jakub Jelinek <jakub@redhat.com>
39993 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
39994 GCC_DIAGNOSTIC_STRINGIFY): Define.
39996 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
39998 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40000 * config/arm/arm.md (<mcrr>): New.
40002 * config/arm/arm.c (arm_arch5te): New.
40003 (arm_option_override): Set arm_arch5te.
40004 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
40006 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
40007 (arm_mcrr_qualifiers): ... this. New.
40008 (MRRC_QUALIFIERS): Define to...
40009 (arm_mrrc_qualifiers): ... this. New.
40010 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
40012 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
40013 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
40014 (MRRCI, mrrc, MRRC): New.
40015 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
40016 VUNSPEC_MRRC2): New.
40018 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40020 * config/arm/arm.md (<mcr>): New.
40022 * config/arm/arm.c (arm_coproc_builtin_available): Add
40023 support for mcr, mrc, mcr2 and mrc2.
40024 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
40025 (arm_mcr_qualifiers): ... this. New.
40026 (MRC_QUALIFIERS): Define to ...
40027 (arm_mrc_qualifiers): ... this. New.
40028 (MCR_QUALIFIERS): Define to ...
40029 (arm_mcr_qualifiers): ... this. New.
40030 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
40032 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
40033 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
40034 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
40035 VUNSPEC_MRC2): New.
40037 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40039 * config/arm/arm.md (*ldc): New.
40043 * config/arm/arm.c (arm_coproc_builtin_available): Add
40044 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
40045 (arm_coproc_ldc_stc_legitimate_address): New.
40046 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
40047 'qualifier_const_pointer'.
40048 (LDC_QUALIFIERS): Define to...
40049 (arm_ldc_qualifiers): ... this. New.
40050 (STC_QUALIFIERS): Define to...
40051 (arm_stc_qualifiers): ... this. New.
40052 * config/arm/arm-protos.h
40053 (arm_coproc_ldc_stc_legitimate_address): New.
40054 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
40055 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
40056 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
40057 stc2, stcl, stc2l): New.
40058 * config/arm/constraints.md (Uz): New.
40059 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
40060 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
40061 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
40062 VUNSPEC_STC2L): New.
40064 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40066 * config/arm/arm.md (<cdp>): New.
40067 * config/arm/arm.c (neon_const_bounds): Rename this ...
40068 (arm_const_bounds): ... this.
40069 (arm_coproc_builtin_available): New.
40070 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
40071 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
40072 (CDP_QUALIFIERS): Define to...
40073 (arm_cdp_qualifiers): ... this. New.
40075 (arm_expand_builtin_args): Add case for 6 arguments.
40076 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
40077 (arm_const_bounds): ... this.
40078 (arm_coproc_builtin_available): New.
40079 * config/arm/arm_acle.h (__arm_cdp): New.
40081 * config/arm/arm_acle_builtins.def (cdp): New.
40083 * config/arm/iterators.md (CDPI,CDP,cdp): New.
40084 * config/arm/neon.md: Rename all 'neon_const_bounds' to
40085 'arm_const_bounds'.
40086 * config/arm/types.md (coproc): New.
40087 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
40088 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
40089 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
40090 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
40092 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40094 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
40095 (UBINOP_QUALIFIERS): New.
40097 (acle_builtin_data): New. Change comment.
40098 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
40099 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
40100 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
40101 arm_acle_builtins.def.
40102 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
40103 (arm_init_acle_builtins): New.
40104 (CRC32_BUILTIN): Remove.
40105 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
40106 crc32cb, crc32ch and crc32cw.
40107 (arm_init_crc32_builtins): Remove.
40108 (arm_init_builtins): Use arm_init_acle_builtins rather
40109 than arm_init_crc32_builtins.
40110 (arm_expand_acle_builtin): New.
40111 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
40112 * config/arm/arm_acle_builtins.def: New.
40114 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40116 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
40117 (arm_builtin_datum): ... this.
40118 (arm_init_neon_builtin): Rename to ...
40119 (arm_init_builtin): ... this. Add a new parameters PREFIX
40120 and USE_SIG_IN_NAME.
40121 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
40122 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
40123 'arm_builtin_datum'.
40124 (arm_init_vfp_builtins): Likewise.
40125 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
40126 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
40127 (arm_expand_neon_args): Rename to ...
