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-dfa.h (get_addr_base_and_unit_offset_1): Return the offset
6 as a poly_int64_pod rather than a HOST_WIDE_INT.
7 (get_addr_base_and_unit_offset): Likewise.
8 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
9 (get_addr_base_and_unit_offset): Likewise.
10 * doc/match-and-simplify.texi: Change off from HOST_WIDE_INT
11 to poly_int64 in example.
12 * fold-const.c (fold_binary_loc): Update call to
13 get_addr_base_and_unit_offset.
14 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
15 (maybe_canonicalize_mem_ref_addr): Likewise.
16 (gimple_fold_stmt_to_constant_1): Likewise.
17 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
19 * ipa-param-manipulation.c (ipa_modify_call_arguments): Likewise.
21 * omp-low.c (lower_omp_target): Likewise.
22 * tree-sra.c (build_ref_for_offset): Likewise.
23 (build_debug_ref_for_model): Likewise.
24 * tree-ssa-address.c (maybe_fold_tmr): Likewise.
25 * tree-ssa-alias.c (ao_ref_init_from_ptr_and_size): Likewise.
26 * tree-ssa-ccp.c (optimize_memcpy): Likewise.
27 * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Likewise.
28 (constant_pointer_difference): Likewise.
29 * tree-ssa-loop-niter.c (expand_simple_operations): Likewise.
30 * tree-ssa-phiopt.c (jump_function_from_stmt): Likewise.
31 * tree-ssa-pre.c (create_component_ref_by_pieces_1): Likewise.
32 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
33 (vn_reference_maybe_forwprop_address, vn_reference_lookup_3): Likewise.
34 (set_ssa_val_to): Likewise.
35 * tree-ssa-strlen.c (get_addr_stridx, addr_stridxptr)
36 (maybe_diag_stxncpy_trunc): Likewise.
37 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
38 * tree.c (build_simple_mem_ref_loc): Likewise.
39 (array_at_struct_end_p): Likewise.
41 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
42 Alan Hayward <alan.hayward@arm.com>
43 David Sherwood <david.sherwood@arm.com>
45 * tree-dfa.h (get_ref_base_and_extent): Return the base, size and
46 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
47 (get_ref_base_and_extent_hwi): Declare.
48 * tree-dfa.c (get_ref_base_and_extent): Return the base, size and
49 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
50 (get_ref_base_and_extent_hwi): New function.
51 * cfgexpand.c (expand_debug_expr): Update call to
52 get_ref_base_and_extent.
53 * dwarf2out.c (add_var_loc_to_decl): Likewise.
54 * gimple-fold.c (get_base_constructor): Return the offset as a
55 poly_int64_pod rather than a HOST_WIDE_INT.
56 (fold_const_aggregate_ref_1): Track polynomial sizes and offsets.
57 * ipa-polymorphic-call.c
58 (ipa_polymorphic_call_context::set_by_invariant)
59 (extr_type_from_vtbl_ptr_store): Track polynomial offsets.
60 (ipa_polymorphic_call_context::ipa_polymorphic_call_context)
61 (check_stmt_for_type_change): Use get_ref_base_and_extent_hwi
62 rather than get_ref_base_and_extent.
63 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
64 * ipa-prop.c (ipa_load_from_parm_agg, compute_complex_assign_jump_func)
65 (get_ancestor_addr_info, determine_locally_known_aggregate_parts):
67 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Update
68 call to get_ref_base_and_extent.
69 * tree-sra.c (create_access, get_access_for_expr): Likewise.
70 * tree-ssa-alias.c (ao_ref_base, aliasing_component_refs_p)
71 (stmt_kills_ref_p): Likewise.
72 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
73 * tree-ssa-scopedtables.c (avail_expr_hash, equal_mem_array_ref_p):
75 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
76 Use get_ref_base_and_extent_hwi rather than get_ref_base_and_extent
77 when calling native_encode_expr.
78 * tree-ssa-structalias.c (get_constraint_for_component_ref): Update
79 call to get_ref_base_and_extent.
80 (do_structure_copy): Use get_ref_base_and_extent_hwi rather than
81 get_ref_base_and_extent.
82 * var-tracking.c (track_expr_p): Likewise.
84 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
85 Alan Hayward <alan.hayward@arm.com>
86 David Sherwood <david.sherwood@arm.com>
88 * ipa-param-manipulation.h (ipa_parm_adjustment::offset): Change from
89 HOST_WIDE_INT to poly_int64_pod.
90 * ipa-param-manipulation.c (ipa_modify_call_arguments): Track
91 polynomail parameter offsets.
93 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
94 Alan Hayward <alan.hayward@arm.com>
95 David Sherwood <david.sherwood@arm.com>
97 * gengtype.c (main): Handle poly_int64_pod.
98 * dwarf2out.h (dw_cfi_oprnd_cfa_loc): New dw_cfi_oprnd_type.
99 (dw_cfi_oprnd::dw_cfi_cfa_loc): New field.
100 (dw_cfa_location::offset, dw_cfa_location::base_offset): Change
101 from HOST_WIDE_INT to poly_int64_pod.
102 * dwarf2cfi.c (queued_reg_save::cfa_offset): Likewise.
103 (copy_cfa): New function.
104 (lookup_cfa_1): Use the cached dw_cfi_cfa_loc, if it exists.
105 (cfi_oprnd_equal_p): Handle dw_cfi_oprnd_cfa_loc.
106 (cfa_equal_p, dwarf2out_frame_debug_adjust_cfa)
107 (dwarf2out_frame_debug_cfa_offset, dwarf2out_frame_debug_expr)
108 (initial_return_save): Treat offsets as poly_ints.
109 (def_cfa_0): Likewise. Cache the CFA in dw_cfi_cfa_loc if either
110 offset is nonconstant.
111 (reg_save): Take the offset as a poly_int64. Fall back to
112 DW_CFA_expression for nonconstant offsets.
113 (queue_reg_save): Take the offset as a poly_int64.
114 * dwarf2out.c (dw_cfi_oprnd2_desc): Handle DW_CFA_def_cfa_expression.
116 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
117 Alan Hayward <alan.hayward@arm.com>
118 David Sherwood <david.sherwood@arm.com>
120 * rtl.h (operand_subword, operand_subword_force): Take the offset
121 as a poly_uint64 an unsigned int.
122 * emit-rtl.c (operand_subword, operand_subword_force): Likewise.
124 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
125 Alan Hayward <alan.hayward@arm.com>
126 David Sherwood <david.sherwood@arm.com>
128 * doc/rtl.texi: Update documentation of SUBREG_BYTE. Document the
129 'p' format code. Use INT_LIST rather than SUBREG as the example of
130 a code with an XINT and an XEXP. Remove the implication that
131 accessing an rtx field using XINT is expected to work.
132 * rtl.def (SUBREG): Change format from "ei" to "ep".
133 * rtl.h (rtunion::rt_subreg): New field.
134 (XCSUBREG): New macro.
135 (SUBREG_BYTE): Use it.
136 (subreg_shape): Change offset from an unsigned int to a poly_uint16.
137 Update constructor accordingly.
138 (subreg_shape::operator ==): Update accordingly.
139 (subreg_shape::unique_id): Return an unsigned HOST_WIDE_INT rather
140 than an unsigned int.
141 (subreg_lsb, subreg_lowpart_offset, subreg_highpart_offset): Return
142 a poly_uint64 rather than an unsigned int.
143 (subreg_lsb_1): Likewise. Take the offset as a poly_uint64 rather
144 than an unsigned int.
145 (subreg_size_offset_from_lsb, subreg_size_lowpart_offset)
146 (subreg_size_highpart_offset): Return a poly_uint64 rather than
147 an unsigned int. Take the sizes as poly_uint64s.
148 (subreg_offset_from_lsb): Return a poly_uint64 rather than
149 an unsigned int. Take the shift as a poly_uint64 rather than
151 (subreg_regno_offset, subreg_offset_representable_p): Take the offset
152 as a poly_uint64 rather than an unsigned int.
153 (simplify_subreg_regno): Likewise.
154 (byte_lowpart_offset): Return the memory offset as a poly_int64
156 (subreg_memory_offset): Likewise. Take the subreg offset as a
157 poly_uint64 rather than an unsigned int.
158 (simplify_subreg, simplify_gen_subreg, subreg_get_info)
159 (gen_rtx_SUBREG, validate_subreg): Take the subreg offset as a
160 poly_uint64 rather than an unsigned int.
161 * rtl.c (rtx_format): Describe 'p' in comment.
162 (copy_rtx, rtx_equal_p_cb, rtx_equal_p): Handle 'p'.
163 * emit-rtl.c (validate_subreg, gen_rtx_SUBREG): Take the subreg
164 offset as a poly_uint64 rather than an unsigned int.
165 (byte_lowpart_offset): Return the memory offset as a poly_int64
167 (subreg_memory_offset): Likewise. Take the subreg offset as a
168 poly_uint64 rather than an unsigned int.
169 (subreg_size_lowpart_offset, subreg_size_highpart_offset): Take the
170 mode sizes as poly_uint64s rather than unsigned ints. Return a
171 poly_uint64 rather than an unsigned int.
172 (subreg_lowpart_p): Treat subreg offsets as poly_ints.
173 (copy_insn_1): Handle 'p'.
174 * rtlanal.c (set_noop_p): Treat subregs offsets as poly_uint64s.
175 (subreg_lsb_1): Take the subreg offset as a poly_uint64 rather than
176 an unsigned int. Return the shift in the same way.
177 (subreg_lsb): Return the shift as a poly_uint64 rather than an
179 (subreg_size_offset_from_lsb): Take the sizes and shift as
180 poly_uint64s rather than unsigned ints. Return the offset as
182 (subreg_get_info, subreg_regno_offset, subreg_offset_representable_p)
183 (simplify_subreg_regno): Take the offset as a poly_uint64 rather than
185 * rtlhash.c (add_rtx): Handle 'p'.
186 * genemit.c (gen_exp): Likewise.
187 * gengenrtl.c (type_from_format, gendef): Likewise.
188 * gensupport.c (subst_pattern_match, get_alternatives_number)
189 (collect_insn_data, alter_predicate_for_insn, alter_constraints)
190 (subst_dup): Likewise.
191 * gengtype.c (adjust_field_rtx_def): Likewise.
192 * genrecog.c (find_operand, find_matching_operand, validate_pattern)
193 (match_pattern_2): Likewise.
194 (rtx_test::SUBREG_FIELD): New rtx_test::kind_enum.
195 (rtx_test::subreg_field): New function.
196 (operator ==, safe_to_hoist_p, transition_parameter_type)
197 (print_nonbool_test, print_test): Handle SUBREG_FIELD.
198 * genattrtab.c (attr_rtx_1): Say that 'p' is deliberately not handled.
199 * genpeep.c (match_rtx): Likewise.
200 * print-rtl.c (print_poly_int): Include if GENERATOR_FILE too.
201 (rtx_writer::print_rtx_operand): Handle 'p'.
202 (print_value): Handle SUBREG.
203 * read-rtl.c (apply_int_iterator): Likewise.
204 (rtx_reader::read_rtx_operand): Handle 'p'.
205 * alias.c (rtx_equal_for_memref_p): Likewise.
206 * cselib.c (rtx_equal_for_cselib_1, cselib_hash_rtx): Likewise.
207 * caller-save.c (replace_reg_with_saved_mem): Treat subreg offsets
209 * calls.c (expand_call): Likewise.
210 * combine.c (combine_simplify_rtx, expand_field_assignment): Likewise.
211 (make_extraction, gen_lowpart_for_combine): Likewise.
212 * loop-invariant.c (hash_invariant_expr_1, invariant_expr_equal_p):
214 * cse.c (remove_invalid_subreg_refs): Take the offset as a poly_uint64
215 rather than an unsigned int. Treat subreg offsets as poly_ints.
216 (exp_equiv_p): Handle 'p'.
217 (hash_rtx_cb): Likewise. Treat subreg offsets as poly_ints.
218 (equiv_constant, cse_insn): Treat subreg offsets as poly_ints.
219 * dse.c (find_shift_sequence): Likewise.
220 * dwarf2out.c (rtl_for_decl_location): Likewise.
221 * expmed.c (extract_low_bits): Likewise.
222 * expr.c (emit_group_store, undefined_operand_subword_p): Likewise.
223 (expand_expr_real_2): Likewise.
224 * final.c (alter_subreg): Likewise.
225 (leaf_renumber_regs_insn): Handle 'p'.
226 * function.c (assign_parm_find_stack_rtl, assign_parm_setup_stack):
227 Treat subreg offsets as poly_ints.
228 * fwprop.c (forward_propagate_and_simplify): Likewise.
229 * ifcvt.c (noce_emit_move_insn, noce_emit_cmove): Likewise.
230 * ira.c (get_subreg_tracking_sizes): Likewise.
231 * ira-conflicts.c (go_through_subreg): Likewise.
232 * ira-lives.c (process_single_reg_class_operands): Likewise.
233 * jump.c (rtx_renumbered_equal_p): Likewise. Handle 'p'.
234 * lower-subreg.c (simplify_subreg_concatn): Take the subreg offset
235 as a poly_uint64 rather than an unsigned int.
236 (simplify_gen_subreg_concatn, resolve_simple_move): Treat
237 subreg offsets as poly_ints.
238 * lra-constraints.c (operands_match_p): Handle 'p'.
239 (match_reload, curr_insn_transform): Treat subreg offsets as poly_ints.
240 * lra-spills.c (assign_mem_slot): Likewise.
241 * postreload.c (move2add_valid_value_p): Likewise.
242 * recog.c (general_operand, indirect_operand): Likewise.
243 * regcprop.c (copy_value, maybe_mode_change): Likewise.
244 (copyprop_hardreg_forward_1): Likewise.
245 * reginfo.c (simplifiable_subregs_hasher::hash, simplifiable_subregs)
246 (record_subregs_of_mode): Likewise.
247 * rtlhooks.c (gen_lowpart_general, gen_lowpart_if_possible): Likewise.
248 * reload.c (operands_match_p): Handle 'p'.
249 (find_reloads_subreg_address): Treat subreg offsets as poly_ints.
250 * reload1.c (alter_reg, choose_reload_regs): Likewise.
251 (compute_reload_subreg_offset): Likewise, and return an poly_int64.
252 * simplify-rtx.c (simplify_truncation, simplify_binary_operation_1):
253 (test_vector_ops_duplicate): Treat subreg offsets as poly_ints.
254 (simplify_const_poly_int_tests<N>::run): Likewise.
255 (simplify_subreg, simplify_gen_subreg): Take the subreg offset as
256 a poly_uint64 rather than an unsigned int.
257 * valtrack.c (debug_lowpart_subreg): Likewise.
258 * var-tracking.c (var_lowpart): Likewise.
259 (loc_cmp): Handle 'p'.
261 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
262 Alan Hayward <alan.hayward@arm.com>
263 David Sherwood <david.sherwood@arm.com>
265 * ira.c (get_subreg_tracking_sizes): New function.
266 (init_live_subregs): Take an integer size rather than a register.
267 (build_insn_chain): Use get_subreg_tracking_sizes. Update calls
268 to init_live_subregs.
270 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
271 Alan Hayward <alan.hayward@arm.com>
272 David Sherwood <david.sherwood@arm.com>
274 * expr.c (store_constructor_field): Change bitsize from a
275 unsigned HOST_WIDE_INT to a poly_uint64 and bitpos from a
276 HOST_WIDE_INT to a poly_int64.
277 (store_constructor): Change size from a HOST_WIDE_INT to
279 (store_field): Likewise bitsize and bitpos.
281 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
282 Alan Hayward <alan.hayward@arm.com>
283 David Sherwood <david.sherwood@arm.com>
285 * expmed.h (store_bit_field): Change bitregion_start and
286 bitregion_end from unsigned HOST_WIDE_INT to poly_uint64.
287 * expmed.c (adjust_bit_field_mem_for_reg, strict_volatile_bitfield_p)
288 (store_bit_field_1, store_integral_bit_field, store_bit_field)
289 (store_fixed_bit_field, store_split_bit_field): Likewise.
290 * expr.c (store_constructor_field, store_field): Likewise.
291 (optimize_bitfield_assignment_op): Likewise. Make the same change
292 to bitsize and bitpos.
293 * machmode.h (bit_field_mode_iterator): Change m_bitregion_start
294 and m_bitregion_end from HOST_WIDE_INT to poly_int64. Make the
295 same change in the constructor arguments.
296 (get_best_mode): Change bitregion_start and bitregion_end from
297 unsigned HOST_WIDE_INT to poly_uint64.
298 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
299 Change bitregion_start and bitregion_end from HOST_WIDE_INT to
301 (bit_field_mode_iterator::next_mode): Update for new types
302 of m_bitregion_start and m_bitregion_end.
303 (get_best_mode): Change bitregion_start and bitregion_end from
304 unsigned HOST_WIDE_INT to poly_uint64.
306 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
307 Alan Hayward <alan.hayward@arm.com>
308 David Sherwood <david.sherwood@arm.com>
310 * rtl.h (simplify_gen_subreg): Add a temporary overload that
311 accepts poly_uint64 offsets.
312 * expmed.h (extract_bit_field): Take bitsize and bitnum as
313 poly_uint64s rather than unsigned HOST_WIDE_INTs.
314 * expmed.c (lowpart_bit_field_p): Likewise.
315 (extract_bit_field_as_subreg): New function, split out from...
316 (extract_bit_field_1): ...here. Take bitsize and bitnum as
317 poly_uint64s rather than unsigned HOST_WIDE_INTs. For vector
318 extractions, check that BITSIZE matches the size of the extracted
319 value and that BITNUM is an exact multiple of that size.
320 If all else fails, try forcing the value into memory if
321 BITNUM is variable, and adjusting the address so that the
322 offset is constant. Split the part that can only handle constant
323 bitsize and bitnum out into...
324 (extract_integral_bit_field): ...this new function.
325 (extract_bit_field): Take bitsize and bitnum as poly_uint64s
326 rather than unsigned HOST_WIDE_INTs.
328 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
329 Alan Hayward <alan.hayward@arm.com>
330 David Sherwood <david.sherwood@arm.com>
332 * expmed.h (store_bit_field): Take bitsize and bitnum as
333 poly_uint64s rather than unsigned HOST_WIDE_INTs.
334 * expmed.c (simple_mem_bitfield_p): Likewise. Add a parameter
335 that returns the byte size.
336 (store_bit_field_1): Take bitsize and bitnum as
337 poly_uint64s rather than unsigned HOST_WIDE_INTs. Update call
338 to simple_mem_bitfield_p. Split the part that can only handle
339 constant bitsize and bitnum out into...
340 (store_integral_bit_field): ...this new function.
341 (store_bit_field): Take bitsize and bitnum as poly_uint64s rather
342 than unsigned HOST_WIDE_INTs.
343 (extract_bit_field_1): Update call to simple_mem_bitfield_p.
345 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
346 Alan Hayward <alan.hayward@arm.com>
347 David Sherwood <david.sherwood@arm.com>
349 * lra-int.h (lra_reg): Change offset from int to poly_int64.
350 (lra_insn_recog_data): Change sp_offset from HOST_WIDE_INT
352 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
353 update_sp_offset from a HOST_WIDE_INT to a poly_int64.
354 (lra_update_reg_val_offset, lra_reg_val_equal_p): Take the
355 offset as a poly_int64 rather than an int.
356 * lra-assigns.c (find_hard_regno_for_1): Handle poly_int64 offsets.
357 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
358 * lra-constraints.c (equiv_address_substitution): Track offsets
360 (emit_inc): Check poly_int_rtx_p instead of CONST_INT_P.
361 (curr_insn_transform): Handle the new form of sp_offset.
362 * lra-eliminations.c (lra_elim_table): Change previous_offset
363 and offset from HOST_WIDE_INT to poly_int64.
364 (print_elim_table, update_reg_eliminate): Update accordingly.
365 (self_elim_offsets): Change from HOST_WIDE_INT to poly_int64_pod.
366 (get_elimination): Update accordingly.
367 (form_sum): Check poly_int_rtx_p instead of CONST_INT_P.
368 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
369 update_sp_offset from a HOST_WIDE_INT to a poly_int64. Handle
370 poly_int64 offsets generally.
371 (curr_sp_change): Change from HOST_WIDE_INT to poly_int64.
372 (mark_not_eliminable, init_elimination): Update accordingly.
373 (remove_reg_equal_offset_note): Return a bool and pass the new
374 offset back by pointer as a poly_int64.
375 * lra-remat.c (change_sp_offset): Take sp_offset as a poly_int64
376 rather than a HOST_WIDE_INT.
377 (do_remat): Track offsets poly_int64s.
378 * lra.c (lra_update_insn_recog_data, setup_sp_offset): Likewise.
380 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
381 Alan Hayward <alan.hayward@arm.com>
382 David Sherwood <david.sherwood@arm.com>
384 * rtl.h (mem_attrs): Add a default constructor. Change size and
385 offset from HOST_WIDE_INT to poly_int64.
386 * emit-rtl.h (set_mem_offset, set_mem_size, adjust_address_1)
387 (adjust_automodify_address_1, set_mem_attributes_minus_bitpos)
388 (widen_memory_access): Take the sizes and offsets as poly_int64s
389 rather than HOST_WIDE_INTs.
390 * alias.c (ao_ref_from_mem): Handle the new form of MEM_OFFSET.
391 (offset_overlap_p): Take poly_int64s rather than HOST_WIDE_INTs
393 (adjust_offset_for_component_ref): Change the offset from a
394 HOST_WIDE_INT to a poly_int64.
395 (nonoverlapping_memrefs_p): Track polynomial offsets and sizes.
396 * cfgcleanup.c (merge_memattrs): Update after mem_attrs changes.
397 * dce.c (find_call_stack_args): Likewise.
398 * dse.c (record_store): Likewise.
399 * dwarf2out.c (tls_mem_loc_descriptor, dw_sra_loc_expr): Likewise.
400 * print-rtl.c (rtx_writer::print_rtx): Likewise.
401 * read-rtl-function.c (test_loading_mem): Likewise.
402 * rtlanal.c (may_trap_p_1): Likewise.
403 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
404 * var-tracking.c (int_mem_offset, track_expr_p): Likewise.
405 * emit-rtl.c (mem_attrs_eq_p, get_mem_align_offset): Likewise.
406 (mem_attrs::mem_attrs): New function.
407 (set_mem_attributes_minus_bitpos): Change bitpos from a
408 HOST_WIDE_INT to poly_int64.
409 (set_mem_alias_set, set_mem_addr_space, set_mem_align, set_mem_expr)
410 (clear_mem_offset, clear_mem_size, change_address)
411 (get_spill_slot_decl, set_mem_attrs_for_spill): Directly
412 initialize mem_attrs.
413 (set_mem_offset, set_mem_size, adjust_address_1)
414 (adjust_automodify_address_1, offset_address, widen_memory_access):
415 Likewise. Take poly_int64s rather than HOST_WIDE_INT.
417 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
418 Alan Hayward <alan.hayward@arm.com>
419 David Sherwood <david.sherwood@arm.com>
421 * rtlanal.c (rtx_addr_can_trap_p_1): Take the offset and size
422 as poly_int64s rather than HOST_WIDE_INTs. Use a size of -1
423 rather than 0 to represent an unknown size. Assert that the size
424 is known when the mode isn't BLKmode.
425 (may_trap_p_1): Use -1 for unknown sizes.
426 (rtx_addr_can_trap_p): Likewise. Pass BLKmode rather than VOIDmode.
428 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
429 Alan Hayward <alan.hayward@arm.com>
430 David Sherwood <david.sherwood@arm.com>
432 * dse.c (store_info): Change offset and width from HOST_WIDE_INT
433 to poly_int64. Update commentary for positions_needed.large.
434 (read_info_type): Change offset and width from HOST_WIDE_INT
436 (set_usage_bits): Likewise.
437 (canon_address): Return the offset as a poly_int64 rather than
438 a HOST_WIDE_INT. Use strip_offset_and_add.
439 (set_all_positions_unneeded, any_positions_needed_p): Use
440 positions_needed.large to track stores with non-constant widths.
441 (all_positions_needed_p): Likewise. Take the offset and width
442 as poly_int64s rather than ints. Assert that rhs is nonnull.
443 (record_store): Cope with non-constant offsets and widths.
444 Nullify the rhs of an earlier store if we can't tell which bytes
446 (find_shift_sequence): Take the access_size and shift as poly_int64s
448 (get_stored_val): Take the read_offset and read_width as poly_int64s
449 rather than HOST_WIDE_INTs.
450 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Handle
451 non-constant offsets and widths.
453 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
454 Alan Hayward <alan.hayward@arm.com>
455 David Sherwood <david.sherwood@arm.com>
457 * inchash.h (inchash::hash::add_poly_int): New function.
458 * tree-ssa-alias.h (ao_ref::offset, ao_ref::size, ao_ref::max_size):
459 Use poly_int64 rather than HOST_WIDE_INT.
460 (ao_ref::max_size_known_p): New function.
461 * tree-ssa-sccvn.h (vn_reference_op_struct::off): Use poly_int64_pod
462 rather than HOST_WIDE_INT.
463 * tree-ssa-alias.c (ao_ref_base): Apply get_ref_base_and_extent
464 to temporaries until its interface is adjusted to match.
465 (ao_ref_init_from_ptr_and_size): Handle polynomial offsets and sizes.
466 (aliasing_component_refs_p, decl_refs_may_alias_p)
467 (indirect_ref_may_alias_decl_p, indirect_refs_may_alias_p): Take
468 the offsets and max_sizes as poly_int64s instead of HOST_WIDE_INTs.
469 (refs_may_alias_p_1, stmt_kills_ref_p): Adjust for changes to
471 * alias.c (ao_ref_from_mem): Likewise.
472 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
473 * tree-ssa-dse.c (valid_ao_ref_for_dse, normalize_ref)
474 (clear_bytes_written_by, setup_live_bytes_from_ref, compute_trims)
475 (maybe_trim_complex_store, maybe_trim_constructor_store)
476 (live_bytes_read, dse_classify_store): Likewise.
477 * tree-ssa-sccvn.c (vn_reference_compute_hash, vn_reference_eq):
478 (copy_reference_ops_from_ref, ao_ref_init_from_vn_reference)
479 (fully_constant_vn_reference_p, valueize_refs_1): Likewise.
480 (vn_reference_lookup_3): Likewise.
481 * tree-ssa-uninit.c (warn_uninitialized_vars): Likewise.
483 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
484 Alan Hayward <alan.hayward@arm.com>
485 David Sherwood <david.sherwood@arm.com>
487 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p)
488 (indirect_refs_may_alias_p): Use ranges_may_overlap_p
489 instead of ranges_overlap_p.
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 * tree-ssa-alias.c (same_addr_size_stores_p): Take the offsets and
496 sizes as poly_int64s rather than HOST_WIDE_INTs.
498 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
499 Alan Hayward <alan.hayward@arm.com>
500 David Sherwood <david.sherwood@arm.com>
502 * gimple-fold.h (fold_ctor_reference): Take the offset and size
503 as poly_uint64 rather than unsigned HOST_WIDE_INT.
504 * gimple-fold.c (fold_ctor_reference): Likewise.
506 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
507 Alan Hayward <alan.hayward@arm.com>
508 David Sherwood <david.sherwood@arm.com>
510 * target.def (dwarf_poly_indeterminate_value): New hook.
511 * targhooks.h (default_dwarf_poly_indeterminate_value): Declare.
512 * targhooks.c (default_dwarf_poly_indeterminate_value): New function.
513 * doc/tm.texi.in (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Document.
514 * doc/tm.texi: Regenerate.
515 * dwarf2out.h (build_cfa_loc, build_cfa_aligned_loc): Take the
516 offset as a poly_int64.
517 * dwarf2out.c (new_reg_loc_descr): Move later in file. Take the
518 offset as a poly_int64.
519 (loc_descr_plus_const, loc_list_plus_const, build_cfa_aligned_loc):
520 Take the offset as a poly_int64.
521 (build_cfa_loc): Likewise. Use loc_descr_plus_const.
522 (frame_pointer_fb_offset): Change to a poly_int64.
523 (int_loc_descriptor): Take the offset as a poly_int64. Use
524 targetm.dwarf_poly_indeterminate_value for polynomial offsets.
525 (based_loc_descr): Take the offset as a poly_int64.
526 Use strip_offset_and_add to handle (plus X (const)).
527 Use new_reg_loc_descr instead of an open-coded version of the
528 previous implementation.
529 (mem_loc_descriptor): Handle CONST_POLY_INT.
530 (compute_frame_pointer_to_fb_displacement): Take the offset as a
531 poly_int64. Use strip_offset_and_add to handle (plus X (const)).
533 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
534 Alan Hayward <alan.hayward@arm.com>
535 David Sherwood <david.sherwood@arm.com>
537 * rtl.h (reg_attrs::offset): Change from HOST_WIDE_INT to poly_int64.
538 (gen_rtx_REG_offset): Take the offset as a poly_int64.
539 * inchash.h (inchash::hash::add_poly_hwi): New function.
540 * gengtype.c (main): Register poly_int64.
541 * emit-rtl.c (reg_attr_hasher::hash): Use inchash. Treat the
542 offset as a poly_int.
543 (reg_attr_hasher::equal): Use must_eq to compare offsets.
544 (get_reg_attrs, update_reg_offset, gen_rtx_REG_offset): Take the
545 offset as a poly_int64.
546 (set_reg_attrs_from_value): Treat the offset as a poly_int64.
547 * print-rtl.c (print_poly_int): New function.
548 (rtx_writer::print_rtx_operand_code_r): Treat REG_OFFSET as
550 * var-tracking.c (track_offset_p, get_tracked_reg_offset): New
552 (var_reg_set, var_reg_delete_and_set, var_reg_delete): Use them.
553 (same_variable_part_p, track_loc_p): Take the offset as a poly_int64.
554 (vt_get_decl_and_offset): Return the offset as a poly_int64.
555 Enforce track_offset_p for parts of a PARALLEL.
556 (vt_add_function_parameter): Use const_offset for the final
557 offset to track. Use get_tracked_reg_offset for the parts
560 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
561 Alan Hayward <alan.hayward@arm.com>
562 David Sherwood <david.sherwood@arm.com>
564 * target.def (truly_noop_truncation): Take poly_uint64s instead of
565 unsigned ints. Change default to hook_bool_puint64_puint64_true.
566 * doc/tm.texi: Regenerate.
567 * hooks.h (hook_bool_uint_uint_true): Delete.
568 (hook_bool_puint64_puint64_true): Declare.
569 * hooks.c (hook_bool_uint_uint_true): Delete.
570 (hook_bool_puint64_puint64_true): New function.
571 * config/mips/mips.c (mips_truly_noop_truncation): Take poly_uint64s
572 instead of unsigned ints.
573 * config/spu/spu.c (spu_truly_noop_truncation): Likewise.
574 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): Likewise.
576 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
577 Alan Hayward <alan.hayward@arm.com>
578 David Sherwood <david.sherwood@arm.com>
580 * optabs.h (expand_operand): Add an int_value field.
581 (create_expand_operand): Add an int_value parameter and use it
582 to initialize the new expand_operand field.
583 (create_integer_operand): Replace with a declaration of a function
584 that accepts poly_int64s. Move the implementation to...
585 * optabs.c (create_integer_operand): ...here.
586 (maybe_legitimize_operand): For EXPAND_INTEGER, check whether
587 the mode preserves the value of int_value, instead of calling
588 const_int_operand on the rtx. Use gen_int_mode to generate
591 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
592 Alan Hayward <alan.hayward@arm.com>
593 David Sherwood <david.sherwood@arm.com>
595 * dumpfile.h (dump_dec): Declare.
596 * dumpfile.c (dump_dec): New function.
597 * pretty-print.h (pp_wide_integer): Turn into a function and
598 declare a poly_int version.
599 * pretty-print.c (pp_wide_integer): New function for poly_ints.
601 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
602 Alan Hayward <alan.hayward@arm.com>
603 David Sherwood <david.sherwood@arm.com>
605 * doc/generic.texi (POLY_INT_CST): Document.
606 * tree.def (POLY_INT_CST): New tree code.
607 * treestruct.def (TS_POLY_INT_CST): New tree layout.
608 * tree-core.h (tree_poly_int_cst): New struct.
609 (tree_node): Add a poly_int_cst field.
610 * tree.h (POLY_INT_CST_P, POLY_INT_CST_COEFF): New macros.
611 (wide_int_to_tree, force_fit_type): Take a poly_wide_int_ref
612 instead of a wide_int_ref.
613 (build_int_cst, build_int_cst_type): Take a poly_int64 instead
615 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
616 instead of an unsigned HOST_WIDE_INT.
617 (build_poly_int_cst, tree_fits_poly_int64_p, tree_fits_poly_uint64_p)
618 (ptrdiff_tree_p): Declare.
619 (tree_to_poly_int64, tree_to_poly_uint64): Likewise. Provide
620 extern inline implementations if the target doesn't use POLY_INT_CST.
621 (poly_int_tree_p): New function.
622 (wi::unextended_tree): New class.
623 (wi::int_traits <unextended_tree>): New override.
624 (wi::extended_tree): Add a default constructor.
625 (wi::extended_tree::get_tree): New function.
626 (wi::widest_extended_tree, wi::offset_extended_tree): New typedefs.
627 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): Use them.
628 (wi::tree_to_poly_widest_ref, wi::tree_to_poly_offset_ref)
629 (wi::tree_to_poly_wide_ref): New typedefs.
630 (wi::ints_for): Provide overloads for extended_tree and
632 (poly_int_cst_value, wi::to_poly_widest, wi::to_poly_offset)
633 (wi::to_wide): New functions.
634 (wi::fits_to_boolean_p, wi::fits_to_tree_p): Handle poly_ints.
635 * tree.c (poly_int_cst_hasher): New struct.
636 (poly_int_cst_hash_table): New variable.
637 (tree_node_structure_for_code, tree_code_size, simple_cst_equal)
638 (valid_constant_size_p, add_expr, drop_tree_overflow): Handle
640 (initialize_tree_contains_struct): Handle TS_POLY_INT_CST.
641 (init_ttree): Initialize poly_int_cst_hash_table.
642 (build_int_cst, build_int_cst_type, build_invariant_address): Take
643 a poly_int64 instead of a HOST_WIDE_INT.
644 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
645 instead of an unsigned HOST_WIDE_INT.
646 (wide_int_to_tree): Rename to...
647 (wide_int_to_tree_1): ...this.
648 (build_new_poly_int_cst, build_poly_int_cst): New functions.
649 (force_fit_type): Take a poly_wide_int_ref instead of a wide_int_ref.
650 (wide_int_to_tree): New function that takes a poly_wide_int_ref.
651 (ptrdiff_tree_p, tree_to_poly_int64, tree_to_poly_uint64)
652 (tree_fits_poly_int64_p, tree_fits_poly_uint64_p): New functions.
653 * lto-streamer-out.c (DFS::DFS_write_tree_body, hash_tree): Handle
655 * tree-streamer-in.c (lto_input_ts_poly_tree_pointers): Likewise.
656 (streamer_read_tree_body): Likewise.
657 * tree-streamer-out.c (write_ts_poly_tree_pointers): Likewise.
658 (streamer_write_tree_body): Likewise.
659 * tree-streamer.c (streamer_check_handled_ts_structures): Likewise.
660 * asan.c (asan_protect_global): Require the size to be an INTEGER_CST.
661 * cfgexpand.c (expand_debug_expr): Handle POLY_INT_CST.
662 * expr.c (expand_expr_real_1, const_vector_from_tree): Likewise.
663 * gimple-expr.h (is_gimple_constant): Likewise.
664 * gimplify.c (maybe_with_size_expr): Likewise.
665 * print-tree.c (print_node): Likewise.
666 * tree-data-ref.c (data_ref_compare_tree): Likewise.
667 * tree-pretty-print.c (dump_generic_node): Likewise.
668 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
669 * tree-vect-data-refs.c (dr_group_sort_cmp): Likewise.
670 * tree-vrp.c (compare_values_warnv): Likewise.
671 * tree-ssa-loop-ivopts.c (determine_base_object, constant_multiple_of)
672 (get_loop_invariant_expr, add_candidate_1, get_computation_aff_1)
673 (force_expr_to_var_cost): Likewise.
674 * tree-ssa-loop.c (for_each_index): Likewise.
675 * fold-const.h (build_invariant_address, size_int_kind): Take a
676 poly_int64 instead of a HOST_WIDE_INT.
677 * fold-const.c (fold_negate_expr_1, const_binop, const_unop)
678 (fold_convert_const, multiple_of_p, fold_negate_const): Handle
680 (size_binop_loc): Likewise. Allow int_const_binop_1 to fail.
681 (int_const_binop_2): New function, split out from...
682 (int_const_binop_1): ...here. Handle POLY_INT_CST.
683 (size_int_kind): Take a poly_int64 instead of a HOST_WIDE_INT.
684 * expmed.c (make_tree): Handle CONST_POLY_INT_P.
685 * gimple-ssa-strength-reduction.c (slsr_process_add)
686 (slsr_process_mul): Check for INTEGER_CSTs before using them
688 * stor-layout.c (bits_from_bytes): New function.
689 (bit_from_pos): Use it.
690 (layout_type): Likewise. For vectors, multiply the TYPE_SIZE_UNIT
691 by BITS_PER_UNIT to get the TYPE_SIZE.
692 * tree-cfg.c (verify_expr, verify_types_in_gimple_reference): Allow
693 MEM_REF and TARGET_MEM_REF offsets to be a POLY_INT_CST.
695 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
696 Alan Hayward <alan.hayward@arm.com>
697 David Sherwood <david.sherwood@arm.com>
699 * doc/rtl.texi (const_poly_int): Document. Also document the
700 rtl sharing behavior.
701 * gengenrtl.c (excluded_rtx): Return true for CONST_POLY_INT.
702 * rtl.h (const_poly_int_def): New struct.
703 (rtx_def::u): Add a cpi field.
704 (CASE_CONST_UNIQUE, CASE_CONST_ANY): Add CONST_POLY_INT.
705 (CONST_POLY_INT_P, CONST_POLY_INT_COEFFS): New macros.
706 (wi::rtx_to_poly_wide_ref): New typedef
707 (const_poly_int_value, wi::to_poly_wide, rtx_to_poly_int64)
708 (poly_int_rtx_p): New functions.
709 (trunc_int_for_mode): Declare a poly_int64 version.
710 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
711 (immed_wide_int_const): Take a poly_wide_int_ref rather than
713 (strip_offset): Declare.
714 (strip_offset_and_add): New function.
715 * rtl.def (CONST_POLY_INT): New rtx code.
716 * rtl.c (rtx_size): Handle CONST_POLY_INT.
717 (shared_const_p): Use poly_int_rtx_p.
718 * emit-rtl.h (gen_int_mode): Take a poly_int64 instead of a
720 (gen_int_shift_amount): Likewise.
721 * emit-rtl.c (const_poly_int_hasher): New class.
722 (const_poly_int_htab): New variable.
723 (init_emit_once): Initialize it when NUM_POLY_INT_COEFFS > 1.
724 (const_poly_int_hasher::hash): New function.
725 (const_poly_int_hasher::equal): Likewise.
726 (gen_int_mode): Take a poly_int64 instead of a HOST_WIDE_INT.
727 (immed_wide_int_const): Rename to...
728 (immed_wide_int_const_1): ...this and make static.
729 (immed_wide_int_const): New function, taking a poly_wide_int_ref
730 instead of a wide_int_ref.
731 (gen_int_shift_amount): Take a poly_int64 instead of a HOST_WIDE_INT.
732 (gen_lowpart_common): Handle CONST_POLY_INT.
733 * cse.c (hash_rtx_cb, equiv_constant): Likewise.
734 * cselib.c (cselib_hash_rtx): Likewise.
735 * dwarf2out.c (const_ok_for_output_1): Likewise.
736 * expr.c (convert_modes): Likewise.
737 * print-rtl.c (rtx_writer::print_rtx, print_value): Likewise.
738 * rtlhash.c (add_rtx): Likewise.
739 * explow.c (trunc_int_for_mode): Add a poly_int64 version.
740 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
741 Handle existing CONST_POLY_INT rtxes.
742 * expmed.h (expand_shift): Take a poly_int64 instead of a
744 * expmed.c (expand_shift): Likewise.
745 * rtlanal.c (strip_offset): New function.
746 (commutative_operand_precedence): Give CONST_POLY_INT the same
747 precedence as CONST_DOUBLE and put CONST_WIDE_INT between that
749 * rtl-tests.c (const_poly_int_tests): New struct.
750 (rtl_tests_c_tests): Use it.
751 * simplify-rtx.c (simplify_const_unary_operation): Handle
753 (simplify_const_binary_operation): Likewise.
754 (simplify_binary_operation_1): Fold additions of symbolic constants
756 (simplify_subreg): Handle extensions and truncations of
758 (simplify_const_poly_int_tests): New struct.
759 (simplify_rtx_c_tests): Use it.
760 * wide-int.h (storage_ref): Add default constructor.
761 (wide_int_ref_storage): Likewise.
762 (trailing_wide_ints): Use GTY((user)).
763 (trailing_wide_ints::operator[]): Add a const version.
764 (trailing_wide_ints::get_precision): New function.
765 (trailing_wide_ints::extra_size): Likewise.
767 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
768 Alan Hayward <alan.hayward@arm.com>
769 David Sherwood <david.sherwood@arm.com>
771 * emit-rtl.h (gen_int_shift_amount): Declare.
772 * emit-rtl.c (gen_int_shift_amount): New function.
773 * asan.c (asan_emit_stack_protection): Use gen_int_shift_amount
775 * calls.c (shift_return_value): Likewise.
776 * cse.c (fold_rtx): Likewise.
777 * dse.c (find_shift_sequence): Likewise.
778 * expmed.c (init_expmed_one_mode, store_bit_field_1, expand_shift_1)
779 (expand_shift, expand_smod_pow2): Likewise.
780 * lower-subreg.c (shift_cost): Likewise.
781 * optabs.c (expand_superword_shift, expand_doubleword_mult)
782 (expand_unop, expand_binop, shift_amt_for_vec_perm_mask)
783 (expand_vec_perm_var): Likewise.
784 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
785 (simplify_binary_operation_1): Likewise.
786 * combine.c (try_combine, find_split_point, force_int_to_mode)
787 (simplify_shift_const_1, simplify_shift_const): Likewise.
788 (change_zero_ext): Likewise. Use simplify_gen_binary.
790 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
792 * poly-int.h (multiple_p): Fix handling of two non-poly_ints.
794 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
796 * doc/invoke.texi (ARM Options): Document accepted extension options
797 for -march=armv8.3-a.
799 2017-12-20 Richard Earnshaw <rearnsha@arm.com>
802 * config.gcc (arm*-*-linux*): When configured with --with-float=hard
803 or --with-float=softfp, set the default CPU to arm10e.
805 2017-12-20 Eric Botcazou <ebotcazou@adacore.com>
807 * config/visium/constraints.md (J, K, L): Use IN_RANGE macro.
808 * config/visium/predicates.md (const_shift_operand): Likewise.
809 * config/visium/visium.c (visium_legitimize_address): Fix oversight.
810 (visium_legitimize_reload_address): Likewise.
812 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
815 * config/arm/arm.h (TEST_REGNO): Adjust comment as expected in r255830.
817 2017-12-20 Jakub Jelinek <jakub@redhat.com>
820 * calls.c (compute_argument_addresses): Ignore TYPE_EMPTY_P arguments.
822 2017-12-20 Julia Koval <julia.koval@intel.com>
824 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VPCLMULQDQ_SET,
825 OPTION_MASK_ISA_VPCLMULQDQ_UNSET): New.
826 (ix86_handle_option): Handle -mvpclmulqdq, move cx6 to flags2.
827 * config.gcc: Include vpclmulqdqintrin.h.
828 * config/i386/cpuid.h: Handle bit_VPCLMULQDQ.
829 * config/i386/driver-i386.c (host_detect_local_cpu): Handle
831 * config/i386/i386-builtin.def (__builtin_ia32_vpclmulqdq_v2di,
832 __builtin_ia32_vpclmulqdq_v4di, __builtin_ia32_vpclmulqdq_v8di): New.
833 * config/i386/i386-c.c (__VPCLMULQDQ__): New.
834 * config/i386/i386.c (isa2_opts): Add -mcx16.
835 (isa_opts): Add -mpclmulqdq, remove -mcx16.
836 (ix86_option_override_internal): Move mcx16 to flags2.
837 (ix86_valid_target_attribute_inner_p): Add vpclmulqdq.
838 (ix86_expand_builtin): Handle OPTION_MASK_ISA_VPCLMULQDQ.
839 * config/i386/i386.h (TARGET_VPCLMULQDQ, TARGET_VPCLMULQDQ_P): New.
840 * config/i386/i386.opt: Add mvpclmulqdq, move mcx16 to flags2.
841 * config/i386/immintrin.h: Include vpclmulqdqintrin.h.
842 * config/i386/sse.md (vpclmulqdq_<mode>): New pattern.
843 * config/i386/vpclmulqdqintrin.h (_mm512_clmulepi64_epi128,
844 _mm_clmulepi64_epi128, _mm256_clmulepi64_epi128): New intrinsics.
845 * doc/invoke.texi: Add -mvpclmulqdq.
847 2017-12-20 Tom de Vries <tom@codesourcery.com>
850 * config/i386/i386.c (ix86_static_chain): Move DECL_STATIC_CHAIN test ...
851 * calls.c (rtx_for_static_chain): ... here. New function.
852 * calls.h (rtx_for_static_chain): Declare.
853 * builtins.c (expand_builtin_setjmp_receiver): Use rtx_for_static_chain
854 instead of targetm.calls.static_chain.
855 * df-scan.c (df_get_entry_block_def_set): Same.
857 2017-12-19 Tom de Vries <tom@codesourcery.com>
859 PR tree-optimization/83493
860 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for): Unshare ub
863 2017-12-19 Jakub Jelinek <jakub@redhat.com>
865 * gimple-ssa-sprintf.c (format_directive): Use inform_n instead of
866 inform with hardcoded english plural handling.
868 2017-12-18 Jeff Law <law@redhat.com>
870 PR tree-optimization/83477
871 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis): For
872 a non-virtual PHI, always push a new range.
874 2017-12-19 Martin Sebor <msebor@redhat.com>
877 * builtins.c (compute_objsize): Handle non-constant offsets.
879 2017-12-19 Jakub Jelinek <jakub@redhat.com>
881 PR tree-optimization/83444
882 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): For the
883 character load case, if get_stridx on MEM_REF's operand doesn't
884 look usable, retry with get_addr_stridx.
886 2017-12-19 Alexandre Oliva <aoliva@redhat.com>
889 * var-tracking.c (vt_debug_insns_local): Do not drop markers.
890 (variable_tracking_main_1): Keep markers even when VTA fails.
893 * cfgexpand.c (expand_gimple_basic_block): Expand label first,
894 even if there are markers before it.
895 * cfgrtl.c (rtl_verify_bb_layout): Reject DEBUG_INSNs outside BBs.
897 2017-12-19 Jakub Jelinek <jakub@redhat.com>
899 * read-rtl.c (parse_reg_note_name): Replace Yoda conditions with
900 typical order conditions.
901 * sel-sched.c (extract_new_fences_from): Likewise.
902 * config/visium/constraints.md (J, K, L): Likewise.
903 * config/visium/predicates.md (const_shift_operand): Likewise.
904 * config/visium/visium.c (visium_legitimize_address,
905 visium_legitimize_reload_address): Likewise.
906 * config/m68k/m68k.c (output_reg_adjust, emit_reg_adjust): Likewise.
907 * config/arm/arm.c (arm_block_move_unaligned_straight): Likewise.
908 * config/avr/constraints.md (Y01, Ym1, Y02, Ym2): Likewise.
909 * config/avr/avr-log.c (avr_vdump, avr_log_set_avr_log,
910 SET_DUMP_DETAIL): Likewise.
911 * config/avr/predicates.md (const_8_16_24_operand): Likewise.
912 * config/avr/avr.c (STR_PREFIX_P, avr_popcount_each_byte,
913 avr_is_casesi_sequence, avr_casei_sequence_check_operands,
914 avr_set_core_architecture, avr_set_current_function,
915 avr_legitimize_reload_address, avr_asm_len, avr_print_operand,
916 output_movqi, output_movsisf, avr_out_plus, avr_out_bitop,
917 avr_out_fract, avr_adjust_insn_length, avr_encode_section_info,
918 avr_2word_insn_p, output_reload_in_const, avr_has_nibble_0xf,
919 avr_map_decompose, avr_fold_builtin): Likewise.
920 * config/avr/driver-avr.c (avr_devicespecs_file): Likewise.
921 * config/avr/gen-avr-mmcu-specs.c (str_prefix_p, print_mcu): Likewise.
922 * config/i386/i386.c (ix86_parse_stringop_strategy_string): Likewise.
923 * config/m32c/m32c-pragma.c (m32c_pragma_memregs): Likewise.
924 * config/m32c/m32c.c (m32c_conditional_register_usage,
925 m32c_address_cost): Likewise.
926 * config/m32c/predicates.md (shiftcount_operand,
927 longshiftcount_operand): Likewise.
928 * config/iq2000/iq2000.c (iq2000_expand_prologue): Likewise.
929 * config/nios2/nios2.c (nios2_handle_custom_fpu_insn_option,
930 can_use_cdx_ldstw): Likewise.
931 * config/nios2/nios2.h (CDX_REG_P): Likewise.
932 * config/cr16/cr16.h (RETURN_ADDR_RTX, REGNO_MODE_OK_FOR_BASE_P):
934 * config/cr16/cr16.md (*mov<mode>_double): Likewise.
935 * config/cr16/cr16.c (cr16_create_dwarf_for_multi_push): Likewise.
936 * config/h8300/h8300.c (h8300_rtx_costs, get_shift_alg): Likewise.
937 * config/vax/constraints.md (U06, U08, U16, CN6, S08, S16): Likewise.
938 * config/vax/vax.c (adjacent_operands_p): Likewise.
939 * config/ft32/constraints.md (L, b, KA): Likewise.
940 * config/ft32/ft32.c (ft32_load_immediate, ft32_expand_prologue):
942 * cfgexpand.c (expand_stack_alignment): Likewise.
943 * gcse.c (insert_expr_in_table): Likewise.
944 * print-rtl.c (rtx_writer::print_rtx_operand_codes_E_and_V): Likewise.
945 * cgraphunit.c (cgraph_node::expand): Likewise.
946 * ira-build.c (setup_min_max_allocno_live_range_point): Likewise.
947 * emit-rtl.c (add_insn): Likewise.
948 * input.c (dump_location_info): Likewise.
949 * passes.c (NEXT_PASS): Likewise.
950 * read-rtl-function.c (parse_note_insn_name,
951 function_reader::read_rtx_operand_r, function_reader::parse_mem_expr):
953 * sched-rgn.c (sched_rgn_init): Likewise.
954 * diagnostic-show-locus.c (layout::show_ruler): Likewise.
955 * combine.c (find_split_point, simplify_if_then_else, force_to_mode,
956 if_then_else_cond, simplify_shift_const_1, simplify_comparison): Likewise.
957 * explow.c (eliminate_constant_term): Likewise.
958 * final.c (leaf_renumber_regs_insn): Likewise.
959 * cfgrtl.c (print_rtl_with_bb): Likewise.
960 * genhooks.c (emit_init_macros): Likewise.
961 * poly-int.h (maybe_ne, maybe_le, maybe_lt): Likewise.
962 * tree-data-ref.c (conflict_fn): Likewise.