40128 (arm_expand_builtin_args): ... this. Rename builtin_arg
40129 enum values and differentiate between ARG_BUILTIN_MEMORY
40130 and ARG_BUILTIN_NEON_MEMORY.
40131 (arm_expand_neon_builtin_1): Rename to ...
40132 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
40133 values, arm_expand_builtin_args and add bool parameter NEON.
40134 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
40135 (arm_expand_vfp_builtin): Likewise.
40136 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
40138 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40140 PR middle-end/77484
40141 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
40142 * predict.c (tree_estimate_probability_bb): Reverse direction of
40143 polymorphic call predictor.
40145 2017-01-06 David Malcolm <dmalcolm@redhat.com>
40147 * passes.c (execute_one_pass): Split out pass-skipping logic into...
40148 (determine_pass_name_match): ...this new function and...
40149 (should_skip_pass_p): ...this new function.
40151 2017-01-06 Nathan Sidwell <nathan@acm.org>
40153 * ipa-visibility.c (function_and_variable_visibility): Reformat
40154 comments and long lines. Remove extrneous if.
40155 * symtab.c (symtab_node::make_decl_local): Fix code format.
40156 (symtab_node::set_section_for_node): Fix comment typo.
40158 2017-01-06 Martin Liska <mliska@suse.cz>
40161 * lra-constraints.c: Rename invariant to lra_invariant.
40162 * predict.c (set_even_probabilities): Initialize e to NULL.
40164 2017-01-05 Martin Sebor <msebor@redhat.com>
40166 PR tree-optimization/78910
40167 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
40168 (format_integer): Correct off-by-one error in the handling
40169 of precision with negative numbers in signed conversions..
40171 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
40173 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
40175 2017-01-05 Jakub Jelinek <jakub@redhat.com>
40177 PR tree-optimization/71016
40178 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
40179 factor_out_conditional_conversion. Formatting fix.
40180 (factor_out_conditional_conversion): Add cond_stmt argument.
40181 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
40182 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
40185 2017-01-05 David Malcolm <dmalcolm@redhat.com>
40187 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
40188 read-rtl-function.o, and selftest-rtl.o.
40189 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
40190 (selftest::aarch64_test_loading_full_dump): New function.
40191 (selftest::aarch64_run_selftests): New function.
40192 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
40193 selftest::aarch64_run_selftests.
40194 * config/i386/i386.c
40195 (selftest::ix86_test_loading_dump_fragment_1): New function.
40196 (selftest::ix86_test_loading_call_insn): New function.
40197 (selftest::ix86_test_loading_full_dump): New function.
40198 (selftest::ix86_test_loading_unspec): New function.
40199 (selftest::ix86_run_selftests): Call the new functions.
40200 * emit-rtl.c (maybe_set_max_label_num): New function.
40201 * emit-rtl.h (maybe_set_max_label_num): New decl.
40202 * function.c (instantiate_decls): Guard call to
40203 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
40204 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
40206 * gensupport.c (gen_reader::gen_reader): Pass "false"
40207 for new "compact" param of rtx_reader.
40208 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
40209 rather than an empty string for NULL strings.
40210 * read-md.c: Potentially include config.h rather than bconfig.h.
40211 Wrap include of errors.h with #ifdef GENERATOR_FILE.
40212 (have_error): New global, copied from errors.c.
40213 (md_reader::read_name): Rename to...
40214 (md_reader::read_name_1): ...this, adding "out_loc" param,
40215 and converting "missing name or number" to returning false, rather
40217 (md_reader::read_name): Reimplement in terms of read_name_1.
40218 (md_reader::read_name_or_nil): New function.
40219 (md_reader::read_string): Handle "(nil)" by returning NULL.
40220 (md_reader::md_reader): Add new param "compact".
40221 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
40222 (md_reader::read_file): New method.
40223 * read-md.h (md_reader::md_reader): Add new param "compact".
40224 (md_reader::read_file): New method.
40225 (md_reader::is_compact): New accessor.
40226 (md_reader::read_name): Convert return type from void to file_location.
40227 (md_reader::read_name_or_nil): New decl.
40228 (md_reader::read_name_1): New decl.
40229 (md_reader::m_compact): New field.
40230 (noop_reader::noop_reader): Pass "false" for new "compact" param
40232 (rtx_reader::rtx_reader): Add new "compact" param.
40233 (rtx_reader::read_rtx_operand): Make virtual and convert return
40234 type from void to rtx.
40235 (rtx_reader::read_until): New decl.
40236 (rtx_reader::handle_any_trailing_information): New virtual function.
40237 (rtx_reader::postprocess): New virtual function.
40238 (rtx_reader::finalize_string): New virtual function.
40239 (rtx_reader::m_in_call_function_usage): New field.
40240 (rtx_reader::m_reuse_rtx_by_id): New field.
40241 * read-rtl-function.c: New file.
40242 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
40243 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
40244 (selftest::verify_three_block_rtl_cfg): New decl.
40245 * read-rtl-function.h: New file.
40246 * read-rtl.c: Potentially include config.h rather than bconfig.h.
40247 For host, include function.h, memmodel.h, and emit-rtl.h.
40248 (one_time_initialization): New function.
40249 (struct compact_insn_name): New struct.
40250 (compact_insn_names): New array.
40251 (find_code): Handle insn codes in compact dumps.
40252 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
40253 (bind_subst_iter_and_attr): Likewise.
40254 (add_condition_to_string): Likewise.
40255 (add_condition_to_rtx): Likewise.
40256 (apply_attribute_uses): Likewise.
40257 (add_current_iterators): Likewise.
40258 (apply_iterators): Likewise.
40259 (initialize_iterators): Guard usage of apply_subst_iterator with
40260 #ifdef GENERATOR_FILE.
40261 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
40262 (md_reader::read_mapping): Likewise.
40263 (add_define_attr_for_define_subst): Likewise.
40264 (add_define_subst_attr): Likewise.
40265 (read_subst_mapping): Likewise.
40266 (check_code_iterator): Likewise.
40267 (rtx_reader::read_rtx): Likewise. Move one-time initialization
40269 (one_time_initialization): New function.
40270 (rtx_reader::read_until): New method.
40271 (read_flags): New function.
40272 (parse_reg_note_name): New function.
40273 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
40274 Handle reuse_rtx ids.
40275 Wrap iterator lookup within #ifdef GENERATOR_FILE.
40276 Add parsing support for RTL dumps, mirroring the special-cases in
40277 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
40278 values, and calling handle_any_trailing_information.
40279 (rtx_reader::read_rtx_operand): Convert return type from void
40280 to rtx, returning return_rtx. Handle case 'e'. Call
40281 finalize_string on XSTR and XTMPL fields.
40282 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
40283 "(nil)" values were omitted. Call the postprocess vfunc on the
40285 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
40286 class ctor. Initialize m_in_call_function_usage. Call
40287 one_time_initialization.
40288 * rtl-tests.c (selftest::test_uncond_jump): Call
40289 set_new_first_and_last_insn.
40290 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
40291 * selftest-rtl.c: New file.
40292 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
40293 (selftest::get_insn_by_uid): New decl.
40294 * selftest-run-tests.c (selftest::run_tests): Call
40295 read_rtl_function_c_tests.
40296 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
40297 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
40300 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
40302 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
40303 operands in a special way. Assert that pos+len <= mode precision.
40305 2017-01-05 Jakub Jelinek <jakub@redhat.com>
40307 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
40308 3 argument Alias with unlimited for the negative form.
40309 (fno-vect-cost-model): Removed.
40311 2017-01-05 Martin Liska <mliska@suse.cz>
40313 * hsa-gen.c (gen_hsa_divmod): New function.
40314 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
40316 2017-01-05 Martin Liska <mliska@suse.cz>
40319 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
40322 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
40324 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
40325 small constant length operands.
40327 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
40329 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
40330 between loop iterations.
40332 2017-01-05 Martin Liska <mliska@suse.cz>
40335 * gimplify.c (gimplify_decl_expr): Compare to
40336 asan_poisoned_variables instread of checking flags.
40337 (gimplify_target_expr): Likewise.
40338 (gimplify_expr): Likewise.
40339 (gimplify_function_tree): Conditionally initialize
40340 asan_poisoned_variables.
40342 2017-01-04 Jeff Law <law@redhat.com>
40344 PR tree-optimizatin/78812
40345 * rtl.h (contains_mem_rtx_p): Prototype.
40346 * ifcvt.c (containts_mem_rtx_p): Move from here to...