963 * selftest.c (assert_streq): Likewise.
964 * expr.c (store_constructor_field, expand_expr_real_1): Likewise.
965 * fold-const.c (fold_range_test, extract_muldiv_1, fold_truth_andor,
966 fold_binary_loc, multiple_of_p): Likewise.
967 * reload.c (push_reload, find_equiv_reg): Likewise.
968 * et-forest.c (et_nca, et_below): Likewise.
969 * dbxout.c (dbxout_symbol_location): Likewise.
970 * reorg.c (relax_delay_slots): Likewise.
971 * dojump.c (do_compare_rtx_and_jump): Likewise.
972 * gengtype-parse.c (type): Likewise.
973 * simplify-rtx.c (simplify_gen_ternary, simplify_gen_relational,
974 simplify_const_relational_operation): Likewise.
975 * reload1.c (do_output_reload): Likewise.
976 * dumpfile.c (get_dump_file_info_by_switch): Likewise.
977 * gengtype.c (type_for_name): Likewise.
978 * gimple-ssa-sprintf.c (format_directive): Likewise.
980 2017-12-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
983 * config/arm/arm.h (TEST_REGNO): Check reg_renumber is set before
984 accessing it. Adjust comment.
986 2017-12-19 Jakub Jelinek <jakub@redhat.com>
989 * predict.c (zero_one_minusone): New function.
990 (apply_return_prediction): Avoid return prediction for functions
991 returning only -1, 0 and 1 values, unless they only return -1 and 0
994 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
996 * config/arc/arc.c (legitimate_scaled_address_p): Clean
997 fall-through warning.
998 (arc_compute_frame_size): Remove unused variables.
999 (arc_print_operand): Fix fprintif format.
1000 (arc_can_follow_jump): Clean fall-through warning.
1002 2017-12-19 Marek Polacek <polacek@redhat.com>
1005 * config/i386/i386.c (init_cumulative_args): Don't check TYPE_EMPTY_P
1008 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
1010 * config/arc/arc.c (overriderregs): New variable.
1011 (arc_override_options): Track fixed/call saved/call options.
1012 (arc_conditional_register_usage): Check against overrideregs
1013 variable whenever we change register properties.
1015 2017-12-19 Nathan Sidwell <nathan@acm.org>
1017 * opts.c (finish_options): Don't prefix dump_base_name if it
1018 already contains directories.
1020 2017-12-19 Martin Liska <mliska@suse.cz>
1022 PR rtl-optimization/82675
1023 * loop-unroll.c (unroll_loop_constant_iterations): Allocate one
1024 more element in sbitmap.
1026 2017-12-19 Paolo Carlini <paolo.carlini@oracle.com>
1028 * gimplify.c (gimplify_expr): Use error_operand_p.
1030 2017-12-19 Sebastian Huber <sebastian.huber@embedded-brains.de>
1033 * config/rs6000/t-rtems (MULTILIB_REQUIRED): Remove 64-bit soft-float
1036 2017-12-19 Daniel Cederman <cederman@gaisler.com>
1038 * config/sparc/sparc.c (sparc_do_work_around_errata): Make sure
1039 the jump is to a label.
1041 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1043 PR tree-optimization/83444
1044 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): Optimize
1049 * ipa-inline.c (flatten_remove_node_hook): New function.
1050 (ipa_inline): Keep only nodes with flatten attribute at the end of
1051 the array in the order from ipa_reverse_postorder, only walk that
1052 portion of array for flattening, if there is more than one such
1053 node, temporarily register a removal hook and ignore removed nodes.
1055 PR tree-optimization/80631
1056 * tree-vect-loop.c (vect_create_epilog_for_reduction): Compare
1057 induc_code against MAX_EXPR or MIN_EXPR instead of reduc_fn against
1058 IFN_REDUC_MAX or IFN_REDUC_MIN.
1060 2017-12-18 Jeff Law <law@redhat.com>
1062 * tree-ssa-dom.c (record_equivalences_from_phis): Do not
1063 record symbolic equivalences from backedges in the CFG.
1066 2017-11-19 Jeff Law <law@redhat.com>
1068 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
1069 of degenerates resulting from ignoring an edge.
1071 2017-12-18 Martin Sebor <msebor@redhat.com>
1074 PR tree-optimization/78450
1075 * tree-ssa-strlen.c (maybe_set_strlen_range): New function.
1076 (handle_builtin_strlen): Call it.
1078 2017-12-18 Segher Boessenkool <segher@kernel.crashing.org>
1080 PR rtl-optimization/83424
1081 * rtlanal.c (dead_or_set_regno_p): Handle CLOBBER just like SET.
1083 2017-12-18 Marek Polacek <polacek@redhat.com>
1086 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
1087 Check if TYPE is INTEGRAL_TYPE_P before accessing its min/max
1090 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
1092 * config/arc/arc.md (maddsidi4, maddsidi4_split): Update pattern.
1093 (umaddsidi4, umaddsidi_split): Likewise.
1095 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
1097 * config/arc/arc.c (arc_legitimate_constant_p): Always check all
1100 2017-12-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1103 * config/s390/s390.c (s390_option_override): Avoid strncpy.
1105 2017-12-18 Richard Biener <rguenther@suse.de>
1107 PR tree-optimization/81877
1108 * tree-ssa-loop-im.c (ref_indep_loop_p): Remove safelen parameters.
1109 (outermost_indep_loop): Adjust.
1110 (ref_indep_loop_p_1): Likewise. Remove safelen handling again.
1111 (can_sm_ref_p): Adjust.
1113 2017-12-18 Richard Biener <rguenther@suse.de>
1116 * tree.c (array_at_struct_end_p): Return true if the underlying
1117 object has space for at least one element in excess of what
1118 the array domain specifies.
1120 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
1122 * doc/extend.texi (x86 Function Attributes): Reformat nocf_check
1123 example to avoid overfull hbox.
1124 * doc/invoke.texi (Option Summary): Add missing @gol.
1125 (C++ Dialect Options): Reformat -Wnoexcept-type example to avoid
1128 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
1129 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1131 * doc/invoke.texi (Option Summary): Add -mverbose-cost-dump
1132 to AArch64 and ARM lists, plus missing -mflip-thumb for ARM.
1133 (AArch64 Options): Document -mverbose-cost-dump.
1134 (ARM Options): Likewise, plus -mflip-thumb.
1136 2017-12-17 Martin Sebor <msebor@redhat.com>
1139 * gimple-ssa-warn-restrict.c
1140 (builtin_memref::offset_out_of_bounds): Correct the handling of
1143 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
1145 * doc/invoke.texi (ARC Options): Add missing -mlra entry.
1147 2017-12-17 John David Anglin <danglin@gcc.gnu.org>
1149 * config/pa/pa.c (pa_som_asm_init_sections): Fix comment.
1151 2017-12-17 Markus Trippelsdorf <markus@trippelsdorf.de>
1153 * x86-tune-costs.h (skylake_cost, core_cost): Decrease r64 multiply
1156 2017-12-16 Sandra Loosemore <sandra@codesourcery.com>
1158 * doc/invoke.texi: Fix some typos.
1160 2017-12-16 Martin Sebor <msebor@redhat.com>
1162 PR tree-optimization/78918
1163 * Makefile.in (OBJS): Add gimple-ssa-warn-restrict.o.
1164 * builtins.c (check_sizes): Rename...
1165 (check_access): ...to this. Rename function arguments for clarity.
1166 (check_memop_sizes): Adjust names.
1167 (expand_builtin_memchr, expand_builtin_memcpy): Same.
1168 (expand_builtin_memmove, expand_builtin_mempcpy): Same.
1169 (expand_builtin_strcat, expand_builtin_stpncpy): Same.
1170 (check_strncat_sizes, expand_builtin_strncat): Same.
1171 (expand_builtin_strncpy, expand_builtin_memset): Same.
1172 (expand_builtin_bzero, expand_builtin_memcmp): Same.
1173 (expand_builtin_memory_chk, maybe_emit_chk_warning): Same.
1174 (maybe_emit_sprintf_chk_warning): Same.
1175 (expand_builtin_strcpy): Adjust.
1176 (expand_builtin_stpcpy): Same.
1177 (expand_builtin_with_bounds): Detect out-of-bounds accesses
1178 in pointer-checking forms of memcpy, memmove, and mempcpy.
1179 (gcall_to_tree_minimal, max_object_size): Define new functions.
1180 * builtins.h (max_object_size): Declare.
1181 * calls.c (alloc_max_size): Call max_object_size instead of
1182 hardcoding ssizetype limit.
1183 (get_size_range): Handle new argument.
1184 * calls.h (get_size_range): Add a new argument.
1185 * cfgexpand.c (expand_call_stmt): Propagate no-warning bit.
1186 * doc/invoke.texi (-Wrestrict): Adjust, add example.
1187 * gimple-fold.c (gimple_fold_builtin_memory_op): Detect overlapping
1189 (gimple_fold_builtin_memory_chk): Same.
1190 (gimple_fold_builtin_stxcpy_chk): New function.
1191 * gimple-ssa-warn-restrict.c: New source.
1192 * gimple-ssa-warn-restrict.h: New header.
1193 * gimple.c (gimple_build_call_from_tree): Propagate location.
1194 * passes.def (pass_warn_restrict): Add new pass.
1195 * tree-pass.h (make_pass_warn_restrict): Declare.
1196 * tree-ssa-strlen.c (handle_builtin_strcpy): Detect overlapping
1198 (handle_builtin_strcat): Same.
1199 (strlen_optimize_stmt): Rename...
1200 (strlen_check_and_optimize_stmt): ...to this. Handle strncat,
1201 stpncpy, strncpy, and their checking forms.
1203 2017-12-16 Jan Hubicka <hubicka@ucw.cz>
1205 PR rtl-optimization/82849
1206 * modulo-sched.c (sms_schedule): Use get_estimated_loop_iterations_int
1207 and get_max_loop_iterations_int.
1209 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1210 Alan Hayward <alan.hayward@arm.com>
1211 David Sherwood <david.sherwood@arm.com>
1213 * machmode.h (mode_for_size, int_mode_for_size, float_mode_for_size)
1214 (smallest_mode_for_size, smallest_int_mode_for_size): Take the mode
1215 size as a poly_uint64.
1216 (mode_for_vector, mode_for_int_vector): Take the number of vector
1217 elements as a poly_uint64.
1218 * stor-layout.c (mode_for_size, smallest_mode_for_size): Take the mode
1219 size as a poly_uint64.
1220 (mode_for_vector, mode_for_int_vector): Take the number of vector
1221 elements as a poly_uint64.
1223 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1224 Alan Hayward <alan.hayward@arm.com>
1225 David Sherwood <david.sherwood@arm.com>
1227 * machmode.h (MACRO_MODE): New macro.
1228 * addresses.h (base_reg_class, ok_for_base_p_1): Use it.
1229 * targhooks.c (default_libcall_value, default_secondary_reload)
1230 (default_memory_move_cost, default_register_move_cost)
1231 (default_class_max_nregs): Likewise.
1233 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1234 Alan Hayward <alan.hayward@arm.com>
1235 David Sherwood <david.sherwood@arm.com>
1237 * doc/sourcebuild.texi: Document IN_TARGET_CODE.
1238 * genattrtab.c (write_header): Define IN_TARGET_CODE to 1 in the
1240 * genautomata.c (main): Likewise.
1241 * genconditions.c (write_header): Likewise.
1242 * genemit.c (main): Likewise.
1243 * genextract.c (print_header): Likewise.
1244 * genopinit.c (main): Likewise.
1245 * genoutput.c (output_prologue): Likewise.
1246 * genpeep.c (main): Likewise.
1247 * genpreds.c (write_insn_preds_c): Likewise.
1248 * genrecog.c (writer_header): Likewise.
1249 * config/aarch64/aarch64-builtins.c (IN_TARGET_CODE): Define.
1250 * config/aarch64/aarch64-c.c (IN_TARGET_CODE): Likewise.
1251 * config/aarch64/aarch64.c (IN_TARGET_CODE): Likewise.
1252 * config/aarch64/cortex-a57-fma-steering.c (IN_TARGET_CODE): Likewise.
1253 * config/aarch64/driver-aarch64.c (IN_TARGET_CODE): Likewise.
1254 * config/alpha/alpha.c (IN_TARGET_CODE): Likewise.
1255 * config/alpha/driver-alpha.c (IN_TARGET_CODE): Likewise.
1256 * config/arc/arc-c.c (IN_TARGET_CODE): Likewise.
1257 * config/arc/arc.c (IN_TARGET_CODE): Likewise.
1258 * config/arc/driver-arc.c (IN_TARGET_CODE): Likewise.
1259 * config/arm/aarch-common.c (IN_TARGET_CODE): Likewise.
1260 * config/arm/arm-builtins.c (IN_TARGET_CODE): Likewise.
1261 * config/arm/arm-c.c (IN_TARGET_CODE): Likewise.
1262 * config/arm/arm.c (IN_TARGET_CODE): Likewise.
1263 * config/arm/driver-arm.c (IN_TARGET_CODE): Likewise.
1264 * config/avr/avr-c.c (IN_TARGET_CODE): Likewise.
1265 * config/avr/avr-devices.c (IN_TARGET_CODE): Likewise.
1266 * config/avr/avr-log.c (IN_TARGET_CODE): Likewise.
1267 * config/avr/avr.c (IN_TARGET_CODE): Likewise.
1268 * config/avr/driver-avr.c (IN_TARGET_CODE): Likewise.
1269 * config/avr/gen-avr-mmcu-specs.c (IN_TARGET_CODE): Likewise.
1270 * config/bfin/bfin.c (IN_TARGET_CODE): Likewise.
1271 * config/c6x/c6x.c (IN_TARGET_CODE): Likewise.
1272 * config/cr16/cr16.c (IN_TARGET_CODE): Likewise.
1273 * config/cris/cris.c (IN_TARGET_CODE): Likewise.
1274 * config/darwin.c (IN_TARGET_CODE): Likewise.
1275 * config/epiphany/epiphany.c (IN_TARGET_CODE): Likewise.
1276 * config/epiphany/mode-switch-use.c (IN_TARGET_CODE): Likewise.
1277 * config/epiphany/resolve-sw-modes.c (IN_TARGET_CODE): Likewise.
1278 * config/fr30/fr30.c (IN_TARGET_CODE): Likewise.
1279 * config/frv/frv.c (IN_TARGET_CODE): Likewise.
1280 * config/ft32/ft32.c (IN_TARGET_CODE): Likewise.
1281 * config/h8300/h8300.c (IN_TARGET_CODE): Likewise.
1282 * config/i386/djgpp.c (IN_TARGET_CODE): Likewise.
1283 * config/i386/driver-i386.c (IN_TARGET_CODE): Likewise.
1284 * config/i386/driver-mingw32.c (IN_TARGET_CODE): Likewise.
1285 * config/i386/host-cygwin.c (IN_TARGET_CODE): Likewise.
1286 * config/i386/host-i386-darwin.c (IN_TARGET_CODE): Likewise.
1287 * config/i386/host-mingw32.c (IN_TARGET_CODE): Likewise.
1288 * config/i386/i386-c.c (IN_TARGET_CODE): Likewise.
1289 * config/i386/i386.c (IN_TARGET_CODE): Likewise.
1290 * config/i386/intelmic-mkoffload.c (IN_TARGET_CODE): Likewise.
1291 * config/i386/msformat-c.c (IN_TARGET_CODE): Likewise.
1292 * config/i386/winnt-cxx.c (IN_TARGET_CODE): Likewise.
1293 * config/i386/winnt-stubs.c (IN_TARGET_CODE): Likewise.
1294 * config/i386/winnt.c (IN_TARGET_CODE): Likewise.
1295 * config/i386/x86-tune-sched-atom.c (IN_TARGET_CODE): Likewise.
1296 * config/i386/x86-tune-sched-bd.c (IN_TARGET_CODE): Likewise.
1297 * config/i386/x86-tune-sched-core.c (IN_TARGET_CODE): Likewise.
1298 * config/i386/x86-tune-sched.c (IN_TARGET_CODE): Likewise.
1299 * config/ia64/ia64-c.c (IN_TARGET_CODE): Likewise.
1300 * config/ia64/ia64.c (IN_TARGET_CODE): Likewise.
1301 * config/iq2000/iq2000.c (IN_TARGET_CODE): Likewise.
1302 * config/lm32/lm32.c (IN_TARGET_CODE): Likewise.
1303 * config/m32c/m32c-pragma.c (IN_TARGET_CODE): Likewise.
1304 * config/m32c/m32c.c (IN_TARGET_CODE): Likewise.
1305 * config/m32r/m32r.c (IN_TARGET_CODE): Likewise.
1306 * config/m68k/m68k.c (IN_TARGET_CODE): Likewise.
1307 * config/mcore/mcore.c (IN_TARGET_CODE): Likewise.
1308 * config/microblaze/microblaze-c.c (IN_TARGET_CODE): Likewise.
1309 * config/microblaze/microblaze.c (IN_TARGET_CODE): Likewise.
1310 * config/mips/driver-native.c (IN_TARGET_CODE): Likewise.
1311 * config/mips/frame-header-opt.c (IN_TARGET_CODE): Likewise.
1312 * config/mips/mips.c (IN_TARGET_CODE): Likewise.
1313 * config/mmix/mmix.c (IN_TARGET_CODE): Likewise.
1314 * config/mn10300/mn10300.c (IN_TARGET_CODE): Likewise.
1315 * config/moxie/moxie.c (IN_TARGET_CODE): Likewise.
1316 * config/msp430/driver-msp430.c (IN_TARGET_CODE): Likewise.
1317 * config/msp430/msp430-c.c (IN_TARGET_CODE): Likewise.
1318 * config/msp430/msp430.c (IN_TARGET_CODE): Likewise.
1319 * config/nds32/nds32-cost.c (IN_TARGET_CODE): Likewise.
1320 * config/nds32/nds32-fp-as-gp.c (IN_TARGET_CODE): Likewise.
1321 * config/nds32/nds32-intrinsic.c (IN_TARGET_CODE): Likewise.
1322 * config/nds32/nds32-isr.c (IN_TARGET_CODE): Likewise.
1323 * config/nds32/nds32-md-auxiliary.c (IN_TARGET_CODE): Likewise.
1324 * config/nds32/nds32-memory-manipulation.c (IN_TARGET_CODE): Likewise.
1325 * config/nds32/nds32-pipelines-auxiliary.c (IN_TARGET_CODE): Likewise.
1326 * config/nds32/nds32-predicates.c (IN_TARGET_CODE): Likewise.
1327 * config/nds32/nds32.c (IN_TARGET_CODE): Likewise.
1328 * config/nios2/nios2.c (IN_TARGET_CODE): Likewise.
1329 * config/nvptx/mkoffload.c (IN_TARGET_CODE): Likewise.
1330 * config/nvptx/nvptx.c (IN_TARGET_CODE): Likewise.
1331 * config/pa/pa.c (IN_TARGET_CODE): Likewise.
1332 * config/pdp11/pdp11.c (IN_TARGET_CODE): Likewise.
1333 * config/powerpcspe/driver-powerpcspe.c (IN_TARGET_CODE): Likewise.
1334 * config/powerpcspe/host-darwin.c (IN_TARGET_CODE): Likewise.
1335 * config/powerpcspe/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
1336 * config/powerpcspe/powerpcspe-c.c (IN_TARGET_CODE): Likewise.
1337 * config/powerpcspe/powerpcspe-linux.c (IN_TARGET_CODE): Likewise.
1338 * config/powerpcspe/powerpcspe.c (IN_TARGET_CODE): Likewise.
1339 * config/riscv/riscv-builtins.c (IN_TARGET_CODE): Likewise.
1340 * config/riscv/riscv-c.c (IN_TARGET_CODE): Likewise.
1341 * config/riscv/riscv.c (IN_TARGET_CODE): Likewise.
1342 * config/rl78/rl78-c.c (IN_TARGET_CODE): Likewise.
1343 * config/rl78/rl78.c (IN_TARGET_CODE): Likewise.
1344 * config/rs6000/driver-rs6000.c (IN_TARGET_CODE): Likewise.
1345 * config/rs6000/host-darwin.c (IN_TARGET_CODE): Likewise.
1346 * config/rs6000/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
1347 * config/rs6000/rs6000-c.c (IN_TARGET_CODE): Likewise.
1348 * config/rs6000/rs6000-linux.c (IN_TARGET_CODE): Likewise.
1349 * config/rs6000/rs6000-p8swap.c (IN_TARGET_CODE): Likewise.
1350 * config/rs6000/rs6000-string.c (IN_TARGET_CODE): Likewise.
1351 * config/rs6000/rs6000.c (IN_TARGET_CODE): Likewise.
1352 * config/rx/rx.c (IN_TARGET_CODE): Likewise.
1353 * config/s390/driver-native.c (IN_TARGET_CODE): Likewise.
1354 * config/s390/s390-c.c (IN_TARGET_CODE): Likewise.
1355 * config/s390/s390.c (IN_TARGET_CODE): Likewise.
1356 * config/sh/sh-c.c (IN_TARGET_CODE): Likewise.
1357 * config/sh/sh-mem.cc (IN_TARGET_CODE): Likewise.
1358 * config/sh/sh.c (IN_TARGET_CODE): Likewise.
1359 * config/sh/sh_optimize_sett_clrt.cc (IN_TARGET_CODE): Likewise.
1360 * config/sh/sh_treg_combine.cc (IN_TARGET_CODE): Likewise.
1361 * config/sparc/driver-sparc.c (IN_TARGET_CODE): Likewise.
1362 * config/sparc/sparc-c.c (IN_TARGET_CODE): Likewise.
1363 * config/sparc/sparc.c (IN_TARGET_CODE): Likewise.
1364 * config/spu/spu-c.c (IN_TARGET_CODE): Likewise.
1365 * config/spu/spu.c (IN_TARGET_CODE): Likewise.
1366 * config/stormy16/stormy16.c (IN_TARGET_CODE): Likewise.
1367 * config/tilegx/mul-tables.c (IN_TARGET_CODE): Likewise.
1368 * config/tilegx/tilegx-c.c (IN_TARGET_CODE): Likewise.
1369 * config/tilegx/tilegx.c (IN_TARGET_CODE): Likewise.
1370 * config/tilepro/mul-tables.c (IN_TARGET_CODE): Likewise.
1371 * config/tilepro/tilepro-c.c (IN_TARGET_CODE): Likewise.
1372 * config/tilepro/tilepro.c (IN_TARGET_CODE): Likewise.
1373 * config/v850/v850-c.c (IN_TARGET_CODE): Likewise.
1374 * config/v850/v850.c (IN_TARGET_CODE): Likewise.
1375 * config/vax/vax.c (IN_TARGET_CODE): Likewise.
1376 * config/visium/visium.c (IN_TARGET_CODE): Likewise.
1377 * config/vms/vms-c.c (IN_TARGET_CODE): Likewise.
1378 * config/vms/vms-f.c (IN_TARGET_CODE): Likewise.
1379 * config/vms/vms.c (IN_TARGET_CODE): Likewise.
1380 * config/xtensa/xtensa.c (IN_TARGET_CODE): Likewise.
1382 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1383 Alan Hayward <alan.hayward@arm.com>
1384 David Sherwood <david.sherwood@arm.com>
1386 * tree-ssa-sccvn.c (vn_reference_lookup_3): Avoid repeated
1389 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1390 Alan Hayward <alan.hayward@arm.com>
1391 David Sherwood <david.sherwood@arm.com>
1393 * doc/generic.texi (VEC_SERIES_EXPR): Document.
1394 * doc/md.texi (vec_series@var{m}): Document.
1395 * tree.def (VEC_SERIES_EXPR): New tree code.
1396 * tree.h (build_vec_series): Declare.
1397 * tree.c (build_vec_series): New function.
1398 * cfgexpand.c (expand_debug_expr): Handle VEC_SERIES_EXPR.
1399 * tree-pretty-print.c (dump_generic_node): Likewise.
1400 * gimple-pretty-print.c (dump_binary_rhs): Likewise.
1401 * tree-inline.c (estimate_operator_cost): Likewise.
1402 * expr.c (expand_expr_real_2): Likewise.
1403 * optabs-tree.c (optab_for_tree_code): Likewise.
1404 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
1405 * fold-const.c (const_binop): Fold VEC_SERIES_EXPRs of constants.
1406 * expmed.c (make_tree): Handle VEC_SERIES.
1407 * optabs.def (vec_series_optab): New optab.
1408 * optabs.h (expand_vec_series_expr): Declare.
1409 * optabs.c (expand_vec_series_expr): New function.
1410 * tree-vect-generic.c (expand_vector_operations_1): Check that
1411 the operands also have vector type.
1413 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1414 Alan Hayward <alan.hawyard@arm.com>
1415 David Sherwood <david.sherwood@arm.com>
1417 * doc/generic.texi (VEC_DUPLICATE_EXPR): Document.
1418 (VEC_COND_EXPR): Add missing @tindex.
1419 * doc/md.texi (vec_duplicate@var{m}): Document.
1420 * tree.def (VEC_DUPLICATE_EXPR): New tree codes.
1421 * tree.c (build_vector_from_val): Add stubbed-out handling of
1422 variable-length vectors, using VEC_DUPLICATE_EXPR.
1423 (uniform_vector_p): Handle VEC_DUPLICATE_EXPR.
1424 * cfgexpand.c (expand_debug_expr): Likewise.
1425 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
1426 * tree-inline.c (estimate_operator_cost): Likewise.
1427 * tree-pretty-print.c (dump_generic_node): Likewise.
1428 * tree-vect-generic.c (ssa_uniform_vector_p): Likewise.
1429 * fold-const.c (const_unop): Fold VEC_DUPLICATE_EXPRs of a constant.
1430 (test_vec_duplicate_folding): New function.
1431 (fold_const_c_tests): Call it.
1432 * optabs.def (vec_duplicate_optab): New optab.
1433 * optabs-tree.c (optab_for_tree_code): Handle VEC_DUPLICATE_EXPR.