40347 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
40348 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
40349 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
40350 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
40352 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
40354 * input.c (assert_char_at_range): Default-initialize actual_range.
40356 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
40358 * df-scan.c (df_ref_create_structure): Make regno unsigned,
40359 to match the caller.
40361 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
40363 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
40364 insns after final jump in test to emit dummy move.
40366 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
40368 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
40369 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
40371 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
40373 * multiple_target.c (create_dispatcher_calls): Init e_next.
40374 * tree-ssa-loop-split.c (split_loop): Init border.
40375 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
40378 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
40383 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
40384 (altivec_register_operand): Do not return true if the operand
40385 contains a SUBREG mixing SImode and SFmode.
40386 (vsx_register_operand): Likewise.
40387 (vsx_reg_sfsubreg_ok): New predicate.
40388 (vfloat_operand): Do not return true if the operand contains a
40389 SUBREG mixing SImode and SFmode.
40390 (vint_operand): Likewise.
40391 (vlogical_operand): Likewise.
40392 (gpc_reg_operand): Likewise.
40393 (int_reg_operand): Likewise.
40394 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
40395 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
40396 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
40398 (rs6000_emit_move_si_sf_subreg): New helper function.
40399 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
40400 fixup SUBREGs involving SImode and SFmode.
40401 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
40402 numbers for the new peephole2 optimization.
40403 (peephole2 for SFmode unions): New peephole2 to optimize cases in
40404 the GLIBC math library that do AND/IOR/XOR operations on single
40405 precision floating point.
40406 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
40407 target macros to say whether we need to avoid SUBREGs mixing
40409 (TARGET_ALLOW_SF_SUBREG): Likewise.
40410 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
40411 (UNSPEC_SI_FROM_SF): Likewise.
40412 (iorxor): Change spacing.
40413 (and_ior_xor): New iterator for AND, IOR, and XOR.
40414 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
40415 (movdi_from_sf_zero_ext): Likewise.
40416 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
40417 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
40418 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
40419 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
40420 (fms<mode>4): Likewise.
40421 (fnma<mode>4): Likewise.
40422 (fnms<mode>4): Likewise.
40423 (nfma<mode>4): Likewise.
40424 (nfms<mode>4): Likewise.
40426 2017-01-04 Marek Polacek <polacek@redhat.com>
40429 * doc/invoke.texi: Document -Wpointer-compare.
40431 2017-01-04 Jakub Jelinek <jakub@redhat.com>
40433 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
40436 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
40437 descriptions for -gdwarf-5 and emit them as uleb128 instead of
40440 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
40443 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
40444 documentation of the powerpc_popcntb_ok attribute.
40445 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
40446 code to issue warning messages if a requested CPU configuration is
40447 not supported by the binary (assembler and loader) toolchain.
40448 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
40449 made to define a built-in function that has been disabled.
40450 (paired_init_builtins): Add assertion to prevent ICE if attempt is
40451 made to define a built-in function that has been disabled.
40452 (altivec_init_builtins): Add comment explaining why definition
40453 of the DST built-in functions is not preceded by an assertion
40454 check. Add assertions to prevent ICE if attempts are made to
40455 define an altivec predicate or an abs* built-in function that has
40457 (htm_init_builtins): Add comment explaining why definition of the
40458 htm built-in functions is not preceded by an assertion check.
40460 2017-01-04 Jeff Law <law@redhat.com>
40462 PR tree-optimizatin/67955
40463 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
40464 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
40465 the points-to solution does not include pt_null. Use DECL_PT_UID
40468 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
40470 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
40471 Use gen_int_mode instead of gen_lopwart for const_int operands.
40473 2017-01-04 Jakub Jelinek <jakub@redhat.com>
40475 PR tree-optimization/71563
40476 * match.pd: Simplify X << Y into X if Y is known to be 0 or
40477 out of range value - has low bits known to be zero.
40479 2017-01-04 Alan Modra <amodra@gmail.com>
40481 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
40482 * configure: Regenerate.
40483 * config.in: Regenerate.
40485 2017-01-04 Jakub Jelinek <jakub@redhat.com>
40488 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
40489 a substring of the message, but strcmp with the whole message. Ifdef
40490 ENABLE_NLS, translate the message first using dgettext.
40492 2017-01-03 Jeff Law <law@redhat.com>
40494 PR tree-optimizatin/78856
40495 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
40496 (mark_threaded_blocks): Remove code to truncate thread paths that
40497 cross multiple loop headers. Instead invalidate the cached loop
40498 iteration information and handle case of a thread path walking
40499 into an irreducible region.