1434 * optabs.h (expand_vector_broadcast): Declare.
1435 * optabs.c (expand_vector_broadcast): Make non-static. Try using
1436 vec_duplicate_optab.
1437 * expr.c (store_constructor): Try using vec_duplicate_optab for
1439 (expand_expr_real_2): Handle VEC_DUPLICATE_EXPR.
1441 2017-12-15 Markus Trippelsdorf <markus@trippelsdorf.de>
1444 * config/i386/x86-tune-costs.h (skylake_cost, core_cost): Increase
1445 div/mod latencies a bit.
1447 2017-12-15 Jeff Law <law@redhat.com>
1449 PR tree-optimization/36550
1450 * tree-ssa-threadupdate.c (count_stmts_and_phis_in_block): New.
1451 (mark_threaded_blocks): Rewrite code to avoid block copying when
1452 optimizing for size. Don't pessimize blocks which will be
1453 copied, but all the statements will be dead.
1455 2017-12-15 Alexandre Oliva <aoliva@redhat.com>
1457 PR tree-optimization/81165
1458 * tree-ssa-threadupdate.c (uses_in_bb): New.
1459 (estimate_threading_killed_stmts): New.
1460 * tree-ssa-threadupdate.h (estimate_threading_killed_stmts): Prototype.
1461 * tree-ssa-threadedge.c
1462 (record_temporary_equivalences_from_stmts_at_dest): Expand limit
1465 2017-12-15 Jeff Law <law@redhat.com>
1467 PR tree-optimization/83410
1468 * tree-ssa-threadupdate.c (thread_block_1): Avoid certain jump
1469 threads when parallelizing loops.
1471 2017-12-15 Jakub Jelinek <jakub@redhat.com>
1473 * tree-core.h (struct attribute_spec): Swap affects_type_identity and
1475 * config/alpha/alpha.c (vms_attribute_table): Swap
1476 affects_type_identity and handler fields, adjust comments.
1477 * config/mips/mips.c (mips_attribute_table): Likewise.
1478 * config/visium/visium.c (visium_attribute_table): Likewise.
1479 * config/epiphany/epiphany.c (epiphany_attribute_table): Likewise.
1480 * config/microblaze/microblaze.c (microblaze_attribute_table):
1482 * config/spu/spu.c (spu_attribute_table): Likewise.
1483 * config/mcore/mcore.c (mcore_attribute_table): Likewise.
1484 * config/arc/arc.c (arc_attribute_table): Likewise.
1485 * config/m68k/m68k.c (m68k_attribute_table): Likewise.
1486 * config/v850/v850.c (v850_handle_interrupt_attribute,
1487 v850_handle_data_area_attribute): Formatting fixes.
1488 (v850_attribute_table): Swap affects_type_identity and handler
1489 fields, adjust comments.
1490 * config/m32r/m32r.c (m32r_attribute_table): Likewise.
1491 * config/arm/arm.c (arm_attribute_table): Likewise.
1492 * config/avr/avr.c (avr_attribute_table): Likewise.
1493 * config/s390/s390.c (s390_attribute_table): Likewise.
1494 * config/sh/sh.c (sh_attribute_table): Likewise.
1495 * config/i386/i386.c (ix86_handle_cconv_attribute,
1496 ix86_handle_callee_pop_aggregate_return): Formatting fixes.
1497 (ix86_attribute_table): Swap affects_type_identity and handler
1498 fields, adjust comments.
1499 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
1500 * config/sparc/sparc.c (sparc_attribute_table): Likewise.
1501 * config/m32c/m32c.c (m32c_attribute_table): Likewise.
1502 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Likewise.
1503 * config/ia64/ia64.c (ia64_attribute_table): Likewise.
1504 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
1505 * config/rx/rx.c (rx_attribute_table): Likewise.
1506 * config/cr16/cr16.c (cr16_attribute_table): Likewise.
1507 * config/h8300/h8300.c (h8300_attribute_table): Likewise.
1508 * config/nvptx/nvptx.c (nvptx_attribute_table): Likewise.
1509 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Likewise.
1510 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
1511 * config/stormy16/stormy16.c (xstormy16_attribute_table): Likewise.
1512 * config/bfin/bfin.c (bfin_attribute_table): Likewise.
1513 * config/rs6000/rs6000.c (rs6000_attribute_table): Likewise.
1514 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
1515 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
1516 * doc/plugins.texi (user_attr): Likewise. Add NULL for
1518 * attribs.c (empty_attribute_table): Swap affects_type_identity and
1520 (register_scoped_attributes, decl_attributes): Formatting fixes.
1522 PR tree-optimization/83269
1523 * fold-const.c (fold_binary_loc): Perform (-A) - B -> (-B) - A
1524 subtraction in arg0's type if type is signed and arg0 is unsigned.
1528 * match.pd ((T)(P + A) - (T)P -> (T) A): Use @@0 instead of @0 and
1529 convert? on @0 instead of convert. Check type of @1, not @0.
1530 ((T)P - (T)(P + A) -> -(T) A): Use @@0 instead of @0 and
1531 convert? on @0 instead of convert. Check type of @1, not @0.
1532 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Use @@0 instead of @0,
1533 only optimize if either both @1 and @2 types are narrower
1534 precision, or both are wider or equal precision, and in the former
1535 case only if both have undefined overflow.
1537 2017-12-15 Richard Biener <rguenther@suse.de>
1540 * internal-fn.def (IFN_NOP): Add.
1541 * internal-fn.c (expand_NOP): Do nothing.
1542 * lto-streamer-in.c (input_function): Instead of removing
1543 sanitizer calls replace them with IFN_NOP calls.
1545 2017-12-15 Richard Sandiford <richard.sandiford@linaro.org>
1546 Alan Hayward <alan.hayward@arm.com>
1547 David Sherwood <david.sherwood@arm.com>
1549 * dse.c (store_info, read_info_type): Replace begin and end with
1551 (print_range): New function.
1552 (set_all_positions_unneeded, any_positions_needed_p)
1553 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Update
1555 (record_store): Likewise. Optimize the case in which all positions
1557 (get_stored_val): Replace read_begin and read_end with read_offset
1559 (replace_read): Update call accordingly.
1561 2017-12-15 Bin Cheng <bin.cheng@arm.com>
1563 * gimple-loop-interchange.cc (STMT_COST_RATIO): New macro.
1564 (loop_cand::m_num_stmts, loop_cand::m_const_init_reduc): New members.
1565 (loop_cand::loop_cand): Initialize above members.
1566 (loop_cand::supported_operations): Delete.
1567 (loop_cand::can_interchange_p): Inline above function.
1568 (loop_cand::classify_simple_reduction): Record number of constant
1569 initialized simple reductions.
1570 (should_interchange_loops): New parameters. Check stmt cost of loops
1572 (tree_loop_interchange::interchange): Prepare stmt cost of outer loop.
1573 Update call to should_interchange_loops.
1574 (should_interchange_loop_nest): Update call to
1575 should_interchange_loops.
1577 2017-12-15 Eric Botcazou <ebotcazou@adacore.com>
1580 * ggc-page.c (HOST_BITS_PER_PTR): Do not define here...
1581 * hwint.h (HOST_BITS_PER_PTR): ...but here instead.
1582 * config/i386/xm-mingw32.h (HOST_BITS_PER_PTR): Delete.
1584 2017-12-15 Julia Koval <julia.koval@intel.com>
1586 * config/i386/i386-builtin.def (__builtin_ia32_vaesenclast_v16qi,
1587 __builtin_ia32_vaesenclast_v32qi, __builtin_ia32_vaesenclast_v64qi): New.
1588 * config/i386/sse.md (vaesenclast_<mode>): New pattern.
1589 * config/i386/vaesintrin.h (_mm256_aesenclast_epi128,
1590 _mm512_aesenclast_epi128, _mm_aesenclast_epi128): New intrinsics.
1592 2017-12-15 Julia Koval <julia.koval@intel.com>
1594 * config/i386/i386-builtin.def (__builtin_ia32_vaesenc_v16qi,
1595 __builtin_ia32_vaesenc_v32qi, __builtin_ia32_vaesenc_v64qi): New.
1596 * config/i386/sse.md (vaesenc_<mode>): New pattern.
1597 * config/i386/vaesintrin.h (_mm256_aesenc_epi128, _mm512_aesenc_epi128,
1598 _mm_aesenc_epi128): New intrinsics.
1600 2017-12-15 Julia Koval <julia.koval@intel.com>
1602 * config/i386/i386-builtin.def (__builtin_ia32_vaesdeclast_v16qi,
1603 __builtin_ia32_vaesdeclast_v32qi, __builtin_ia32_vaesdeclast_v64qi): New.
1604 * config/i386/sse.md (vaesdeclast_<mode>): New pattern.
1605 * config/i386/vaesintrin.h (_mm256_aesdeclast_epi128,
1606 _mm512_aesdeclast_epi128, _mm_aesdeclast_epi128): New intrinsics.
1608 2017-12-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1610 * gimple-ssa-strength-reduction.c (analyze_increments):
1611 Distinguish replacement costs for constant strides from those for
1614 2017-12-14 Jakub Jelinek <jakub@redhat.com>
1616 * var-tracking.c (variable_tracking_main_1): Formatting fix.
1618 2017-12-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
1620 * doc/invoke.texi: Document -Wcast-function-type.
1621 * recog.h (stored_funcptr): Change signature.
1622 * tree-dump.c (dump_node): Avoid warning.
1623 * typed-splay-tree.h (typed_splay_tree): Avoid warning.
1625 2017-12-14 Qing Zhao <qing.zhao@oracle.com>
1628 * gimple-fold.c (get_range_strlen): Add the handling of non-member
1631 2017-12-14 David Malcolm <dmalcolm@redhat.com>
1633 PR tree-optimization/83312
1634 * domwalk.h (dom_walker::dom_walker): Fix typo in comment.
1635 * tree-cfg.c (find_taken_edge): Update to handle NULL_TREE for
1636 "val" param, and to cope with arbitrary basic blocks.
1637 (find_taken_edge_cond_expr): Add "cond_stmt" param and use it to
1638 handle NULL_TREE for "val", dropping "bb" param.
1639 (find_taken_edge_switch_expr): Make "switch_stmt" param const and
1640 drop "bb" param. Handle NULL_TREE for "val".
1641 (find_case_label_for_value): Make "switch_stmt" param const.
1642 * tree-vrp.c (class check_array_bounds_dom_walker): New subclass
1644 (vrp_prop::check_all_array_refs): Reimplement as...
1645 (check_array_bounds_dom_walker::before_dom_children): ...this new
1646 vfunc. Replace linear search through BB block list, excluding
1647 those with non-executable in-edges via dominator walk.
1649 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1651 * config/arm/arm.opt (mverbose-cost-dump): New option.
1652 * config/arm/arm.c (arm_rtx_costs): Use it.
1654 2017-12-14 Andreas Schwab <schwab@linux-m68k.org>
1657 * reload1.c (emit_input_reload_insns): Skip debug markers.
1659 2017-12-14 Alexandre Oliva <aoliva@redhat.com>
1661 * config/i386/i386.c (rest_of_insert_endbranch): Use call loc
1665 * config/arc/arc.c (hwloop_optimize): Skip debug insns.
1666 * config/sh/sh-protos.h (sh_find_set_of_reg): Adjust.
1667 * config/sh/sh.c: Skip debug insns besides notes.
1668 * config/sh/sh.md: Likewise.
1669 * config/sh/sh_treg_combine.cc: Likewise.
1670 * config/sh/sync.md: Likewise.
1672 2017-12-14 Tom de Vries <tom@codesourcery.com>
1674 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
1675 Add item for weak_undefined.
1677 2017-12-14 Richard Biener <rguenther@suse.de>
1679 PR tree-optimization/67842
1680 * sese.h (bb_in_region): Remove #if 0'ed code.
1682 2017-12-14 Richard Biener <rguenther@suse.de>
1684 PR tree-optimization/83326
1685 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
1686 may_be_zero parameter and handle it by not marking the first
1687 peeled copy as not exiting the loop.
1688 (try_peel_loop): Likewise.
1689 (canonicalize_loop_induction_variables): Use number_of_iterations_exit
1690 to handle the case of constant or zero iterations and perform
1691 loop header copying on-the-fly.
1693 2017-12-14 Richard Biener <rguenther@suse.de>
1695 PR tree-optimization/83418
1696 * vr-values.c (vr_values::extract_range_for_var_from_comparison_expr):
1697 Instead of asserting we don't get unfolded comparisons deal with
1700 2017-12-14 Jakub Jelinek <jakub@redhat.com>
1703 * var-tracking.c (vt_initialize): Ignore non-DEBUG_INSNs outside of
1704 basic blocks. Assert debug bind insns don't appear outside of bbs,
1705 don't reset them. Assert insns without BLOCK_FOR_INSN are outside of
1708 PR tree-optimization/83198
1709 * gimple-ssa-sprintf.c (format_floating): Set type solely based on
1710 dir.modifier, regardless of TREE_TYPE (arg). Assume non-REAL_CST
1711 value if arg is a REAL_CST with incompatible type.
1713 2017-12-14 Sudakshina Das <sudi.das@arm.com>
1714 Bin Cheng <bin.cheng@arm.com>
1717 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Move LTGT
1719 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Add
1722 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1724 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75,
1725 cortex-a75.cortex-a55): Specify AARCH64_FL_F16 in the arch features.
1727 2017-12-14 Richard Sandiford <richard.sandiford@linaro.org>
1728 Alan Hayward <alan.hayward@arm.com>
1729 David Sherwood <david.sherwood@arm.com>
1731 * poly-int.h: New file.
1732 * poly-int-types.h: Likewise.
1733 * coretypes.h: Include them.
1734 (POLY_INT_CONVERSION): Define.
1735 * target.def (estimated_poly_value): New hook.
1736 * doc/tm.texi.in (TARGET_ESTIMATED_POLY_VALUE): New hook.
1737 * doc/tm.texi: Regenerate.
1738 * doc/poly-int.texi: New file.
1739 * doc/gccint.texi: Include it.
1740 * doc/rtl.texi: Describe restrictions on subreg modes.
1741 * Makefile.in (TEXI_GCCINT_FILES): Add poly-int.texi.
1742 * genmodes.c (NUM_POLY_INT_COEFFS): Provide a default definition.
1743 (emit_insn_modes_h): Emit a definition of NUM_POLY_INT_COEFFS.
1744 * targhooks.h (default_estimated_poly_value): Declare.
1745 * targhooks.c (default_estimated_poly_value): New function.
1746 * target.h (estimated_poly_value): Likewise.
1747 * wide-int.h (WI_UNARY_RESULT): Use wi::binary_traits.
1748 (wi::unary_traits): Delete.
1749 (wi::binary_traits::signed_shift_result_type): Define for
1750 offset_int << HOST_WIDE_INT, etc.
1751 (generic_wide_int::operator <<=): Define for all types and use
1752 wi::lshift instead of <<.
1753 (wi::hwi_with_prec): Add a default constructor.
1754 (wi::ints_for): New class.
1755 (operator <<): Define for all wide-int types.
1756 (operator /): New function.
1757 (operator %): Likewise.
1758 * selftest.h (ASSERT_KNOWN_EQ, ASSERT_KNOWN_EQ_AT, ASSERT_MAYBE_NE)
1759 (ASSERT_MAYBE_NE_AT): New macros.
1761 2017-12-13 Eric Botcazou <ebotcazou@adacore.com>
1762 Dominik Vogt <vogt@linux.vnet.ibm.com>
1765 * emit-rtl.c (init_emit): Remove ??? comment.
1766 * explow.c (get_dynamic_stack_size): Take known alignment of stack
1767 pointer + STACK_DYNAMIC_OFFSET into account in lieu of STACK_BOUNDARY.
1768 * config/sparc/sparc.h (INIT_EXPANDERS): In 32-bit mode, lower the
1769 alignment of 3 virtual registers to BITS_PER_WORD.
1771 * config/sparc/sparc.c (sparc_compute_frame_size): Simplify.
1773 2017-12-13 Peter Bergner <bergner@vnet.ibm.com>
1775 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_HTM_NO_SUSPEND): New define.
1776 * config/rs6000/rs6000.c (cpu_supports_info): Use it.
1778 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
1781 * reload1.c (eliminate_regs_in_insn): Skip debug markers.
1783 2017-12-13 Jakub Jelinek <jakub@redhat.com>
1785 * tree-cfg.c (verify_gimple_in_cfg): Verify no non-label stmts
1786 with the exception of debug begin stmt markers appear before
1790 * final.c (rest_of_handle_final): Call variable_tracking_main only
1791 if !flag_var_tracking.
1793 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
1794 Jakub Jelinek <jakub@redhat.com>
1798 * tree-cfgcleanup.c (remove_forwarder_block): Keep after
1799 labels debug stmts that can only appear after labels.
1801 2017-12-13 Alexander Monakov <amonakov@ispras.ru>
1803 PR rtl-optimization/82398
1804 * sel-sched.c (sel_rank_for_schedule): Fix check for zero
1805 EXPR_USEFULNESS in priority comparison.
1807 2017-12-13 Segher Boessenkool <segher@kernel.crashing.org>
1809 PR rtl-optimization/83393
1810 * combine.c (move_deaths): If reg_stat points to a too new insn in
1811 last_death, do not use it: find the proper insn instead.
1813 2017-12-12 Jeff Law <law@redhat.com>
1815 PR tree-optimization/83298
1816 PR tree-optimization/83362
1817 PR tree-optimization/83383
1818 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Make
1819 push_value_range a public interface. Add new argument to
1820 record_ranges_from_stmt.
1821 * gimple-ssa-evrp-analyze.c
1822 (evrp_range_analyzer::record_ranges_from_stmt): Add new argument.
1823 Update comments. Handle recording temporary equivalences.
1824 * tree-ssa-dom.c (dom_opt_opt_walker::before_dom_children): Add
1825 new argument to call to evrp_range_analyzer::record_ranges_from_stmt.
1826 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Likewise.
1827 * tree-ssa-threadedge.c: Include alloc-pool.h, vr-values.h and
1828 gimple-ssa-evrp-analyze.h.
1829 (record_temporary_equivalences_from_phis): Add new argument. When
1830 the PHI arg is an SSA_NAME, set the result's range to the range
1832 (record_temporary_equivalences_from_stmts_at_dest): Record ranges
1833 from statements too.
1834 (thread_through_normal_block): Accept new argument, evrp_range_analyzer.
1835 Pass it down to children as needed.
1836 (thread_outgoing_edges): Likewise.
1837 (thread_across_edge): Likewise. Push/pop range state as needed.
1838 * tree-ssa-threadedge.h (thread_outgoing_edges): Update prototype.
1840 2017-12-12 Julia Koval <julia.koval@intel.com>
1842 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
1843 (PTA_CANNONLAKE): Remove PTA_CLWB.
1845 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1848 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
1849 (vcondv2div2df): Likewise.
1850 (vconduv2dfv2di): Likewise.
1852 2017-12-12 Jakub Jelinek <jakub@redhat.com>
1854 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
1855 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
1856 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
1857 ATTR_MATHFN_FPROUNDING_ERRNO.
1859 2017-12-12 Richard Biener <rguenther@suse.de>
1861 PR tree-optimization/83385
1862 * graphite-scop-detection.c (get_order, order): Remove.
1863 (bb_to_rpo): New global.
1865 (build_scops): Sort pbbs in RPO order.
1867 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
1869 * combine.c (simplify_set): Do not transform subregs to zero_extends
1870 if the destination mode is a vector mode.
1872 2017-12-12 Jakub Jelinek <jakub@redhat.com>
1874 PR tree-optimization/83359
1875 * tree-cfg.h (fold_loop_internal_call): Declare.
1876 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
1877 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
1878 (find_loop_dist_alias): New function.
1879 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
1880 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
1883 PR tree-optimization/80631
1884 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
1885 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
1886 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
1887 hardcoding zero as the value if COND_EXPR is never true. For
1888 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
1889 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
1890 hardcoding MAX_EXPR as the reduction operation.
1891 (is_nonwrapping_integer_induction): Allow negative step.
1892 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
1893 vect_create_epilog_for_reduction, if no value is suitable, don't
1894 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
1896 2017-12-12 Richard Biener <rguenther@suse.de>
1898 PR tree-optimization/81889
1899 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
1900 range info from the non-wrapping IV instead of just the range
1903 2017-12-12 Julia Koval <julia.koval@intel.com>
1905 * config.gcc: Add vaesintrin.h.
1906 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
1908 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
1909 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
1911 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
1912 * config/i386/immintrin.h: Include vaesintrin.h.
1913 * config/i386/sse.md (vaesdec_<mode>): New pattern.
1914 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
1915 _mm_aesdec_epi128): New intrinsics.
1917 2017-12-12 Julia Koval <julia.koval@intel.com>
1919 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
1920 OPTION_MASK_ISA_VAES_UNSET): New.
1921 (ix86_handle_option): Handle -mvaes.
1922 * config/i386/cpuid.h: Define bit_VAES.
1923 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
1924 * config/i386/i386-c.c (__VAES__): New.
1925 * config/i386/i386.c (ix86_target_string): Add -mvaes.
1926 (ix86_valid_target_attribute_inner_p): Ditto.
1927 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
1928 * config/i386/i386.opt: Add -mvaes.
1929 * doc/invoke.texi: Ditto.
1931 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
1933 * debug.h (gcc_debug_hooks): Add inline_entry.
1934 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
1935 * debug.c (do_nothing_debug_hooks): Likewise.
1936 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
1937 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
1938 (dwarf2_lineno_debug_hooks): Likewise.
1940 * common.opt (gstatement-frontiers): New, setting
1941 debug_nonbind_markers_p.
1942 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
1943 * toplev.c (process_options): Autodetect value for debug statement
1945 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
1946 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
1948 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
1949 markers. Integrate source bind into debug stmt expand loop.
1950 (pass_expand::execute): Check debug marker limit. Avoid deep
1951 TER and expand debug locations for debug bind insns only.
1952 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
1954 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
1955 * final.c (reemit_insn_block_notes): Take current block from
1956 nonbind markers. Declare note where it's first set.
1957 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
1958 begin stmt markers if enabled.
1959 (notice_source_line): Handle nonbind markers. Fail if their
1960 location is unknown or that of builtins.
1961 (rest_of_handle_final): Convert begin stmt markers to notes if
1962 var-tracking didn't run.
1963 (rest_of_clean_state): Skip begin stmt markers.
1964 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
1966 * function.c (allocate_struct_function): Set begin_stmt_markers.
1967 * function.h (struct function): Add debug_marker_count counter
1968 and debug_nonbind_markers flag.
1969 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
1970 * gimple-low.c (lower_function_body): Adjust
1971 debug_nonbind_markers.
1972 (lower_stmt): Drop or skip gimple debug stmts.
1973 (lower_try_catch): Skip debug stmts.
1974 * gimple.c (gimple_build_debug_begin_stmt): New.
1975 (gimple_copy): Increment debug_marker_count if copying one.
1976 * gimple.h (gimple_build_debug_begin_stmt): Declare.
1977 * gimplify.c (rexpr_location): New.
1978 (rexpr_has_location): New.
1979 (warn_switch_unreachable_r): Handle gimple debug stmts.
1980 (shortcut_cond_r): Call expr_location.
1982 (find_goto_label): New.
1983 (shortcut_cond_expr): Call expr_has_location, expr_location, and
1985 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
1987 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
1988 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
1989 (LANG_HOOKS_INITIALIZER): ... this.
1990 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
1991 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
1993 (update_ebb_live_info): Skip debug insn markers.
1994 * lra.c (debug_insn_static_data): Rename to...
1995 (debug_bind_static_data): ... this.
1996 (debug_marker_static_data): New.
1997 (lra_set_insn_recog_data): Select one of the above depending
1999 (lra_update_isn_regno_info): Don't assume debug insns have
2001 (push_insns): Skip debug insns.
2002 * lto-streamer-in.c (input_function): Drop debug stmts
2003 depending on active options. Adjust debug_nonbind_markers.
2004 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
2005 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
2006 begin stmt marker notes.
2007 (print_insn): Likewise.
2008 * recog.c (extract_insn): Recognize rtl for debug markers.
2009 * rtl.def (DEBUG_MARKER): New.
2010 * tree-inline.c: Include params.h.
2011 (remap_gimple_stmt): Handle nonbind markers.
2012 (maybe_move_debug_stmts_to_successors): Likewise.
2013 (copy_debug_stmt): Likewise.
2014 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
2015 markers regardless of no side effects.
2016 (tsi_link_before): Don't update container's side effects when adding
2017 a begin stmt marker.
2018 (tsi_link_after): Likewise.
2019 (expr_first): Skip begin stmt markers.
2020 (expr_last): Likewise.
2021 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
2022 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
2023 Disregard nonbind markers.
2024 * tree.c (make_node_stat): Don't set side effects for begin stmt
2026 (build1_stat): Likewise.
2027 * tree.def (DEBUG_BEGIN_STMT): New.
2028 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
2029 * var-tracking.c (delete_debug_insns): Renamed to...
2030 (delete_vta_debug_insns): ... this.
2031 (reemit_marker_as_note): New.
2032 (vt_initialize): Reemit markers.
2033 (delete_vta_debug_insns): Likewise.
2034 (vt_debug_insns_local): Reemit or delete markers.
2035 (variable_tracking_main_1): Likewise.
2036 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
2037 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
2038 (gimple_debug_nonbind_marker_p): New.
2039 (gimple_build_debug_bind): Adjust.
2040 (gimple_build_debug_begin_stmt): New.
2041 * doc/invoke.texi (max-debug-marker-count): New param.
2042 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
2043 (debug_parameter_ref, debug_marker): New.
2044 (NOTE_INSN_BEGIN_STMT): New.
2045 (DEBUG_INSN): Describe begin stmt markers.
2047 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
2048 without debug insns, we wouldn't, but clean up debug insns
2049 after a control flow insn nevertheless.
2051 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
2052 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
2053 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
2054 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
2055 (rtl_tidy_fallthru_edge): Likewise.
2056 (rtl_verify_fallthru): Likewise.
2057 (rtl_verify_bb_layout): Likewise.
2058 (skip_insns_after_block): Likewise.
2059 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
2060 * dwarf2out.c: Include print-rtl.h.
2061 (dwarf2out_next_real_insn): New.
2062 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
2063 Dump debug binds in asm comments.
2064 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
2065 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
2066 callers to use gsi_start_nondebug_bb instead.
2067 (gsi_after_labels): Skip gimple debug stmts.
2068 (gsi_start_nondebug): New.
2069 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
2070 (proper_loop_form_for_interchange): Adjust.
2071 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
2072 * gimple.h (gimple_seq_last_nondebug_stmt): New.
2073 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
2074 (collect_fallthrough_labels): Likewise.
2075 (should_warn_for_implicit_fallthrough): Likewise.
2076 (warn_implicit_fallthrough_r): Likewise.
2077 (expand_FALLTHROUGH_r): Likewise.
2078 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
2079 (graphite_copy_stmts_from_block): Skip nonbind markers.
2080 * haifa-sched.c (sched_extend_bb): Skip debug insns.
2081 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
2082 * jump.c (clean_barriers): Skip debug insns.
2083 * omp-expand.c (expand_parallel_call): Skip debug insns.
2084 (expand_task_call): Likewise.
2085 (remove_exit_barrier): Likewise.
2086 (expand_omp_taskreg): Likewise.
2087 (expand_omp_for_init_counts): Likewise.
2088 (expand_omp_for_generic): Likewise.
2089 (expand_omp_for_static_nochunk): Likewise.
2090 (expand_omp_for_static_chunk): Likewise.
2091 (expand_omp_simd): Likewise.
2092 (expand_omp_taskloop_for_outer): Likewise.
2093 (expand_omp_taskloop_for_inner): Likewise.
2094 (expand_oacc_for): Likewise.
2095 (expand_omp_sections): Likewise.
2096 (expand_omp_single): Likewise.
2097 (expand_omp_synch): Likewise.
2098 (expand_omp_atomic_load): Likewise.
2099 (expand_omp_atomic_store): Likewise.
2100 (expand_omp_atomic_fetch_op): Likewise.
2101 (expand_omp_atomic_pipeline): Likewise.
2102 (expand_omp_atomic_mutex): Likewise.
2103 (expand_omp_target): Likewise.
2104 (grid_expand_omp_for_loop): Likewise.
2105 (grid_expand_target_grid_body): Likewise.
2106 (build_omp_regions_1): Likewise.
2107 * omp-low.c (check_combined_parallel): Skip debug stmts.
2108 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
2109 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
2110 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
2112 * tree-cfg.c (make_blobs_1): Skip debug stmts.
2113 (make_edges): Likewise.
2114 (cleanup_dead_labels): Likewise.
2115 (gimple_can_merge_blocks_p): Likewise.
2116 (stmt_starts_bb_p): Likewise.
2117 (gimple_block_label): Likewise.
2118 (gimple_redirect_edge_and_branch): Likewise.
2119 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
2121 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
2123 * tree-pretty-print (print_declaration): Omit initializer in slim
2125 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
2127 (eliminate_unnecessary_stmts): Stabilize block removal order.
2128 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
2129 * var-tracking.c (get_first_insn): New.
2130 (vt_emit_notes): Call it.
2131 (vt_initialize): Walk any insns before the first BB.
2132 (delete_debug_insns): Likewise.
2134 * gimple.h (enum gimple_debug_subcode): Add
2135 GIMPLE_DEBUG_BEGIN_STMT.
2136 (gimple_debug_begin_stmt_p): New.
2137 (gimple_debug_nonbind_marker_p): New.
2138 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
2139 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
2140 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
2141 * insn-notes.def (BEGIN_STMT): New.
2142 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
2143 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
2144 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
2145 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
2146 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
2147 (INSN_DEBUG_MARKER_KIND): New.
2148 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
2149 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
2150 (INSN_VAR_LOCATION_PTR): New.
2151 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
2153 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
2154 deps only if debug bind insns are enabled.
2155 (pass_expand::execute): Avoid deep TER and expand
2156 debug locations for debug bind insns only.
2157 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
2158 debug stmts special handling down to debug bind stmts.
2159 * combine.c (try_combine): Narrow debug insns special handling
2160 down to debug bind insns.
2161 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
2162 Narrow debug insns preexisting special handling down to debug
2164 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
2165 handling down to debug bind insns.
2166 * function.c (instantiate_virtual_regs): Skip debug markers,
2167 adjust handling of debug binds.
2168 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
2169 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
2170 * haifa-sched.c (schedule_insn): Narrow special handling of debug
2171 insns to debug bind insns.
2172 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
2173 special handling of debug stmts to debug bind stmts.
2174 * ipa-split.c (split_function): Likewise.
2175 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
2176 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
2178 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
2179 * regrename.c (build_def_use): Likewise.
2180 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
2181 (pass_cprop_hardreg): Narrow special casing of debug insns to
2183 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
2184 * reload1.c (reload): Likewise.
2185 * sese.c (sese_insert_phis_for_liveouts): Narrow special
2186 casing of debug stmts to debug bind stmts.
2187 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
2188 * ssa-iterators.h (num_imm_uses): Likewise.
2189 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
2190 debug stmts to debug bind stmts.
2191 * tree-inline.c (tree_function_versioning): Narrow special casing
2192 of debug stmts to debug bind stmts.
2193 * tree-loop-distribution.c (generate_loops_for_partition):
2194 Narrow special casing of debug stmts to debug bind stmts.
2195 * tree-sra.c (analyze_access_subtree): Narrow special casing
2196 of debug stmts to debug bind stmts.
2197 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
2198 stmts to debug bind stmts.
2199 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
2200 casing of debug stmts to debug bind stmts.
2201 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
2202 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
2203 casing of debug stmts to debug bind stmts.
2204 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
2206 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
2207 debug stmts to debug bind stmts.
2208 (gimple_replace_ssa_lhs): Likewise.
2209 (insert_debug_temp_for_var_def): Likewise.
2210 (insert_debug_temps_for_defs): Likewise.
2211 (reset_debug_uses): Likewise.
2212 * tree-ssanames.c (release_ssa_name_fn): Likewise.
2213 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
2214 (adjust_debug_stmts): Likewise.
2215 (adjust_phi_and_debug_stmts): Likewise.
2216 (vect_do_peeling): Likewise.
2217 * tree-vect-loop.c (vect_transform_loop): Likewise.
2218 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
2219 * var-tracking.c (adjust_mems): Narrow special casing of debug
2220 insns to debug bind insns.
2221 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
2222 (compute_bb_dataflow, vt_find_locations): Likewise.
2223 (vt_expand_loc, emit_notes_for_changes): Likewise.
2224 (vt_init_cfa_base): Likewise.
2225 (vt_emit_notes): Likewise.
2226 (vt_initialize): Likewise.
2227 (vt_finalize): Likewise.
2229 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
2230 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
2231 (next_nonnote_nondebug_insn_bb): New.
2232 (prev_nonnote_nondebug_insn_bb): New.
2233 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
2234 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
2235 (prev_nonnote_nondebug_insn_bb): Declare.
2236 (next_nonnote_nondebug_insn_bb): Declare.
2237 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
2238 * cfgrtl.c (get_last_bb_insn): Likewise.
2239 * lra.c (push_insns): Likewise.
2241 2017-12-11 David Malcolm <dmalcolm@redhat.com>
2244 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
2246 * selftest.c (selftest::test_runner::test_runner): New ctor.
2247 (selftest::test_runner::~test_runner): New dtor.
2248 * selftest.h (class selftest::test_runner): New class.
2250 2017-12-11 Carl Love <cel@us.ibm.com>
2252 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
2253 vec_extract_fp32_from_shortl]): Add #defines.
2254 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
2255 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
2256 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
2257 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
2258 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
2259 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
2260 * doc/extend.texi: Add documentation for the added builtins.
2262 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
2264 PR rtl-optimization/80693
2265 PR rtl-optimization/81019
2266 PR rtl-optimization/81020
2267 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
2268 are not mentioned in i3. Place the REG_UNUSED note on i2,
2269 possibly modified to REG_DEAD, if it did not originate in i3.
2271 2017-12-11 Jakub Jelinek <jakub@redhat.com>
2273 * recog.c (store_data_bypass_p_1): New function.
2274 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
2275 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
2277 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
2279 PR rtl-optimization/83361
2280 * ifcvt.c (if_convert): Call fixup_partitions.
2282 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
2284 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
2285 early folding of splat_u{8,16,32}.
2287 2017-12-11 Jakub Jelinek <jakub@redhat.com>
2289 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
2290 output_operand_lossage first argument with capital letter.
2291 (aarch64_override_options): Don't start error and sorry first argument
2292 with capital letter.
2294 2017-12-11 Andi Kleen <ak@linux.intel.com>
2296 PR gcov-profile/83355
2297 * auto-profile.c (string_table::get_index_by_decl): Don't
2298 recurse when abstract origin points to itself.
2300 2017-12-11 Bin Cheng <bin.cheng@arm.com>
2302 PR tree-optimization/83320
2303 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
2304 (prune_datarefs_not_in_loop): Ditto.
2306 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
2308 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
2310 2017-12-10 Jakub Jelinek <jakub@redhat.com>
2312 PR tree-optimization/83337
2313 * gimple-loop-interchange.cc (compute_access_stride): Handle
2314 bitfield DRs properly.
2316 2017-12-09 Jakub Jelinek <jakub@redhat.com>
2318 PR tree-optimization/83338
2319 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
2320 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
2321 vector of unsigned integers to vector of signed integers.
2323 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
2325 PR rtl-optimization/83317
2326 * lra-constraints.c (process_address_1): Add insn code check.
2328 2017-12-08 Michael Matz <matz@suse.de>
2330 Fix PR tree-optimization/83323
2331 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
2332 head-controlled loops and loop BBs.
2333 * common.opt (funroll-and-jam): Remove, instead ...
2334 (floop-unroll-and-jam): ... reuse this option.
2335 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
2336 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
2337 (-floop-unroll-and-jam): ... this option.
2339 2017-12-08 Jakub Jelinek <jakub@redhat.com>
2341 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
2342 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
2343 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
2345 2017-12-08 Richard Biener <rguenther@suse.de>
2348 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
2349 handle accesses outside of zero-sized vars.
2351 2017-12-08 Martin Jambor <mjambor@suse.cz>
2353 PR tree-optimization/83141
2354 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
2355 test for MEM_REFs implicitely changing types with padding. Remove
2357 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
2359 2017-12-08 Jakub Jelinek <jakub@redhat.com>
2361 * config/arc/arc.c (arc_attribute_table): Add exclusions to
2363 * config/avr/avr.c (avr_attribute_table): Likewise.
2364 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
2365 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
2366 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
2367 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
2368 of struct attribute_spec.
2369 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
2371 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
2374 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
2376 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
2379 * profile-count.c (profile_count::from_gcov_type): Move from
2380 profile-count.h; handle overflow.
2381 * profile-count.h (profile_count::from_gcov_type): Move offline.
2383 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
2385 PR rtl-optimization/83304
2386 * combine.c (move_deaths): If we do not know where a register died,
2389 2017-12-08 Richard Biener <rguenther@suse.de>
2391 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
2392 Provide -fopt-info-loop feedback when we interchange in a nest.
2394 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
2396 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
2397 for armv6 ARM CPU IDs.
2399 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
2401 * common/config/arm/arm-common.c: Include <algorithm>.
2402 (INCLUDE_VECTOR): Define.
2403 (compare_opt_names): New function.
2404 (arm_rewrite_selected_arch): Only strip out extensions that can be
2405 expressed through -mfpu. Sort the remaining extensions
2408 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
2410 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
2411 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
2412 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
2413 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
2415 2017-06-08 Tristan Gingold <gindold@adacore.com>
2418 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
2419 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
2421 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2423 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
2424 of struct attribute_spec.
2426 2017-12-08 Julia Koval <julia.koval@intel.com>
2428 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
2429 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
2430 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
2431 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
2432 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
2433 _mm_maskz_dpwssds_epi32): Ditto.
2435 2017-12-08 Richard Biener <rguenther@suse.de>
2437 PR tree-optimization/81303
2438 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
2439 conditions try to create a comparison vector type matching
2440 the data vector type.
2441 (vectorizable_condition): Adjust.
2442 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
2443 Leave invariant conditions alone in case we can vectorize those.
2445 2017-12-08 Julia Koval <julia.koval@intel.com>
2447 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
2448 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
2449 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
2450 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
2451 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
2453 2017-12-08 Julia Koval <julia.koval@intel.com>
2455 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
2456 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
2457 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
2458 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
2459 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
2460 _mm_maskz_dpbusds_epi32): New intrinsics.
2462 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
2464 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
2466 (ld<bh>io_signed): Likewise.
2467 (st<bhw>io): Likewise.
2468 * config/nios2/predicates.md (ldstio_memory_operand): Allow
2469 SMALL_INT12 constant integer operand.
2471 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
2473 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
2475 (nios2_large_constant_allowed): ...this. Adjust uses.
2476 (nios2_plus_symbolic_constant_p): Rename to...
2477 (nios2_plus_large_constant_p): ...this. Adjust uses.
2478 (nios2_legitimate_address_p): Correct CONST_INT handling.
2479 (nios2_symbolic_memory_operand_p): Rename to...
2480 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
2481 (nios2_large_constant_p): Check for large constant integers too.
2482 (nios2_split_large_constant): Handle constant integers.
2483 (nios2_split_symbolic_memory_operand): Rename to...
2484 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
2485 (nios2_legitimize_constant_address): Handle constant integers.
2486 (r0rel_constant_p): Handle small constant integers.
2487 (nios2_print_operand_address): Handle r0-relative integer addresses.
2488 * config/nios2/nios2-protos.h: Adjust for renamed functions.
2489 * config/nios2/nios2.md: Adjust for renamed functions.
2491 2017-12-07 Andrew Waterman <andrew@sifive.com>
2493 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
2494 (TARGET_HAVE_SRODATA_SECTION): New define.
2495 (riscv_select_section): New function.
2497 2017-12-08 Joseph Myers <joseph@codesourcery.com>
2498 Alexander Monakov <amonakov@ispras.ru>
2499 Jakub Jelinek <jakub@redhat.com>
2502 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
2504 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2506 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
2507 a bool success value. Don't call output_operand_lossage here.
2508 (aarch64_print_ldpstp_address): Return a bool success value.
2509 (aarch64_print_operand_address): Call output_addr_const if
2510 aarch64_print_address_internal fails.
2511 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
2512 'y'; call output_operand_lossage instead. Call output_operand_lossage
2513 if aarch64_print_ldpstp_address fails.
2515 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2517 * tree-vector-builder.h
2518 (tree_vector_builder::binary_encoded_nelts): Declare.
2519 * tree-vector-builder.c
2520 (tree_vector_builder::binary_encoded_nelts): New function.
2521 * fold-const.c (negate_expr_p): Likewise.
2522 (operand_equal_p, fold_checksum_tree): Likewise.
2523 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
2524 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
2525 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
2526 (uniform_vector_p): Likewise.
2527 * varasm.c (const_hash_1, compare_constant): Likewise.
2528 * tree-ssa-ccp.c: Include tree-vector-builder.h.
2529 (valid_lattice_transition): Operate directly on the VECTOR_CST
2531 * ipa-icf.c: Include tree-vector-builder.h.
2532 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
2533 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
2535 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2537 * tree.c (build_vector): Delete.
2538 * tree.h (build_vector): Make static and move into the self-testing
2541 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2543 * vector-builder.h (vector_builder::derived): New const overload.
2544 (vector_builder::elt): New function.
2545 * tree-vector-builder.h (tree_vector_builder::type): New function.
2546 (tree_vector_builder::apply_step): Declare.
2547 * tree-vector-builder.c (tree_vector_builder::apply_step): New
2549 * gimple-fold.h (tree_vector_builder): Declare.
2550 (gimple_build_vector): Take a tree_vector_builder instead of a
2551 type and vector of elements.
2552 * gimple-fold.c (gimple_build_vector): Likewise.
2553 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
2555 (get_initial_defs_for_reduction): Likewise.
2556 (vectorizable_induction): Likewise.
2558 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2560 * tree-vector-builder.h
2561 (tree_vector_builder::new_binary_operation): Declare.
2562 * tree-vector-builder.c
2563 (tree_vector_builder::new_binary_operation): New function.
2564 * fold-const.c (fold_relational_const): Use it.
2565 (const_binop): Likewise. Check that both input vectors have
2566 the same number of elements, thus excluding things like WIDEN_SUM.
2567 Check whether it is possible to operate directly on the encodings
2570 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2572 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
2573 new_unary_operation, operating only on the encoded elements.
2574 (const_unop): Likewise.
2575 (exact_inverse): Likewise.
2576 (distributes_over_addition_p): New function.
2577 (const_binop): Use tree_vector_builder and new_unary_operation
2578 for combinations of VECTOR_CST and INTEGER_CST. Operate only
2579 on the encoded elements unless the encoding is strided and the
2580 operation does not distribute over addition.
2581 (fold_convert_const): Use tree_vector_builder and
2582 new_unary_operation. Operate only on the encoded elements
2583 for truncating integer conversions, or for non-stepped encodings.
2585 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2587 * config/sparc/sparc.c: Include tree-vector-builder.h.
2588 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
2589 * expmed.c: Include tree-vector-builder.h.
2590 (make_tree): Use tree_vector_builder instead of build_vector.
2591 * fold-const.c: Include tree-vector-builder.h.
2592 (const_binop): Use tree_vector_builder instead of build_vector.
2593 (const_unop): Likewise.
2594 (native_interpret_vector): Likewise.
2595 (fold_vec_perm): Likewise.
2596 (fold_ternary_loc): Likewise.
2597 * gimple-fold.c: Include tree-vector-builder.h.
2598 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
2600 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
2601 (simplify_vector_constructor): Use tree_vector_builder instead
2603 * tree-vect-generic.c: Include tree-vector-builder.h.
2604 (add_rshift): Use tree_vector_builder instead of build_vector.
2605 (expand_vector_divmod): Likewise.
2606 (optimize_vector_constructor): Likewise.
2607 * tree-vect-loop.c: Include tree-vector-builder.h.
2608 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
2609 of build_vector. Explicitly use a stepped encoding for
2611 * tree-vect-slp.c: Include tree-vector-builder.h.
2612 (vect_get_constant_vectors): Use tree_vector_builder instead
2614 (vect_transform_slp_perm_load): Likewise.
2615 (vect_schedule_slp_instance): Likewise.
2616 * tree-vect-stmts.c: Include tree-vector-builder.h.
2617 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
2618 (vect_gen_perm_mask_any): Likewise.
2619 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
2620 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
2622 (build_vector_from_val): Likewise. Explicitly use a duplicate
2625 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
2627 * doc/generic.texi (VECTOR_CST): Describe new representation of
2629 * vector-builder.h: New file.
2630 * tree-vector-builder.h: Likewise.
2631 * tree-vector-builder.c: Likewise.
2632 * Makefile.in (OBJS): Add tree-vector-builder.o.
2633 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
2634 * tree-core.h (tree_base): Add a vector_cst field to the u union.
2635 (tree_vector): Change the number of elements to
2636 vector_cst_encoded_nelts.
2637 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
2638 (VECTOR_CST_ELTS): Delete.
2639 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
2640 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
2641 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
2642 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
2643 (VECTOR_CST_ENCODED_ELT): Likewise.
2644 (vector_cst_encoded_nelts): New function.
2645 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
2646 VECTOR_CST_NELTS_PER_PATTERN as arguments.
2647 (vector_cst_int_elt, vector_cst_elt): Declare.
2648 * tree.c: Include tree-vector-builder.h.
2649 (tree_code_size): Abort if passed VECTOR_CST.
2650 (tree_size): Update for new VECTOR_CST layout.
2651 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
2652 VECTOR_CST_NELTS_PER_PATTERN as arguments.
2653 (build_vector): Use tree_vector_builder.
2654 (vector_cst_int_elt, vector_cst_elt): New functions.
2655 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
2656 encoded elements and then create the vector in the canonical form.
2657 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
2658 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
2659 (tree_c_tests): Call test_vector_cst_patterns.
2660 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
2662 (hash_tree): Likewise.
2663 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
2664 (streamer_write_tree_header): Likewise.
2665 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
2666 (streamer_alloc_tree): Likewise. Update call to make_vector.
2667 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
2669 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2671 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
2672 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
2673 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
2675 2017-12-07 Bin Cheng <bin.cheng@arm.com>
2676 Richard Biener <rguenther@suse.de>
2678 PR tree-optimization/81303
2679 * Makefile.in (gimple-loop-interchange.o): New object file.
2680 * common.opt (floop-interchange): Reuse the option from graphite.
2681 * doc/invoke.texi (-floop-interchange): Ditto. New document for
2682 -floop-interchange and mention it for -O3.
2683 * opts.c (default_options_table): Enable -floop-interchange at -O3.
2684 * gimple-loop-interchange.cc: New file.
2685 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
2686 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
2687 * passes.def (pass_linterchange): New pass.
2688 * timevar.def (TV_LINTERCHANGE): New time var.
2689 * tree-pass.h (make_pass_linterchange): New declaration.
2690 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
2691 interchange. Record IV before/after increment in new parameters.
2692 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
2693 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
2695 (check_reduction_path): ...New function here.
2696 * tree-vectorizer.h (check_reduction_path): New declaration.
2698 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
2701 PR rtl-optimization/80818
2702 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
2703 always early clobbered.
2704 * lra-lives.c (process_bb_lives): Check input hard regs for early
2705 clobbered non-operand hard reg.
2707 2017-12-07 Jakub Jelinek <jakub@redhat.com>
2710 * tree-cfg.c (verify_gimple_assign_binary): Don't require
2711 types_compatible_p, just that TYPE_MODE is the same.
2713 2017-12-07 Martin Sebor <msebor@redhat.com>
2716 * attribs.c (empty_attribute_table): Initialize new member of
2717 struct attribute_spec.
2718 (decl_attributes): Add argument. Handle mutually exclusive
2719 combinations of attributes.
2720 (selftests::test_attribute_exclusions): New function.
2721 (selftests::attribute_c_tests): Ditto.
2722 * attribs.h (decl_attributes): Add default argument.
2723 * selftest.h (attribute_c_tests): Declare.
2724 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
2725 * tree-core.h (attribute_spec::exclusions, exclude): New type and
2727 * doc/extend.texi (Common Function Attributes): Update const and pure.
2728 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
2729 of struct attribute_spec.
2730 * config/arc/arc.c (arc_attribute_table): Same.
2731 * config/arm/arm.c (arm_attribute_table): Same.
2732 * config/avr/avr.c ( avr_attribute_table): Same.
2733 * config/bfin/bfin.c (bfin_attribute_table): Same.
2734 * config/cr16/cr16.c (cr16_attribute_table): Same.
2735 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
2736 * config/h8300/h8300.c (h8300_attribute_table): Same.
2737 * config/i386/i386.c (ix86_attribute_table): Same.
2738 * config/ia64/ia64.c (ia64_attribute_table): Same.
2739 * config/m32c/m32c.c (m32c_attribute_table): Same.
2740 * config/m32r/m32r.c (m32r_attribute_table): Same.
2741 * config/m68k/m68k.c (m68k_attribute_table): Same.
2742 * config/mcore/mcore.c (mcore_attribute_table): Same.
2743 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
2744 * config/mips/mips.c (mips_attribute_table): Same.
2745 * config/msp430/msp430.c (msp430_attribute_table): Same.
2746 * config/nds32/nds32.c (nds32_attribute_table): Same.
2747 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
2748 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
2749 * config/rl78/rl78.c (rl78__attribute_table): Same.
2750 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
2751 * onfig/rx/rx.c (rx_attribute_table): Same.
2752 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
2753 * config/sh/sh.c (sh_attribute_table): Same.
2754 * config/sparc/sparc.c (sparc_attribute_table): Same.
2755 * config/spu/spu.c (spu_attribute_table): Same.
2756 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
2757 * config/v850/v850.c (v850_attribute_table): Same.
2758 * config/visium/visium.c (visium_attribute_table): Same.
2760 2017-12-07 Tamar Christina <tamar.christina@arm.com>
2763 * config/arm/arm.c (INCLUDE_STRING): Define.
2764 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
2765 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
2768 2017-12-07 Michael Matz <matz@suse.de>
2770 Add unroll and jam pass
2772 * gimple-loop-jam.c: New file.
2773 * Makefile.in (OBJS): Add gimple-loop-jam.o.
2774 * common.opt (funroll-and-jam): New option.
2775 * opts.c (default_options_table): Add unroll-and-jam at -O3.
2776 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
2777 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
2778 * passes.def: Add pass_loop_jam.
2779 * timevar.def (TV_LOOP_JAM): Add.
2780 * tree-pass.h (make_pass_loop_jam): Declare.
2781 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
2782 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
2783 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
2784 to flow_loop_tree_node_add.
2785 (duplicate_subloops, copy_loops_to): Append to sibling list.
2786 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
2787 * doc/invoke.texi (-funroll-and-jam): Document new option.
2788 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
2790 2017-12-07 Richard Biener <rguenther@suse.de>
2792 PR tree-optimization/83296
2793 PR tree-optimization/67769
2794 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
2795 flow sensitive info in an unrelated BB.
2796 (value_replacement): Use reset_flow_sensitive_info.
2797 (minmax_replacement): Reset flow sensitive info on the def
2798 we move. Do not reset flow sensitive info in the whole BB
2799 we move the stmt to.
2800 (abs_replacement): Likewise.
2802 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
2805 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
2806 rs6000_cpu to the given -mcpu=, or to the default processor.
2808 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
2810 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
2811 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
2812 instead of rs6000_cpu_attr.