40501 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
40504 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
40505 assertions. Add support for doing the signbit if the IEEE 128-bit
40506 floating point value is in a GPR.
40507 * config/rs6000/rs6000.md (Fsignbit): Delete.
40508 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
40509 Update the length attribute if the value is in a GPR.
40510 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
40511 the sign or zero extension instruction, since the value is always 0/1.
40512 (signbit<mode>2_dm2): Delete using <Fsignbit>.
40515 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
40516 extracting SImode to a GPR register so that we can generate a
40517 store, limit the vector to be in a traditional Altivec register
40518 for the vextuwrx instruction.
40520 2017-01-03 Ian Lance Taylor <iant@google.com>
40522 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
40524 2017-01-03 Martin Sebor <msebor@redhat.com>
40526 PR tree-optimization/78696
40527 * gimple-ssa-sprintf.c (format_floating): Correct handling of
40528 precision. Use MPFR for %f for greater fidelity. Correct handling
40530 (pass_sprintf_length::compute_format_length): Set width and precision
40531 specified by asrerisk to void_node for vararg functions.
40532 (try_substitute_return_value): Adjust dump output.
40534 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
40536 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
40538 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
40540 * doc/invoke.texi (SPARC options): Document -mlra as the default.
40541 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
40542 -mlra/-mno-lra was passed to the compiler.
40544 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
40546 PR rtl-optimization/65618
40547 * emit-rtl.c (try_split): Move initialization of "before" and
40548 "after" to just before the call to emit_insn_after_setloc.
40550 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
40552 * doc/md.texi (Standard Names): Remove reference to Java frontend.
40554 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
40556 * dwarf2out.c (gen_enumeration_type_die): When
40557 -gno-strict-dwarf, add a DW_AT_encoding attribute.
40559 2017-01-03 Jakub Jelinek <jakub@redhat.com>
40561 PR tree-optimization/78965
40562 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
40563 Change first argument from const call_info & to call_info &. For %n
40564 set info.nowrite to false.
40566 PR middle-end/78901
40567 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
40568 possibly throwing calls.
40570 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
40571 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
40572 and fns handling, rather than in a separate case SSA_NAME.
40574 2017-01-02 Jeff Law <law@redhat.com>
40576 * config/darwin-driver.c (darwin_driver_init): Const-correctness
40577 fixes for first_period and second_period variables.
40579 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
40582 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
40583 (*insvqi_1): New insn pattern.
40584 (*insvqi_1_mem_rex64): Ditto.
40585 (*insvqi_2): Ditto.
40586 (*insvqi_3): Rename from *insvqi.
40588 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
40590 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
40592 * doc/cfg.texi (Edges): Remove reference to Java.
40593 (Maintaining the CFG): Ditto.
40595 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40597 PR middle-end/77674
40598 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
40599 transparent aliases.
40601 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40603 PR middle-end/77484
40604 * predict.def (PRED_CALL): Update hitrate.
40605 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
40606 * predict.c (tree_estimate_probability_bb): Split CALL predictor
40607 into direct/indirect/polymorphic variants.
40609 2017-01-01 Jakub Jelinek <jakub@redhat.com>
40611 Update copyright years.
40613 * gcc.c (process_command): Update copyright notice dates.
40614 * gcov-dump.c (print_version): Ditto.
40615 * gcov.c (print_version): Ditto.
40616 * gcov-tool.c (print_version): Ditto.
40617 * gengtype.c (create_file): Ditto.
40618 * doc/cpp.texi: Bump @copying's copyright year.
40619 * doc/cppinternals.texi: Ditto.
40620 * doc/gcc.texi: Ditto.
40621 * doc/gccint.texi: Ditto.
40622 * doc/gcov.texi: Ditto.
40623 * doc/install.texi: Ditto.
40624 * doc/invoke.texi: Ditto.
40626 Copyright (C) 2017 Free Software Foundation, Inc.
40628 Copying and distribution of this file, with or without modification,
40629 are permitted in any medium without royalty provided the copyright
40630 notice and this notice are preserved.