2813 (rs6000_adjust_cost): Ditto.
2814 (is_microcoded_insn): Ditto.
2815 (rs6000_adjust_priority): Ditto.
2816 (rs6000_issue_rate): Ditto.
2817 (rs6000_use_sched_lookahead): Ditto.
2818 (rs6000_use_sched_lookahead_guard): Ditto.
2819 (rs6000_sched_reorder): Ditto.
2820 (force_new_group): Ditto.
2821 * config/rs6000/rs6000.md (cpu attribute): Ditto.
2822 (group_ending_nop): Ditto.
2824 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
2826 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
2827 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
2828 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
2829 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
2831 (rs6000_reassociation_width): Ditto.
2832 (rs6000_emit_epilogue): Ditto.
2833 (rs6000_adjust_cost): Ditto.
2834 (is_microcoded_insn): Ditto.
2835 (is_cracked_insn): Ditto.
2836 (rs6000_adjust_priority): Ditto.
2837 (rs6000_sched_reorder): Ditto.
2838 (rs6000_sched_reorder2): Ditto.
2839 (insn_must_be_first_in_group): Ditto.
2840 (insn_must_be_last_in_group): Ditto.
2841 (rs6000_register_move_cost): Ditto.
2842 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
2845 2017-12-07 Julia Koval <julia.koval@intel.com>
2847 * config.gcc: Add vaesintrin.h.
2848 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
2850 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
2851 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
2853 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
2854 * config/i386/immintrin.h: Include vaesintrin.h.
2855 * config/i386/sse.md (vaesdec_<mode>): New pattern.
2856 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
2857 _mm_aesdec_epi128): New intrinsics.
2859 2017-12-06 David Malcolm <dmalcolm@redhat.com>
2861 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
2862 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
2863 c-family/c-spellcheck.cc.
2864 (best_macro_match::best_macro_match): Likewise.
2866 (struct edit_distance_traits<cpp_hashnode *>): Move to
2867 c-family/c-spellcheck.h.
2868 (class best_macro_match): Likewise.
2870 2017-12-06 Jakub Jelinek <jakub@redhat.com>
2872 PR tree-optimization/83293
2873 * gimple-ssa-strength-reduction.c (insert_initializers): Use
2874 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
2875 might insert into empty bb.
2878 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
2879 simplify for plus with :c added, and pointer_plus without that.
2880 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
2881 with undefined overflow and the conversion is not widening,
2882 perform negation in utype and only convert to type afterwards.
2883 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
2884 simplify for plus with :c added, and pointer_plus without that.
2885 If type is integral with undefined overflow and the conversion is
2886 not widening, perform minus in utype and only convert to type
2887 afterwards. Move the last pointer_diff_expr simplify into the
2890 2017-12-06 Martin Sebor <msebor@redhat.com>
2892 PR tree-optimization/82646
2893 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
2894 strncpy, not maxlen.
2896 2017-12-06 Martin Sebor <msebor@redhat.com>
2898 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
2901 PR tree-optimization/83075
2902 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
2903 strncat/strncpy don't change length of source string.
2905 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
2908 2017-11-29 Martin Aberg <maberg@gaisler.com>
2910 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
2911 to prevent b2bst errata sequence.
2912 (sqrtdf2_fix): Likewise.
2914 2017-12-06 Jakub Jelinek <jakub@redhat.com>
2916 PR tree-optimization/81945
2917 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
2918 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
2919 to dest_cfun has orig_loop_num set, either remap it to the new
2920 loop number if the loop got moved too, or clear it.
2922 2017-12-05 Steve Ellcey <sellcey@cavium.com>
2924 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
2926 (thunderx2t99_nothing): New insn reservation.
2927 (thunderx2t99_mrs): New insn reservation.
2928 (thunderx2t99_multiple): New insn reservation.
2929 (thunderx2t99_alu_basi): Add bfx to reservation.
2930 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
2932 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
2935 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
2937 2017-12-05 Bin Cheng <bin.cheng@arm.com>
2939 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
2940 tree-ssa-pre.c::remove_dead_inserted_code.
2941 * tree-ssa-dce.h: New file.
2942 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
2943 (remove_dead_inserted_code): Move and rename to function
2944 tree-ssa-dce.c::simple_dce_from_worklist.
2945 (pass_pre::execute): Update use.
2947 2017-12-05 Richard Biener <rguenther@suse.de>
2949 PR tree-optimization/83277
2950 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
2951 to code-gen liveout vars.
2953 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
2955 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
2956 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
2957 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
2960 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
2962 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
2964 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
2965 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
2968 2017-12-05 Richard Biener <rguenther@suse.de>
2970 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
2971 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
2972 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
2973 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
2975 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2977 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
2979 2017-12-05 Martin Liska <mliska@suse.cz>
2980 Jakub Jelinek <jakub@redhat.com>
2982 * doc/invoke.texi: Document the options.
2983 * flag-types.h (enum sanitize_code): Add
2984 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
2985 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
2986 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
2987 * opts.c: Define new sanitizer options.
2988 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
2989 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
2991 2017-12-05 Julia Koval <julia.koval@intel.com>
2993 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
2994 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
2995 (ix86_handle_option): Handle -mavx512vnni.
2996 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
2997 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
2998 * config/i386/i386-c (__AVX512VNNI__): New.
2999 * config/i386/i386.c (ix86_target_string): Handle new option.
3000 (ix86_valid_target_attribute_inner_p): Handle new option.
3001 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
3002 * config/i386/i386.opt (mavx512vnni): New option.
3004 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
3007 * athlon.md: Disable for generic.
3008 * haswell.md: Enable for generic.
3009 * i386.c (ix86_sched_init_global): Add core hooks for generic.
3010 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
3012 (ix86_adjust_cost): Move generic to haswell path.
3014 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
3016 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
3017 instead of MEM_P in a couple more places. Fix formatting issues.
3019 2017-12-04 Jim Wilson <jimw@sifive.com>
3021 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
3022 instead of GP_REG_LAST-1.
3023 (riscv_adjust_libcall_cfi_prologue): Likewise.
3024 (riscv_adjust_libcall_cri_epilogue): Likewise.
3025 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
3028 2017-12-04 Luis Machado <luis.machado@linaro.org>
3030 * ipa-pure-const.c (check_decl): Add missing newline.
3031 (state_from_flags): Likewise.
3033 2017-12-04 Jeff Law <law@redhat.com>
3035 PR tree-optimizatin/78496
3036 * gimple-ssa-evrp-analyze.h
3037 (evrp_range_analyzer::get_vr_values): Simplify.
3038 * gimple-ssa-evrp-analyze.c: Corresponding changes.
3039 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
3040 and gimple-ssa-evrp-analyze.h.
3041 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
3042 (simplify_stmt_for_jump_threading): Copy a blob of code from
3043 tree-vrp.c to use ranges to simplify statements.
3044 (dom_opt_dom_walker::before_dom_children): Call
3045 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
3046 (dom_opt_dom_walker::after_dom_children): Similarly for
3047 evrp_range_analyzer::leave.
3048 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
3051 * gimple-ssa-evrp-analyze.c
3052 (evrp_range_analyzer::extract_range_from_stmt): Always use
3053 vr_values::update_value_range so preexisting range info is
3054 medged with new range info, even if the new range is VR_VARYING.
3056 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
3058 * combine.c: Adjust comment.
3059 (use_crosses_set_p): Delete.
3060 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
3061 (try_combine): Ditto.
3063 2017-12-04 Richard Biener <rguenther@suse.de>
3065 PR tree-optimization/83255
3066 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
3067 Re-add zero-iteration check.
3069 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
3071 PR rtl-optimization/83245
3072 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
3073 hard registers as earlyclobber, also if not in an asm.
3075 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
3079 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3082 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
3083 rs6000_cpu based on cpu_index, not tune_index.
3085 2017-12-04 Richard Biener <rguenther@suse.de>
3087 PR tree-optimization/83238
3088 * graphite-scop-detection.c (scop_detection::merge_sese): Make
3089 code match comment, rejecting invalid SESE regions.
3091 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
3093 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
3094 require base operand is a REG_POINTER prior to reload on targets
3095 with non-equivalent space registers.
3097 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
3099 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
3100 (update_specialized_profile): Fix updating of counts.
3101 (perhaps_add_new_callers): Likewise.
3103 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
3106 * x86-tune.def: Remove obsolette FIXMEs.
3107 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
3108 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
3109 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
3111 (X86_TUNE_PAD_RETURNS): Disable for generic.
3113 2017-12-02 Jakub Jelinek <jakub@redhat.com>
3115 PR tree-optimization/83170
3116 PR tree-optimization/83241
3117 * gimple-ssa-store-merging.c
3118 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
3119 gimple_vuse (ins_stmt) in case it has changed.
3120 (imm_store_chain_info::output_merged_store): Likewise.
3122 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
3126 * tree-cfg.c (pass_warn_function_return::execute): Handle
3127 __builtin_ubsan_handle_missing_return like __builtin_unreachable
3128 with BUILTINS_LOCATION.
3132 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
3133 is BLKmode for vector field with vector raw mode, use TYPE_MODE
3134 instead of DECL_MODE.
3136 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
3137 last argument to rtx pointer.
3138 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
3139 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
3140 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
3141 depending on the chosen ISAs.
3142 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
3143 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
3144 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
3146 * config/i386/sse.md (mov<mode>_internal): Likewise.
3147 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
3149 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3151 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
3153 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3155 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
3156 parameters from prototype.
3157 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
3158 parameters. Don't print a comment.
3159 (emit_fusion_gpr_load): Adjust.
3160 (emit_fusion_load_store): Adjust.
3161 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
3162 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
3163 comment on the second line.
3165 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3168 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
3169 rs6000_cpu based on cpu_index, not tune_index.
3171 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3173 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
3174 which_alternative instead of which_alternative + 1.
3175 (output_asm_insn): Print an extra tab if the template is short.
3177 2017-12-01 Jim Wilson <jimw@sifive.com>
3179 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
3181 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
3182 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
3184 * doc/tm.texi: Regenerate.
3186 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
3189 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
3190 whether we can allocate pseudos before trying to fix an address.
3191 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
3192 memory address is indexed or indirect.
3193 (floatuns_<mode>si2_hw2): Likewise.
3195 2017-12-01 Jason Merrill <jason@redhat.com>
3197 * Makefile.in (TAGS): Add c-family/*.cc.
3199 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
3201 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
3202 (call_value_insn): Likewise.
3203 (sibcall_insn): Likewise.
3204 (sibcall_value_insn): Likewise.
3205 (movsi_aarch64): Likewise.
3206 (movdi_aarch64): Likewise.
3207 (add_losym_): Likewise.
3208 (ldr_got_small_): Likewise.
3209 (ldr_got_small_sidi): Likewise.
3210 (ldr_got_small_28k_): Likewise.
3211 (ldr_got_small_28k_sidi): Likewise.
3212 * config/aarch64/aarch64.c (aarch64_print_address_internal):
3213 Move output_addr_const to symbolic case. Add error check.
3215 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3217 * config/s390/predicates.md (plus16_Q_operand): New predicate.
3218 * config/s390/s390.md: Disable MVC merging peephole if it would
3219 disable operand forwarding.
3220 (new peephole2): Split MVCs if it would turn them into up to 2
3223 2017-12-01 Richard Biener <rguenther@suse.de>
3225 PR tree-optimization/83232
3226 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
3227 detection of same access. Instead of breaking the group here
3228 do not consider the duplicate. Add comment explaining real fix.
3230 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
3232 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
3234 2017-12-01 Sudakshina Das <sudi.das@arm.com>
3236 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
3238 2017-12-01 Jakub Jelinek <jakub@redhat.com>
3240 * function.h (struct function): Remove cilk_frame_decl,
3241 is_cilk_function and calls_cilk_spawn fields.
3242 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
3244 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
3245 cilk_elemental field.
3246 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
3247 * target.def: Adjust comment.
3248 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
3249 Don't test cilk_elemental.
3251 PR tree-optimization/83233
3252 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
3253 bswap_stat name for the struct.
3256 * tree.h (SWITCH_BREAK_LABEL_P): Define.
3257 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
3258 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
3259 SWITCH_BREAK_LABEL_P set on the label.
3260 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
3261 added for default case if it was missing and not all cases covered.
3262 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
3263 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
3265 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
3266 to avoid -Wimplicit-fallthrough warning.
3267 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
3268 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
3270 PR tree-optimization/83221
3271 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
3273 (init_reassoc): Formatting fix.
3276 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
3277 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
3281 * tree-cfg.c: Include asan.h.
3282 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
3284 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
3286 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
3288 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
3290 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
3291 vpcmpeqd instruction.
3293 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
3295 * config/i386/i386.c (standard_sse_constant_opcode): Fix
3296 registers type for 128bit mode.
3298 2017-11-30 Jakub Jelinek <jakub@redhat.com>
3300 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
3302 * gengtype-state.c (read_state_pair): Likewise.
3303 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
3304 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
3305 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
3306 * ifcvt.c (noce_try_store_flag_constants): Likewise.
3307 * tree-ssa-ccp.c (ccp_finalize): Likewise.
3308 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
3309 * builtins.c (fold_builtin_3): Likewise.
3310 * graphite-scop-detection.c
3311 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
3312 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
3315 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
3318 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
3319 parameter. Return true if ignore_decl_rtl_set_p is true and other
3320 conditions are satisfied.
3321 * asan.h (asan_protect_global): Add new parameter.
3322 * varasm.c (categorize_decl_for_section): Pass true as second parameter
3323 to asan_protect_global calls.
3325 2017-11-30 Jim Wilson <jimw@sifive.com>
3327 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
3328 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
3329 -mstrict-align, add info on default value. Delete redundant lines for
3330 -mabi. Add missing -mexplicit-relocs docs.
3332 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
3334 * config/arc/arc.md (trap): New pattern.
3336 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
3338 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
3339 instruction to end into a delay slot.
3340 * config/arc/arc.md (cond_delay_insn): Check if the instruction
3341 can be placed into a delay slot against reg_note.
3343 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
3345 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
3346 labels number of usages.
3348 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
3350 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
3352 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
3354 2017-11-30 Jakub Jelinek <jakub@redhat.com>
3357 * internal-fn.c (expand_mul_overflow): Optimize unsigned
3358 multiplication by power of 2 constant into two shifts + comparison.
3360 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
3363 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
3365 2017-11-30 Richard Biener <rguenther@suse.de>
3367 PR tree-optimization/83202
3368 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
3369 allow_peel argument and guard peeling.
3370 (canonicalize_loop_induction_variables): Likewise.
3371 (canonicalize_induction_variables): Pass false.
3372 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
3373 peeling from cunrolli.
3375 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
3377 * combine.c (try_combine): Print a message to dump file whenever
3378 I0, I1, or I2 cannot be combined into I3.
3380 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
3382 PR rtl-optimization/83156
3383 PR rtl-optimization/82621
3384 * combine.c (try_combine): Don't split an I2 if one of the dests is
3385 set again before I3. Allow unused dests.
3387 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
3389 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
3391 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
3393 PR rtl-optimization/80818
3394 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
3395 recursively. Use insn code for clobber.
3396 (lra_set_insn_recog_data): Pass the new arg to
3397 collect_non_operand_hard_regs.
3398 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
3400 (lra_update_insn_regno_info): Pass insn to
3401 add_regs_to_insn_regno_info.
3403 2017-11-29 Jim Wilson <jimw@sifive.com>
3404 Andrew Waterman <andrew@sifive.com>
3406 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
3407 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
3408 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
3409 use SINGLE_SHIFT_COST.
3410 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
3411 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
3413 2017-11-29 Julia Koval <julia.koval@intel.com>
3415 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
3416 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
3417 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
3418 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
3419 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
3420 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
3421 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
3422 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
3423 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
3424 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
3425 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
3426 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
3427 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
3428 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
3429 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
3430 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
3431 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
3432 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
3433 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
3434 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
3435 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
3436 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
3437 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
3438 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
3439 __builtin_ia32_vpshldv_v2di_mask,
3440 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
3441 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
3442 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
3444 2017-11-29 Julia Koval <julia.koval@intel.com>
3446 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
3447 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
3448 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
3449 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
3450 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
3451 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
3452 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
3453 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
3454 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
3455 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
3456 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
3457 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
3458 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
3459 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
3460 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
3461 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
3462 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
3463 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
3464 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
3465 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
3466 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
3468 2017-11-29 Daniel Cederman <cederman@gaisler.com>
3470 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
3471 movsi_pic_gotdata_op instruction as a load for the UT699 errata
3474 2017-11-29 Martin Aberg <maberg@gaisler.com>
3476 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
3477 to prevent b2bst errata sequence.
3478 (sqrtdf2_fix): Likewise.
3480 2017-11-29 Daniel Cederman <cederman@gaisler.com>
3482 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
3483 (div_sqrt_insn_p): New function.
3484 (sparc_do_work_around_errata): Insert NOP instructions to
3485 prevent sequences that could trigger the TN-0013 errata for
3486 certain LEON3 processors.
3487 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
3488 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
3489 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
3490 (in_branch_delay): Prevent div and sqrt in delay slot if
3492 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
3494 2017-11-29 Daniel Cederman <cederman@gaisler.com>
3496 * config/sparc/sparc.c (atomic_insn_p): New function.
3497 (sparc_do_work_around_errata): Insert NOP instructions to
3498 prevent sequences that could trigger the TN-0010 errata for
3500 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
3501 instruction referable in atomic_insns_p.
3503 2017-11-29 Daniel Cederman <cederman@gaisler.com>
3505 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
3506 (atomic_compare_and_swap_leon3_1): Likewise.
3509 2017-11-29 Daniel Cederman <cederman@gaisler.com>
3511 * config/sparc/sparc.c (fpop_insn_p): New function.
3512 (sparc_do_work_around_errata): Insert NOP instructions to
3513 prevent sequences that could trigger the TN-0012 errata for
3515 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
3516 * config/sparc/sparc.md (fix_gr712rc): New attribute.
3517 (in_branch_annul_delay): Prevent floating-point instructions
3518 in delay slot of annulled integer branch.
3520 2017-11-29 Richard Biener <rguenther@suse.de>
3522 PR tree-optimization/83202
3523 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
3525 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
3526 nodes vectorized to the same stmts multiple times.
3527 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
3528 (vect_analyze_slp_instance): Adjust.
3529 (scalar_stmts_to_slp_tree_map_t): New typedef.
3530 (vect_schedule_slp_instance): Add a map recording the SLP node
3531 representing the vectorized stmts for a set of scalar stmts.
3532 Avoid code-generating redundancies.
3533 (vect_schedule_slp): Allocate map and pass it down.
3535 2017-11-29 Nathan Sidwell <nathan@acm.org>
3538 * tree.c (build_complex_type): Fix canonicalization. Only fill in
3541 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
3543 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
3544 cases for printing LDP/STP memory addresses.
3545 (aarch64_print_address_internal): Renamed from
3546 aarch64_print_operand_address, added parameter, add Pmode check.
3547 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
3548 (aarch64_print_operand_address): Indirect to
3549 aarch64_print_address_internal.
3550 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
3551 'y' operand output specifier.
3553 2017-11-29 Jakub Jelinek <jakub@redhat.com>
3556 * tree.c (build_simple_mem_ref_loc): Handle
3557 get_addr_base_and_unit_offset returning a MEM_REF.
3560 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
3563 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
3566 2017-11-28 David Malcolm <dmalcolm@redhat.com>
3568 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
3570 (layout::move_to_column): Likewise.
3572 2017-11-28 Jakub Jelinek <jakub@redhat.com>
3575 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
3576 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
3578 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3579 Martin Jambor <mjambor@suse.cz>
3582 * tree.h (expr_type_first_operand_type_p): Declare
3583 * tree.c (expr_type_first_operand_type_p): New function.
3584 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
3585 (ipa_value_from_jfunc): Adjust declaration.
3586 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
3587 Use it as result type for arithmetics, unless it is NULL in which case
3588 be more conservative.
3589 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
3590 ipa_get_jf_pass_through_result.
3591 (propagate_vals_across_pass_through): Likewise.
3592 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
3593 is to propagate_vals_across_pass_through.
3594 (propagate_constants_across_call): Pass PARM_TYPE to
3595 propagate_scalar_across_jump_function.
3596 (find_more_scalar_values_for_callers_subset): Pass parameter type to
3597 ipa_value_from_jfunc.
3598 (cgraph_edge_brings_all_scalars_for_node): Likewise.
3599 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
3600 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
3601 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
3602 target_type, pass it to ipa_value_from_jfunc.
3603 (update_indirect_edges_after_inlining): Pass parameter type to
3604 try_make_edge_direct_simple_call.
3606 2017-11-28 Jeff Law <law@redhat.com>
3608 * gimple-ssa-evrp-analyze.c
3609 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
3610 refine ranges if scev_initialized_p returns true.
3611 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
3613 2017-11-28 Julia Koval <julia.koval@intel.com>
3615 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
3616 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
3617 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
3618 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
3619 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
3620 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
3621 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
3622 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
3623 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
3624 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
3625 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
3626 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
3627 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
3628 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
3629 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
3630 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
3631 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
3632 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
3633 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
3634 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
3635 __builtin_ia32_vpshrd_v2di_mask): New builtins.
3636 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
3638 2017-11-28 Julia Koval <julia.koval@intel.com>
3640 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
3641 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
3642 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
3643 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
3644 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
3645 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
3646 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
3647 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
3648 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
3649 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
3650 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
3651 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
3652 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
3653 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
3654 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
3655 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
3656 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
3657 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
3658 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
3659 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
3660 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
3661 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
3662 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
3663 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
3664 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
3665 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
3666 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
3667 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
3668 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
3669 __builtin_ia32_vpshld_v2di_mask): New builtins.
3670 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
3671 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
3673 2017-11-28 Richard Biener <rguenther@suse.de>
3675 PR tree-optimization/80776
3676 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
3678 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
3680 (evrp_range_analyzer::record_ranges_from_incoming_edges):
3681 If the incoming edge is an effective fallthru because the other
3682 edge only reaches a __builtin_unreachable () then record ranges
3683 derived from the controlling condition in SSA info.
3684 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
3685 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
3687 2017-11-28 Olivier Hainque <hainque@adacore.com>
3689 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
3690 on mingw build hosts.
3692 2017-11-28 Jakub Jelinek <jakub@redhat.com>
3695 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
3697 2017-11-28 Richard Biener <rguenther@suse.de>
3700 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
3701 copies generated from memcpy use a character array as reference
3704 2017-11-28 Julia Koval <julia.koval@intel.com>
3705 Sebastian Peryt <sebastian.peryt@intel.com>
3707 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
3708 c-family/c-cilkplus.o, c-family/array-notation-common.o,
3709 cilk-common.o, cilk.h, cilk-common.c): Remove.
3711 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
3712 * builtins.c (is_builtin_name): Remove cilkplus condition.
3713 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
3714 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
3715 cilk-builtins.def, cilkplus.def): Remove.
3716 * cif-code.def (CILK_SPAWN): Remove.
3717 * cilk-builtins.def: Delete.
3718 * cilk-common.c: Ditto.
3720 * cilkplus.def: Ditto.
3721 * config/darwin.h (fcilkplus): Delete.
3722 * cppbuiltin.c: Ditto.
3723 * doc/extend.texi: Remove cilkplus doc.
3724 * doc/generic.texi: Ditto.
3725 * doc/invoke.texi: Ditto.
3726 * doc/passes.texi: Ditto.
3727 * gcc.c (fcilkplus): Remove.
3728 * gengtype.c (cilk.h): Remove.
3729 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
3731 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
3733 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
3735 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
3736 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
3737 cilkplus conditions.
3738 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
3739 inline_read_section): Ditto.
3740 * ipa-inline-analysis.c (cilk.h): Remove.
3741 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
3742 * lto-wrapper.c (merge_and_complain, append_compiler_options,
3743 append_linker_options): Remove condition for fcilkplus.
3744 * lto/lto-lang.c (cilk.h): Remove.
3745 (lto_init): Remove condition for fcilkplus.
3746 * omp-expand.c (expand_cilk_for_call): Delete.
3747 (expand_omp_taskreg, expand_omp_for_static_chunk,
3748 expand_omp_for): Remove cilkplus
3750 (expand_cilk_for): Delete.
3751 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
3752 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
3753 execute_lower_omp, diagnose_sb_0): Ditto.
3754 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
3755 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
3756 * tree-nested.c: Ditto.
3757 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
3758 (dump_generic_node): Ditto.
3759 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
3760 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
3762 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
3764 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
3766 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
3767 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
3768 for vector mode and !TARGET_SIMD.
3770 2017-11-28 Jakub Jelinek <jakub@redhat.com>
3772 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
3774 * tree.h (SWITCH_LABELS): Remove.
3775 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
3776 assert SWITCH_BODY is non-NULL.
3777 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
3779 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
3781 PR tree-optimization/80788
3782 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
3783 has TREE_OVERFLOW set, call drop_tree_overflow.
3785 2017-11-28 Richard Biener <rguenther@suse.de>
3787 PR tree-optimization/83158
3788 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
3790 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
3793 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
3794 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
3796 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
3798 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
3799 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
3801 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
3804 * varasm.c (compare_constant): Take the mode of the constants into
3805 account when comparing floating point constants.
3807 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
3809 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
3810 from explicit instantiation of debug_helper.
3811 * vec.h (DEFINE_DEBUG_VEC): Ditto.
3813 2017-11-27 Richard Biener <rguenther@suse.de>
3815 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
3818 2017-11-27 Richard Biener <rguenther@suse.de>
3820 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
3821 (type_hash_canon): Also clear int_cst_hash_table entry for
3823 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
3825 2017-11-27 Tamar Christina <tamar.christina@arm.com>
3827 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
3828 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
3829 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
3831 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
3833 * hash-map.h (gt_cleare_cache): Avoid UB.
3835 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
3837 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
3838 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
3839 (decide_unroll_runtime_iterations): Likewise.
3840 (decide_unroll_stupid): Likewise.
3842 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3845 * config/i386/i386.md: Add a loop with incssp.
3847 2017-11-27 Martin Jambor <mjambor@suse.cz>
3849 PR tree-optimization/81248
3850 * tree-sra.c (splice_param_accesses): Remove size check.
3851 (decide_one_param_reduction): Fix size check.
3852 * gimple-pretty-print.c (dump_profile): Silence warning.
3853 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
3855 2017-11-27 Jakub Jelinek <jakub@redhat.com>
3858 * dbxout.c (lastlineno): New variable.
3859 (dbx_debug_hooks): Use dbxout_switch_text_section as
3860 switch_text_section debug hook.
3861 (dbxout_function_end): Switch to current_function_section
3862 rather than function_section. If crtl->has_bb_partition,
3863 output just one N_FUN, depending on in_cold_section_p.
3864 (dbxout_source_line): Remember last lineno in lastlineno.
3865 (dbxout_switch_text_section): New function.
3866 (dbxout_function_decl): Adjust dbxout_block caller.
3867 (dbx_block_with_cold_children): New function.
3868 (dbxout_block): Return true if any LBRAC/RBRAC have been
3869 emitted. Use dbx_block_with_cold_children at depth == 0
3870 in second partition. Add PARENT_BLOCKNUM argument, pass
3871 it optionally adjusted to children. Output LBRAC/RBRAC
3872 around recursive call only if the block is in the current
3873 partition, if not and anything was output, emit empty
3875 * final.c (final_scan_insn): Compute cold_function_name
3876 before calling switch_text_section debug hook. Call
3877 that hook even if dwarf2out_do_frame if not emitting
3881 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
3882 TREE_READONLY decls.
3884 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
3886 PR rtl-optimization/82488
3887 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
3889 2017-11-26 Julia Koval <julia.koval@intel.com>
3891 * config/i386/i386.c (processor_target_table): Add skylake_cost for
3893 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
3896 2017-11-26 Julia Koval <julia.koval@intel.com>
3898 * config/i386/driver-i386.c (host_detect_local_cpu):
3899 Detect skylake-avx512.
3901 2017-11-26 Julia Koval <julia.koval@intel.com>
3903 * config.gcc: Add -march=cannonlake.
3904 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
3905 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
3906 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
3907 (PTA_CANNONLAKE): New.
3908 (processor_target_table): Add cannonlake.
3909 (ix86_option_override_internal): Ditto.
3910 (fold_builtin_cpu): Ditto.
3911 (get_builtin_code_for_version): Handle cannonlake.
3912 (M_INTEL_COREI7_CANNONLAKE): New.
3913 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
3914 * doc/invoke.texi: Add -march=cannonlake.
3916 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
3918 * plugin.c (add_new_plugin): Use platform-specific library extensions.
3919 (try_init_one_plugin): Alternative implementation for MinGW.
3920 * Makefile.in (plugin_implib): New.
3921 (gengtype-lex.c): Fix broken AIX workaround.
3922 * configure: Regenerate.
3923 * doc/plugins.texi: Document support for MinGW.
3925 2017-11-25 Jakub Jelinek <jakub@redhat.com>
3927 PR rtl-optimization/81553
3928 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
3929 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
3930 is a shift where C1 has different mode than the whole shift, use C1's
3931 mode for MULT rather than the shift's mode.
3934 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
3935 builtins not enabled in the currently selected ISA.
3937 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
3939 PR tree-optimization/71026
3940 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
3941 (insert_reciprocals): Change to insert reciprocals before a division
3942 by a square and to insert the square of a reciprocal.
3943 (execute_cse_reciprocals_1): Change to consider division by a square.
3944 (register_division_in): Add importance parameter.
3946 2017-11-24 Richard Biener <rguenther@suse.de>
3948 PR tree-optimization/82402
3949 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
3950 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
3952 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
3954 * match.pd (0-ptr): New transformation.
3956 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
3959 * ipa-inline.c (inline_small_functions): Set current badnes correctly
3960 when skipping checking.
3962 2017-11-24 Richard Biener <rguenther@suse.de>
3964 PR tree-optimization/83128
3965 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
3966 (vn_reference_lookup_3): Likewise.
3968 2017-11-24 Jakub Jelinek <jakub@redhat.com>
3971 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
3972 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
3975 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
3976 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
3977 check for the latter with gimple_call_builtin_p. Do not handle
3978 BUILT_IN_STPNCPY_CHK which is not a pass through call.
3980 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
3982 * config/arm/arm_neon.h: Fix pragma GCC push_options before
3985 2017-11-23 Julia Koval <julia.koval@intel.com>
3987 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
3988 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
3989 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
3990 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
3991 _mm512_maskz_expandloadu_epi16): New intrinsics.
3992 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
3993 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
3994 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
3995 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
3996 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
3997 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
3998 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
3999 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
4000 _mm256_maskz_expandloadu_epi8): New intrinsics.
4001 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
4002 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
4003 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
4004 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
4005 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
4006 * config/i386/sse.md (VI248_VLBW): New iterator.
4007 (expand<mode>_mask, expand<mode>_maskz): New patterns.
4009 2017-11-23 Julia Koval <julia.koval@intel.com>
4011 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
4012 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
4013 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
4014 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
4015 _mm512_mask_compressstoreu_epi16): New.
4016 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
4017 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
4018 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
4019 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
4020 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
4021 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
4022 _mm256_mask_compressstoreu_epi8): New.
4023 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
4024 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
4025 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
4026 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
4027 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
4028 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
4029 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
4030 __builtin_ia32_compresshi128_mask,
4031 __builtin_ia32_compressstoreuqi512_mask,
4032 __builtin_ia32_compressstoreuhi512_mask,
4033 __builtin_ia32_compressstoreuqi256_mask,
4034 __builtin_ia32_compressstoreuqi128_mask,
4035 __builtin_ia32_compressstoreuhi256_mask,
4036 __builtin_ia32_compressstoreuhi128_mask): New builtins.
4037 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
4039 (ix86_expand_special_args_builtin): Handle new types.
4040 (s4fma_expand): Handle new builtin array.
4041 * config/i386/immintrin.h: Include new headers.
4042 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
4043 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
4045 2017-11-23 Jakub Jelinek <jakub@redhat.com>
4047 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
4048 fixes. Declare temp and g variables at the top in order to avoid
4049 {} in most of the cases.
4051 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
4053 * match.pd (ptr-0): New transformation.
4055 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
4057 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
4058 (struct addr_mode_cost_table): New.
4059 (struct tune_params): Add field addr_mode_costs.
4060 * config/arm/arm.c (generic_addr_mode_costs): New.
4061 (arm_slowmul_tune): Initialise addr_mode_costs field.
4062 (arm_fastmul_tune): Likewise.
4063 (arm_strongarm_tune): Likewise.
4064 (arm_xscale_tune): Likewise.
4065 (arm_9e_tune): Likewise.
4066 (arm_marvell_pj4_tune): Likewise.
4067 (arm_v6t2_tune): Likewise.
4068 (arm_cortex_tune): Likewise.
4069 (arm_cortex_a8_tune): Likewise.
4070 (arm_cortex_a7_tune): Likewise.
4071 (arm_cortex_a15_tune): Likewise.
4072 (arm_cortex_a35_tune): Likewise.
4073 (arm_cortex_a53_tune): Likewise.
4074 (arm_cortex_a57_tune): Likewise.
4075 (arm_exynosm1_tune): Likewise.
4076 (arm_xgene1_tune): Likewise.
4077 (arm_cortex_a5_tune): Likewise.
4078 (arm_cortex_a9_tune): Likewise.
4079 (arm_cortex_a12_tune): Likewise.
4080 (arm_cortex_a73_tune): Likewise.
4081 (arm_v7m_tune): Likewise.
4082 (arm_cortex_m7_tune): Likewise.
4083 (arm_v6m_tune): Likewise.
4084 (arm_fa726te_tune): Likewise.
4085 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
4087 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
4089 * config/arm/arm.c (arm_mem_costs): New function.
4090 (arm_rtx_costs_internal): Use arm_mem_costs.
4092 2017-11-23 Mark Wielaard <mark@klomp.org>
4094 * dwarf2out.c (init_sections_and_labels): Use generation to create
4095 unique ranges_section_label and ranges_base_label. Return generation.
4096 (output_rnglists): Add generation argument. Use generation to create
4097 unique ranges labels.
4098 (dwarf2out_finish): Get generation from init_sections_and_labels
4099 and pass generation to output_rnglists.
4101 2017-11-23 Mike Stump <mikestump@comcast.net>
4102 Eric Botcazou <ebotcazou@adacore.com>
4104 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
4105 * cfgloop.h (struct loop): Add unroll field.
4106 * function.h (struct function): Add has_unroll bitfield.
4107 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
4108 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
4109 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
4110 (pass_rtl_unroll_loops::gate): Likewise.
4111 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
4112 for which loop->unroll==1.
4113 (decide_unroll_constant_iterations): Use note for consistency and
4114 take loop->unroll into account. Return early if loop->unroll is set.
4115 Fix thinko in existing test.
4116 (decide_unroll_runtime_iterations): Use note for consistency and
4117 take loop->unroll into account.
4118 (decide_unroll_stupid): Likewise.
4119 * lto-streamer-in.c (input_cfg): Read loop->unroll.
4120 * lto-streamer-out.c (output_cfg): Write loop->unroll.
4121 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
4123 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
4124 (print_loop): Print loop->unroll if set.
4125 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
4126 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
4127 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
4129 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
4130 loop->unroll is set and smaller than the trip count. Otherwise bypass
4131 entirely the heuristics if loop->unroll is set. Remove dead note.
4132 Fix off-by-one bug in other note.
4133 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
4134 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
4136 (tree_unroll_loops_completely): Make static.
4137 (pass_complete_unroll::execute): Use correct type for variable.
4138 (pass_complete_unrolli::execute): Fix formatting.
4139 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
4141 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
4143 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
4144 enable when TARGET_PREFER_AVX128 is set.
4146 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
4148 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
4150 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
4151 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
4153 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
4155 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
4156 frequencies. Fix estimation of aggregate parameters.
4158 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
4160 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
4163 2017-11-23 Tom de Vries <tom@codesourcery.com>
4165 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
4167 2017-11-23 Tom de Vries <tom@codesourcery.com>
4169 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
4172 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
4175 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
4176 sibcall_value_pcrel_fdpic): Use local variable instead of
4178 (calli_tbr_rel): Add missing operand 2.
4179 (call_valuei_tbr_rel): Add missing operand 3.
4181 2017-11-23 Jakub Jelinek <jakub@redhat.com>
4184 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
4185 bitpos/bitsize covering the whole destination, use store_expr only if
4186 the complex mode is the same. Otherwise, use expand_normal and if
4187 it returns CONCAT, subreg each part separately instead of trying to
4188 subreg the whole result.
4190 2017-11-23 Richard Biener <rguenther@suse.de>
4192 PR tree-optimization/23094
4193 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
4194 come here from walking over backedges in the first iteration.
4195 (vn_reference_lookup_3): Skip clobbers that store the same value.
4197 2017-11-23 Richard Biener <rguenther@suse.de>
4199 PR tree-optimization/81403
4200 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
4201 a block we need a leader relative to.
4202 (phi_translate_1): For nary processing require a leader from
4203 get_representative_for given we run expression simplification
4204 using match-and-simplify. Remove previous fix.
4206 2017-11-22 Jeff Law <law@redhat.com>
4208 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
4209 Use new method allocate_value_range rather than accessing the
4210 vrp_value_range_pool data member directly.
4211 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
4212 to use extract_range_from_stmt method to avoid need for
4213 extract_range_from_assignment method.
4214 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
4215 method rather than setting values_propgated data member directly.
4216 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
4217 and values propagated data members and extract_range_from_assignment
4218 method. Reorder private data members to conform to standards.
4219 Add new methods set_lattice_propagation_complete and
4220 allocate_value_range.
4222 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
4224 PR rtl-optimization/83030
4225 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
4226 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
4227 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
4228 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
4229 CROSSING_JUMP_P flag is already set before setting it.
4230 * cfgrtl.c (fixup_partition_crossing): Likewise.
4231 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
4234 2017-11-22 Jakub Jelinek <jakub@redhat.com>
4236 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
4237 Handle the case where both arguments are using gen_const_vec_series.
4239 2017-11-22 David Malcolm <dmalcolm@redhat.com>
4242 * pretty-print.c (pp_format): Move quoting implementation to
4243 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
4244 to pass address of "quote" local.
4245 (pp_begin_quote): New function.
4246 (pp_end_quote): New function.
4247 * pretty-print.h (printer_fn): Convert penultimate param from bool
4249 (pp_begin_quote): New decl.
4250 (pp_end_quote): New decl.
4251 * tree-diagnostic.c (default_tree_printer): Convert penultimate
4252 param from bool to bool *.
4253 * tree-diagnostic.h (default_tree_printer): Likewise.
4255 2017-11-22 Jeff Law <law@redhat.com>
4257 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
4258 Initialize vr_values.
4259 (evrp_range_analyzer::try_find_new_range): Call methods attached to
4260 vr_values via vr_values class instance rather than delegators.
4261 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
4262 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
4263 (evrp_range_analyzer::push_value_range): Likewise.
4264 (evrp_range_analyzer::pop_value_range): Likewise.
4265 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
4266 most delegators. Those remaining are exposed as public interfaces.
4267 Make vr_values a pointer and private.
4268 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
4270 (evrp_range_analyzer::get_vr_value): New method.
4271 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
4272 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
4273 (class evrp_dom_walker): Attach evrp_folder class, initialize
4274 it in the ctor. Remove temporary delegators.
4275 (evrp_dom_walker::before_dom_children): Call methods in attached
4276 evrp_range_analyzer class via class instance pointer. Use
4277 free value_range_constant_singleton to remove need for
4278 op_with_constant_singleton_value delegator method. Do not
4279 create a vrp_prop class instance for every call! Narrow
4280 scope of a couple variables.
4281 (evrp_dom_walker::cleanup): Call methods in attached
4282 evrp_range_analyzer class via class instance pointer.
4283 * vr-values.h (class vr_values): Privatize many methods and
4286 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
4288 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
4289 * cfgexpand.c (expand_debug_expr): Remove handling for them.
4290 * expr.c (expand_expr_real_2): Likewise.
4291 * fold-const.c (const_unop): Likewise.
4292 * optabs-tree.c (optab_for_tree_code): Likewise.
4293 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
4294 * tree-inline.c (estimate_operator_cost): Likewise.
4295 * tree-pretty-print.c (dump_generic_node): Likewise.
4296 (op_code_prio): Likewise.
4297 (op_symbol_code): Likewise.
4298 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
4299 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
4300 * internal-fn.c (direct_internal_fn_optab): New function.
4301 (direct_internal_fn_array, direct_internal_fn_supported_p
4302 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
4303 * fold-const-call.c (fold_const_reduction): New function.
4304 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
4306 * tree-vect-loop.c: Include internal-fn.h.
4307 (reduction_code_for_scalar_code): Rename to...
4308 (reduction_fn_for_scalar_code): ...this and return an internal
4310 (vect_model_reduction_cost): Take an internal_fn rather than
4312 (vect_create_epilog_for_reduction): Likewise. Build calls rather
4314 (vectorizable_reduction): Use internal functions rather than tree
4315 codes for the reduction operation. Update calls to the functions
4317 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
4318 Use calls to internal functions rather than REDUC tree codes.
4319 * config/aarch64/aarch64-simd.md: Update comment accordingly.
4321 2017-11-22 Olivier Hainque <hainque@adacore.com>
4323 * config/vxworks.c (vxworks_override_options): Pick default
4324 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
4325 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
4326 DWARF_GNAT_ENCODINGS_DEFAULT.
4327 * config/vxworksae.h: Likewise.
4329 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
4331 PR tree-optimization/83104
4332 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
4335 2017-11-22 Marek Polacek <polacek@redhat.com>
4336 H.J. Lu <hongjiu.lu@intel.com>
4337 Jason Merrill <jason@redhat.com>
4342 * calls.c (initialize_argument_information): Call
4343 warn_parameter_passing_abi target hook.
4344 (store_one_arg): Use 0 for empty record size. Don't push 0 size
4345 argument onto stack.
4346 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
4347 * common.opt: Update -fabi-version description.
4348 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
4349 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
4351 (ix86_is_empty_record): New function.
4352 (ix86_warn_parameter_passing_abi): New function.
4353 (TARGET_EMPTY_RECORD_P): Redefine.
4354 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
4355 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
4356 * doc/tm.texi: Regenerated.
4357 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
4358 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
4359 * dwarf2out.c (get_ultimate_context): Move to tree.c.
4360 * explow.c (hard_function_value): Call arg_int_size_in_bytes
4361 instead of int_size_in_bytes.
4362 * expr.c (copy_blkmode_to_reg): Likewise.
4363 * function.c (aggregate_value_p): Return 0 for empty types.
4364 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
4366 (locate_and_pad_parm): Call arg size_in_bytes instead
4368 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
4369 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
4370 * target.def (empty_record_p, warn_parameter_passing_abi): New target
4372 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
4373 (std_gimplify_va_arg_expr): Skip empty records. Call
4374 arg_size_in_bytes instead size_in_bytes.
4375 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
4376 * tree-core.h (tree_type_common): Add empty_flag.
4377 (tree_decl_common): Update comments.
4378 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
4380 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
4381 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
4383 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
4384 * tree.c (default_is_empty_type): New function.
4385 (default_is_empty_record): New function.
4386 (arg_int_size_in_bytes): New function.
4387 (arg_size_in_bytes): New function.
4388 (get_ultimate_context): New function.
4389 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
4390 TRANSLATION_UNIT_WARN_EMPTY_P.
4391 (default_is_empty_record, arg_int_size_in_bytes,
4392 arg_size_in_bytes, get_ultimate_context): Declare.
4394 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
4396 * config/arm/arm.c (cmse_clear_registers): New function.
4397 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
4398 code by call to cmse_clear_registers.
4399 (cmse_nonsecure_entry_clear_before_return): Likewise.
4401 2017-11-22 Tamar Christina <tamar.christina@arm.com>
4403 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
4404 (vdot_s32, vdotq_s32): New.
4405 (vdot_lane_u32, vdotq_lane_u32): New.
4406 (vdot_lane_s32, vdotq_lane_s32): New.
4409 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
4412 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
4413 for unsigned values with fewer HWIs than the precision.
4414 (test_overflow): New function.
4415 (wide_int_cc_tests): Call it.
4417 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
4419 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
4420 has the same mode class as Pmode.
4422 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
4424 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
4425 padding_bits_to_clear_ptr.
4426 (cmse_nonsecure_entry_clear_before_return): Likewise.
4428 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
4430 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
4431 auto_sbitap instead of integer bitfield to control register needing
4434 2017-11-22 Jakub Jelinek <jakub@redhat.com>
4436 PR tree-optimization/83044
4437 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
4438 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
4439 that need the upper bound. Subtract offset from
4440 get_addr_base_and_unit_offset only if positive and subtract it
4441 before division by eltsize rather than after it.
4444 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
4445 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
4446 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
4450 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
4451 expand_binop with *mul_widen_optab, make sure at least one of the
4452 operands doesn't have VOIDmode.
4455 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
4457 PR rtl-optimization/82044
4458 PR tree-optimization/82042
4459 * dse.c (record_store): Check for overflow.
4460 (check_mem_read_rtx): Properly check for overflow if width == -1, call
4461 add_wild_read instead of clear_rhs_from_active_local_stores on
4462 overflow and log it into dump_file.
4464 2017-11-22 Richard Biener <rguenther@suse.de>
4466 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
4467 fake edges to exit when looking for a place to insert.
4468 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
4470 (insert_into_preds_of_block): Commit edge insertion immediately,
4471 assert that doesn't require new BBs.
4472 (fini_pre): Release expressions.
4473 (pass_pre::execute): Shuffle things around a bit, if the fn
4474 is too large do not compute AVAIL either as this is really the
4477 2017-11-22 Richard Biener <rguenther@suse.de>
4479 PR tree-optimization/83089
4480 * tree-if-conv.c (pass_if_conversion::execute): If anything
4481 changed reset SCEV and free the number of iteration estimates.
4483 2017-11-21 Martin Sebor <msebor@redhat.com>
4485 PR tree-optimization/82945
4486 * calls.h (warn_nonstring_bound): Remove unused function.
4488 2017-11-21 Martin Sebor <msebor@redhat.com>
4490 PR tree-optimization/82945
4491 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
4492 * calls.h (maybe_warn_nonstring_arg): Declare new function.
4493 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
4495 (initialize_argument_information): Call maybe_warn_nonstring_arg.
4496 * calls.h (get_attr_nonstring_decl): Declare new function.
4497 * doc/extend.texi (attribute nonstring): Update.
4498 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
4499 get_attr_nonstring_decl and handle it.
4500 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
4501 detection of nul-termination.
4502 (strlen_to_stridx): Change to a pointer.
4503 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
4504 (pass_strlen::execute): Same.
4506 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
4508 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
4509 for the new option -mprefer-vector-width=[none|128|256|512].
4510 * config/i386/i386.c (ix86_target_string): remove old style options
4511 -mprefer-avx256 and make -mprefer-avx128 as alias.
4512 (ix86_option_override_internal): Apply defaults for the
4513 -mprefer-vector-width=[128|256] option.
4514 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
4515 Implement macros to work with -mprefer-vector-width=.
4516 * config/i386/i386.opt: Implemented option
4517 -mprefer-vector-width=[none|128|256|512].
4518 * doc/invoke.texi: Documentation for
4519 -mprefer-vector-width=[none|128|256|512].
4521 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
4523 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
4525 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
4527 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
4529 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
4531 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
4533 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
4534 * expr.c (expand_expr_real_2): Likewise.
4535 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
4536 fold_binary_loc): Likewise.
4537 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
4538 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
4539 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
4540 MINUS_EXPR transformations.
4541 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
4542 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
4543 * tree-inline.c (estimate_operator_cost): Likewise.
4544 * tree-pretty-print.c (dump_generic_node, op_code_prio,
4545 op_symbol_code): Likewise.
4546 * tree-vect-stmts.c (vectorizable_operation): Likewise.
4547 * vr-values.c (extract_range_from_binary_expr): Likewise.
4548 * varasm.c (initializer_constant_valid_p_1): Likewise.
4549 * tree.def: New tree code POINTER_DIFF_EXPR.
4551 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
4553 * config/i386/i386.md (*bswap<mode>2_movbe): Add
4554 integer suffix to movbe mnemonic.
4555 (*bswaphi2_movbe): Ditto.
4556 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
4558 2017-11-21 Jakub Jelinek <jakub@redhat.com>
4561 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
4562 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
4563 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
4565 2017-11-21 Martin Liska <mliska@suse.cz>
4567 * tree-inline.c (expand_call_inline): Remove not needed
4570 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
4571 Jakub Jelinek <jakub@redhat.com>
4574 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
4575 Remove static keyword from f variable.
4577 2017-11-21 Jakub Jelinek <jakub@redhat.com>
4579 PR tree-optimization/83086
4580 * gimple-ssa-store-merging.c
4581 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
4582 rather than n.base_addr.
4584 2017-11-21 Martin Liska <mliska@suse.cz>
4586 PR rtl-optimization/82044
4587 PR tree-optimization/82042
4588 * dse.c (check_mem_read_rtx): Check for overflow.
4590 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
4592 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
4595 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
4597 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
4598 and its corresponding call arg location note.
4600 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
4601 Andrew Burgess <andrew.burgess@embecosm.com>
4603 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
4605 (arc_return_slot_offset): Likewise.
4606 (arc_eh_return_address_location): New declaration.
4607 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
4608 (MUST_SAVE_REGISTER): Add exception handler case.
4609 (MUST_SAVE_RETURN_ADDR): Likewise.
4610 (arc_frame_pointer_required): Likewise.
4611 (arc_frame_pointer_needed): New function.
4612 (arc_compute_frame_size): Changed.
4613 (arc_expand_prologue): Likewise.
4614 (arc_expand_epilogue): Likewise.
4615 (arc_initial_elimination_offset): Likewise.
4616 (arc_return_slot_offset): Delete.
4617 (arc_eh_return_address_location): New function.
4618 (arc_builtin_setjmp_frame_value): Likewise.
4619 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
4620 (EH_RETURN_STACKADJ_RTX): Define.
4621 (EH_RETURN_HANDLER_RTX): Likewise.
4622 * config/arc/arc.md (eh_return): Delete.
4624 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
4626 * print-tree.h (debug_vec_tree): Remove prototype.
4627 * gdbinit.in (pvt): Remove macro.
4629 2017-11-21 Jakub Jelinek <jakub@redhat.com>
4631 PR tree-optimization/83047
4632 * gimple-ssa-store-merging.c
4633 (imm_store_chain_info::output_merged_store): If the loads with the
4634 same vuse are in different basic blocks, for load_gsi pick a load
4635 location that is dominated by the other loads.
4638 * config/i386/i386.c (ix86_memmodel_check): Start
4639 -Winvalid-memory-model diagnostics with lowercase letter.
4642 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
4643 set in_cold_section_p to the partition containing loc_list->first.
4644 When seeing loc_list->last_before_switch node, update secname and
4645 perform range_across_switch second partition handling only after that.
4648 * run-rtl-passes.c: Include debug.h.
4649 (run_rtl_passes): Call debug_hooks->assembly_start.
4650 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
4654 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
4655 OPTAB_DIRECT in calls to expand_simple_binop.
4657 2017-11-20 David Malcolm <dmalcolm@redhat.com>
4660 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
4662 2017-11-20 David Malcolm <dmalcolm@redhat.com>
4665 * spellcheck.h (best_match::blithely_get_best_candidate): New
4668 2017-11-20 Jakub Jelinek <jakub@redhat.com>
4670 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
4671 with lower case letter.
4673 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
4675 * config/i386/i386.md (bswaphi2): New expander.
4676 (*bswaphi2_movbe): New insn pattern.
4677 (bswaphi -> rorhi pepehole2): New peephole pattern.
4679 2017-11-20 Jeff Law <law@redhat.com>
4681 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
4682 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
4683 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
4684 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
4685 gimple-ssa-evrp-analyze.h.
4687 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
4688 set BB_VISITED here.
4689 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
4691 2017-11-20 Bin Cheng <bin.cheng@arm.com>
4693 * tree-predcom.c: Add general comment on Store-Store chains.
4694 (split_data_refs_to_components): Postpone clearing eliminate_store_p
4696 (get_chain_last_ref_at): Rename into...
4697 (get_chain_last_write_at): ...this.
4698 (get_chain_last_write_before_load): New function.
4699 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
4700 CT_STORE_STORE when write reference is added.
4701 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
4702 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
4703 (initialize_root_vars_store_elim_1): Ditto.
4704 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
4705 definition is created.
4706 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
4707 chain by replacing it with dominant stored value.
4709 2017-11-20 Bin Cheng <bin.cheng@arm.com>
4711 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
4713 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
4715 * vr-values.c (extract_range_from_binary_expr): Use a full range
4718 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
4720 * config/arm/arm.md (R4_REGNUM): Define constant.
4721 (nonsecure_call_internal): Remove r4 clobber.
4722 (nonsecure_call_value_internal): Likewise.
4723 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
4724 clobber and resequence match_operands.
4725 (nonsecure_call_value_reg_thumb1_v5): Likewise.
4726 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
4727 (nonsecure_call_value_reg_thumb2): Likewise.
4729 2017-11-20 Jakub Jelinek <jakub@redhat.com>
4731 PR tree-optimization/78821
4732 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
4733 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
4734 to the address of the base rather than the base itself.
4735 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
4736 (find_bswap_or_nop_finalize): New function.
4737 (find_bswap_or_nop): Use it.
4738 (bswap_replace): Return a tree rather than bool, change first
4739 argument from gimple * to gimple_stmt_iterator, allow inserting
4740 into an empty sequence, allow ins_stmt to be NULL - then emit
4741 all stmts into gsi. Fix up MEM_REF address gimplification.
4742 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
4743 (struct store_immediate_info): Add N and INS_STMT non-static
4745 (store_immediate_info::store_immediate_info): Initialize them
4746 from newly added ctor args.
4747 (merged_store_group::apply_stores): Formatting fixes. Sort by
4749 (stmts_may_clobber_ref_p): For stores call also
4750 refs_anti_dependent_p.
4751 (gather_bswap_load_refs): New function.
4752 (imm_store_chain_info::try_coalesce_bswap): New method.
4753 (imm_store_chain_info::coalesce_immediate_stores): Use it.
4754 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
4755 (imm_store_chain_info::output_merged_store): Fail if number of
4756 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
4757 and NOP_EXPR rhs_code.
4758 (pass_store_merging::process_store): Compute n and ins_stmt, if
4759 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
4760 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
4762 (pass_store_merging::execute): Calculate dominators.
4764 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
4765 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
4766 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
4767 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
4768 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
4769 class pass_optimize_bswap, bswap_replace,
4770 pass_optimize_bswap::execute): Moved to ...
4771 * gimple-ssa-store-merging.c: ... this file.
4772 Include optabs-tree.h.
4773 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
4774 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
4775 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
4776 anonymous namespace, remove static keywords.
4777 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
4778 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
4780 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
4783 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
4785 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
4787 * vec.h (debug_helper): New function.
4788 (DEFINE_DEBUG_VEC): New macro.
4789 * hash-set.h (debug_helper): New function.
4790 (DEFINE_DEBUG_HASH_SET): New macro.
4791 * cfg.c (debug_slim (edge)): New function.
4792 Call DEFINE_DEBUG_VEC for edges.
4793 Call DEFINE_DEBUG_HASH_SET for edges.
4794 * cfghooks.c (debug_slim (basic_block)): New function.
4795 Call DEFINE_DEBUG_VEC for basic blocks.
4796 Call DEFINE_DEBUG_HASH_SET for basic blocks.
4797 * print-tree.c (debug_slim): New function to handle trees.
4798 Call DEFINE_DEBUG_VEC for trees.
4799 Call DEFINE_DEBUG_HASH_SET for trees.
4800 (debug (vec<tree, va_gc>) &): Remove.
4801 (debug (<vec<tree, va_gc>) *): Remove.
4802 * print-rtl.c (debug_slim): New function to handle const_rtx.
4803 Call DEFINE_DEBUG_VEC for rtx_def.
4804 Call DEFINE_DEBUG_VEC for rtx_insn.
4805 Call DEFINE_DEBUG_HASH_SET for rtx_def.
4806 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
4807 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
4808 (debug (vec<rtx_insn *> *ptr): Remove.
4809 (debug_insn_vector): Remove.
4810 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
4812 2017-11-20 Tom de Vries <tom@codesourcery.com>
4814 PR rtl-optimization/82020
4815 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
4816 IF_THEN_ELSE condition.
4818 2017-11-19 Jeff Law <law@redhat.com>
4820 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
4821 of degenerates resulting from ignoring an edge.
4823 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
4826 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
4828 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
4831 * profile-count.c (profile_count::to_sreal_scale): Fix return value
4832 for uninitialied counts.
4834 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
4837 * tree-inline.c (estimate_num_insns): Set to 1 at least.
4839 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
4842 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
4844 2017-11-19 Tom de Vries <tom@codesourcery.com>
4846 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
4849 2017-11-19 Tom de Vries <tom@codesourcery.com>
4851 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
4854 2017-11-19 Tom de Vries <tom@codesourcery.com>
4856 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
4859 2017-11-19 Tom de Vries <tom@codesourcery.com>
4861 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
4863 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
4865 2017-11-19 Tom de Vries <tom@codesourcery.com>
4867 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
4869 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
4870 semicolon after MCORE_EXPORT_NAME call.
4872 2017-11-19 Tom de Vries <tom@codesourcery.com>
4875 * vmsdbgout.c (vmsdbgout_early_finish): New function.
4876 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
4878 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
4880 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
4881 (cgraph_edge::clone): Cleanup updating of profile.
4882 * ipa-cp.c (update_profiling_info): Likewise.
4883 * ipa-inline-transform.c (inline_transform): Likewise.
4884 * ipa-inline.c (inline_small_functions): Add missing space to dump.
4885 * ipa-split.c (execute_split_functions): Do not split when function
4887 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
4888 * profile-count.c (profile_count::dump): Add global0.
4889 (profile_count::to_cgraph_frequency): Do not ICE when entry is
4891 (profile_count::to_sreal_scale): Likewise.
4892 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
4893 (profile_count::combine_with_ipa_count): New function.
4894 * profile-count.h (profile_guessed_global0adjusted): New.
4895 (profile_count::adjusted_zero): New.
4896 (profile_count::global0adjusted): New.
4897 (profile_count::combine_with_ipa_count): New.
4898 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
4899 correct profile of return block of split functions.
4900 (copy_cfg_body): Remove unused profile_count.
4901 (copy_body): Likewise.
4902 (expand_call_inline): Update.
4903 (tree_function_versioning): Update.
4905 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
4907 * hash-set.h (hash_set::empty): New.
4908 * tree-ssa-threadbackward.h: Delete.
4909 * tree-ssa-threadbackward.c (class thread_jumps): New.
4910 Move max_threaded_paths into class.
4911 (fsm_find_thread_path): Remove arguments that are now in class.
4912 (profitable_jump_thread_path): Rename to...
4913 (thread_jumps::profitable_jump_thread_path): ...this.
4914 (convert_and_register_jump_thread_path): Rename to...
4915 (thread_jumps::convert_and_register_current_path): ...this.
4916 (check_subpath_and_update_thread_path): Rename to...
4917 (thread_jumps::check_subpath_and_update_thread_path): ...this.
4918 (register_jump_thread_path_if_profitable): Rename to...
4919 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
4920 (handle_phi): Rename to...
4921 (thread_jumps::handle_phi): ...this.
4922 (handle_assignment): Rename to...
4923 (thread_jumps::handle_assignment): ...this.
4924 (fsm_find_control_statement_thread_paths): Rename to...
4925 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
4926 (find_jump_threads_backwards): Rename to...
4927 (thread_jumps::find_jump_threads_backwards): ...this.
4928 Initialize path local data.
4929 (pass_thread_jumps::execute): Call find_jump_threads_backwards
4930 from within thread_jumps class.
4931 (pass_early_thread_jumps::execute): Same.
4933 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
4935 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
4937 2017-11-17 Steve Ellcey <sellcey@cavium.com>
4939 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
4940 to canonical location.
4942 2017-11-17 Steve Ellcey <sellcey@cavium.com>
4945 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
4947 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
4949 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
4951 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
4952 rcrt1.o%s/grcrt1.o%s for -static-pie.
4954 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
4956 * i386.c (ix86_multiplication_cost, ix86_division_cost,
4957 ix86_shift_rotate_cost): Break out from ...
4958 (ix86_rtx_costs): ... here.
4959 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
4962 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
4964 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
4965 when entry block was promoted unlikely.
4966 (estimate_bb_frequencies): Increase frequency scale.
4967 * profile-count.h (profile_count): Export precision info.
4969 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
4971 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
4972 disturbin profile of entry block.
4974 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
4976 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
4979 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
4981 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
4982 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
4984 2017-11-17 Jeff Law <law@redhat.com>
4986 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
4987 from evrp_dom_walker class. Various methods moved into new class.
4988 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
4989 (evrp_range_analyzer::enter): New method.
4990 (evrp_range_analyzer::leave): New method.
4991 (evrp_dom_walker): Remove delegators no longer needed by this class.
4992 Replace vr_values data member with evrp_range_analyzer
4994 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
4995 method extracted from evrp_dom_walker::before_dom_children.
4996 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
4997 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
4999 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
5000 Add private copy constructor and move assignment operators.
5001 Privatize methods and class data where trivially possible.
5002 (evrp_dom_walker::cleanup): New function, extracted from
5003 execute_early_vrp. Simplify access to class data.
5005 * vr-values.h (get_output_for_vrp): Prototype.
5006 * vr-values.c (get_output_for_vrp): New function extracted from
5007 vrp_visit_assignment_or_call and extract_range_from_stmt.
5008 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
5010 2017-11-17 Luis Machado <luis.machado@linaro.org>
5012 * config/aarch64/aarch64.c
5013 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
5014 (qdf24xx_tunings) <autoprefetcher_model>: Set to
5015 tune_params::AUTOPREFETCHER_WEAK.
5017 2017-11-17 Tamar Christina <tamar.christina@arm.com>
5020 * config/arm/arm.c (arm_valid_target_attribute_rec):
5021 Parse "arch=" and "+<ext>".
5022 (arm_valid_target_attribute_tree): Re-init global options.
5023 (arm_option_override): Make non-static.
5024 (arm_options_perform_arch_sanity_checks): Make errors fatal.
5025 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
5026 (__ARM_FEATURE_CRC32): Support undef.
5027 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
5028 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
5030 2017-11-17 David Malcolm <dmalcolm@redhat.com>
5032 * gdbinit.in (break-on-diagnostic): New command.
5034 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5036 * config/i386/i386.c (ix86_expand_epilogue): Change simple
5037 return to indirect jump for EH return if control-flow
5038 protection is enabled. Change explicit 'false' argument in
5039 pro_epilogue_adjust_stack with a value of
5041 * config/i386/i386.md (simple_return_indirect_internal):
5042 Remove SImode restriction to support 64-bit.
5044 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
5046 * combine.c (added_notes_insn): New.
5047 (try_combine): Handle added_notes_insn like added_links_insn.
5048 Rewrite return value code.
5049 (distribute_notes): Set added_notes_insn to the earliest insn we added
5052 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
5054 PR rtl-optimization/82621
5055 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
5056 dest of one of those SETs is unused.
5058 2017-11-17 Richard Biener <rguenther@suse.de>
5061 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
5062 * tree-pretty-print.c (dump_generic_node): Handle
5063 annot_expr_parallel_kind.
5064 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
5065 * gimplify.c (gimple_boolify): Likewise.
5067 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5069 * config.gcc (extra_headers): Add cet.h for x86 targets.
5070 * config/i386/cet.h: New file.
5071 * doc/install.texi: Add --enable-cet/--disable-cet.
5073 2017-11-17 Richard Biener <rguenther@suse.de>
5075 PR tree-optimization/83017
5076 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
5077 (gen_parallel_loop): Properly count iterations.
5078 (parallelize_loops): Handle loop->can_be_parallel independent
5079 of flag_loop_parallelize_all. Make static profitability test match
5081 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
5082 * invoke.texi (parloops-min-per-thread): Document.
5084 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
5086 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
5087 upstreaming review comments.
5089 2017-11-17 Tamar Christina <tamar.christina@arm.com>
5091 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
5092 with fast unaligned access.
5093 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
5095 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
5097 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
5098 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
5100 2017-11-17 Richard Biener <rguenther@suse.de>
5102 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
5103 folding of references.
5105 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
5108 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
5109 of replacing call to strncmp with corresponding call to strcmp when
5112 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
5114 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
5115 option prefer-avx256 for skylake-avx512 configuration.
5116 * config/i386/i386.c (ix86_option_override_internal): Ditto.
5117 (get_builtin_code_for_version): Ditto.
5119 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
5120 Monk Chiang <sh.chiang04@gmail.com>
5122 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
5123 (FIXED_REGISTERS): Reserve more register numbers.
5124 (CALL_USED_REGISTERS): Likewise.
5125 (REG_ALLOC_ORDER): Likewise.
5126 (REG_CLASS_CONTENTS): Likewise.
5127 (REGISTER_NAMES): Likewise.
5129 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
5130 Kito Cheng <kito.cheng@gmail.com>
5132 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
5134 * config/nds32/iterators.md: Add vector mode iterators and attributes.
5136 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
5138 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
5139 parameter list for vec_splats.
5141 2017-11-16 Joseph Myers <joseph@codesourcery.com>
5143 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
5145 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
5147 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
5150 * doc/install.texi: Document incompatibility of Perl >=5.6.26
5151 with the required version of automake 1.11.6.
5153 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
5155 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
5157 (DU_C2_3_power9): Correct reservation combinations.
5158 (FP_DIV_power9, VEC_DIV_power9): New.
5159 (power9-alu): Split out rotate/shift...
5160 (power9-rot): ...to here, correct dispatch resource.
5161 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
5163 (power9-fp): Correct latency.
5164 (power9-sdiv): Add div/sqrt resource.
5165 (power9-ddiv): Correct latency, add div/sqrt resource.
5166 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
5167 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
5169 (power9-qpdiv, power9-qpmul): Adjust resource usage.
5171 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
5173 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
5174 switch statement mapping KF built-ins to TF built-ins if we don't
5175 have the proper ISA 3.0 assembler support.
5177 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5179 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
5180 (lower_emutls_function_body): Do not compute it.
5182 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5184 * ipa-split.c (split_bb_info): Turn time to sreal.
5185 (split_point): Likewise.
5186 (dump_split_point): Likewise.
5187 (fine_split_points): Likewise.
5188 (execute_split_functions): Only zero split_bbs; turn time to sreals.
5190 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5192 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
5195 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5197 * predict.c (combine_predictions_for_bb): Preserve zero predicted
5199 (expensive_function_p): Remove useless assert.
5200 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
5202 2017-11-16 Martin Sebor <msebor@redhat.com>
5204 PR tree-optimization/82588
5205 PR tree-optimization/82583
5206 * tree-vrp.c (check_array_ref): Handle flexible array members,
5207 string literals, and inner indices.
5208 (search_for_addr_array): Add detail to diagnostics.
5210 2017-11-16 Nathan Sidwell <nathan@acm.org>
5214 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
5215 (SET_DECL_ASSEMBLER_NAME): Forward to
5216 overwrite_decl_assembler_name.
5217 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
5218 (overwrite_decl_assembler_name): Declare.
5219 * tree.c (overwrite_decl_assembler_name): New.
5220 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
5221 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
5222 (LANG_HOOKS_INITIALIZER): Add it.
5223 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
5224 * langhooks.c (lhd_set_decl_assembler_name): Use
5225 SET_DECL_ASSEMBLER_NAME.
5226 (lhd_overwrite_decl_assembler_name): Default implementation.
5228 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
5229 Jackson Woodruff <jackson.woodruff@arm.com>
5231 PR tree-optimization/71026
5232 * match.pd: Canonicalize constant multiplies in division.
5234 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
5236 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
5237 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
5238 Remove OPT_fomit_frame_pointer entry.
5239 * common/config/arc/arc-common.c: Likewise.
5240 * common/config/arm/arm-common.c: Likewise.
5241 * common/config/avr/avr-common.c: Likewise.
5242 * common/config/c6x/c6x-common.c: Likewise.
5243 * common/config/cr16/cr16-common.c: Likewise.
5244 * common/config/cris/cris-common.c: Likewise.
5245 * common/config/epiphany/epiphany-common.c: Likewise.
5246 * common/config/fr30/fr30-common.c: Likewise.
5247 * common/config/frv/frv-common.c: Likewise.
5248 * common/config/ia64/ia64-common.c: Likewise.
5249 * common/config/iq2000/iq2000-common.c: Likewise.
5250 * common/config/lm32/lm32-common.c: Likewise.
5251 * common/config/m32r/m32r-common.c: Likewise.
5252 * common/config/mcore/mcore-common.c: Likewise.
5253 * common/config/microblaze/microblaze-common.c: Likewise.
5254 * common/config/mips/mips-common.c: Likewise.
5255 * common/config/mmix/mmix-common.c: Likewise.
5256 * common/config/mn10300/mn10300-common.c: Likewise.
5257 * common/config/nios2/nios2-common.c: Likewise.
5258 * common/config/pa/pa-common.c: Likewise.
5259 * common/config/pdp11/pdp11-common.c: Likewise.
5260 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
5261 * common/config/riscv/riscv-common.c: Likewise.
5262 * common/config/rs6000/rs6000-common.c: Likewise.
5263 * common/config/rx/rx-common.c: Likewise.
5264 * common/config/s390/s390-common.c: Likewise.
5265 * common/config/sh/sh-common.c: Likewise.
5266 * common/config/sparc/sparc-common.c: Likewise.
5267 * common/config/tilegx/tilegx-common.c: Likewise.
5268 * common/config/tilepro/tilepro-common.c: Likewise.
5269 * common/config/v850/v850-common.c: Likewise.
5270 * common/config/visium/visium-common.c: Likewise.
5271 * common/config/xstormy16/xstormy16-common.c: Likewise.
5272 * common/config/xtensa/xtensa-common.c: Likewise.
5273 * invoke.texi (-fomit-frame-pointer): Update documentation.
5275 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5277 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
5279 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5281 * cfg.c (scale_bbs_frequencies_int,
5282 cale_bbs_frequencies_gcov_type): Remove.
5283 * cfg.h (scale_bbs_frequencies_int,
5284 cale_bbs_frequencies_gcov_type): Remove.
5286 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5288 * tree-ssa-loop-manip.c
5289 (scale_dominated_blocks_in_loop): Update to profile counts.
5290 (tree_transform_and_unroll_loop): Likewise.
5292 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5294 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
5295 scale_bbs_frequencies_int.
5297 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5299 * final.c (compute_alignments): Use counts rather than frequencies.
5301 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5303 * cfgloopanal.c: Include sreal.h
5304 (average_num_loop_insns): Use counts and sreal for accounting.
5306 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5308 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
5311 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
5312 Kito Cheng <kito.cheng@gmail.com>
5314 * config/nds32/constraints.md: Provide more constraints.
5315 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
5316 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
5317 support constraints usage.
5319 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
5321 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
5323 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
5324 Kito Cheng <kito.cheng@gmail.com>
5326 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
5327 * config/nds32/nds32.opt: Refine the layout.
5328 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
5329 TARGET_EXT_STRING): Support new options.
5330 * config/nds32/nds32.h: Likewise.
5331 * config/nds32/nds32.md: Likewise.
5332 * config/nds32/nds32-predicates.c: Likewise.
5333 * config/nds32/constraints.md: Likewise.
5334 * common/config/nds32/nds32-common.c: Likewise.
5336 2017-11-16 Julia Koval <julia.koval@intel.com>
5339 * config/i386/gfniintrin.h: Add sse check.
5340 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
5342 2017-11-16 Julia Koval <julia.koval@intel.com>
5344 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
5345 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
5346 (ix86_handle_option): Handle -mavx512vbmi2.
5347 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
5348 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
5349 * config/i386/i386-c.c (__AVX512VBMI2__): New.
5350 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
5351 (ix86_valid_target_attribute_inner_p): Ditto.
5352 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
5353 * config/i386/i386.opt (mavx512vbmi2): New option.
5354 * doc/invoke.texi: Add new option.
5356 2017-11-16 Julia Koval <julia.koval@intel.com>
5358 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
5359 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
5360 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
5361 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
5362 _mm512_gf2p8mul_epi8): New intrinsics.
5363 * config/i386/i386-builtin-types.def
5364 (V64QI_FTYPE_V64QI_V64QI): New type.
5365 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
5366 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
5367 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
5368 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
5369 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
5370 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
5372 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
5374 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
5375 explicitly as a stream of bytes.
5377 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5379 * config/rs6000/altivec.h (vec_xst_be): New #define.
5380 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
5381 and externalize from *altivec_vperm_<mode>_internal.
5382 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
5384 (XL_BE_V8HI): Likewise.
5385 (XL_BE_V4SI): Likewise.
5386 (XL_BE_V4SI): Likewise.
5387 (XL_BE_V2DI): Likewise.
5388 (XL_BE_V4SF): Likewise.
5389 (XL_BE_V2DF): Likewise.
5390 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
5391 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
5392 all array entries with these keys: VSX_BUILTIN_VEC_XL,
5393 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
5394 VSX_BUILTIN_VEC_XST_BE.
5395 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
5396 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
5398 (altivec_init_builtins): Replace conditional calls to def_builtin
5399 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
5400 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
5401 with unconditional calls. Remove calls to def_builtin for
5402 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
5403 __builtin_vec_xst_be.
5404 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
5405 to define_expand, and add alternate RTL generation for P8.
5406 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
5407 vsx_ld_elemrev_v8hi.
5408 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
5409 add alternate RTL generation for P8.
5410 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
5411 vsx_ld_elemrev_v16qi.
5412 (vsx_st_elemrev_v8hi): Convert define_insn
5413 to define_expand, and add alternate RTL generation for P8.
5414 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
5415 vsx_st_elemrev_v8hi.
5416 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
5417 add alternate RTL generation for P8.
5418 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
5419 vsx_st_elemrev_v16qi.
5421 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
5424 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
5425 TARGET_AVX512ER check.
5426 (ix86_option_override_internal): Set MASK_VZEROUPPER if
5427 neither -mzeroupper nor -mno-zeroupper is used and
5428 TARGET_EMIT_VZEROUPPER is set.
5429 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
5430 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
5432 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
5434 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
5435 folding of vector compares.
5436 (fold_build_vec_cmp): New helper function.
5437 (fold_compare_helper): New helper function.
5438 (builtin_function_type): Add compare builtins to the list of functions
5439 having unsigned arguments. Cosmetic updates to comment indentation.
5440 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
5441 the not+eq combination.
5443 2017-11-15 Bin Cheng <bin.cheng@arm.com>
5445 PR tree-optimization/82726
5446 PR tree-optimization/70754
5447 * tree-predcom.c (order_drefs_by_pos): New function.
5448 (combine_chains): Move code setting has_max_use_after to...
5449 (try_combine_chains): ...here. New parameter. Sort combined chains
5450 according to position information.
5451 (tree_predictive_commoning_loop): Update call to above function.
5452 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
5454 2017-11-15 Bin Cheng <bin.cheng@arm.com>
5456 PR tree-optimization/82726
5458 2017-01-23 Bin Cheng <bin.cheng@arm.com>
5460 PR tree-optimization/70754
5461 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
5462 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
5463 combined stmt before it if not NULL.
5464 (combine_chains): Process refs reversely and compute dominance point
5468 2017-02-23 Bin Cheng <bin.cheng@arm.com>
5470 PR tree-optimization/79663
5471 * tree-predcom.c (combine_chains): Process refs in reverse order
5472 only for ZERO length chains, and add explaining comment.
5474 2017-11-15 Tamar Christina <tamar.christina@arm.com>
5476 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
5477 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
5478 * doc/invoke.texi (ARM Options): Add armv8.3-a.
5480 2017-11-15 Tamar Christina <tamar.christina@arm.com>
5482 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
5484 2017-11-15 Martin Liska <mliska@suse.cz>
5486 * tree-cfg.c (pass_warn_function_return::execute):
5487 Compare warn_return_type for greater than zero.
5489 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
5493 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
5494 to return true on Xeon and not on Xeon Phi.
5495 (ix86_check_avx256_register): Changed to ...
5496 (ix86_check_avx_upper_register): ... this. Add extra check for
5497 VALID_AVX512F_REG_OR_XI_MODE.
5498 (ix86_avx_u128_mode_needed): Changed
5499 ix86_check_avx256_register to ix86_check_avx_upper_register.
5500 (ix86_check_avx256_stores): Changed to ...
5501 (ix86_check_avx_upper_stores): ... this. Changed
5502 ix86_check_avx256_register to ix86_check_avx_upper_register.
5503 (ix86_avx_u128_mode_after): Changed
5504 avx_reg256_found to avx_upper_reg_found. Changed
5505 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
5506 (ix86_avx_u128_mode_entry): Changed
5507 ix86_check_avx256_register to ix86_check_avx_upper_register.
5508 (ix86_avx_u128_mode_exit): Ditto.
5509 * config/i386/i386.h: (host_detect_local_cpu): New define.
5511 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
5513 * config/arm/xgene1.md (xgene1): Split into automatons
5514 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
5515 (xgene1_f_load): Adjust reservations and/or types.
5516 (xgene1_f_store): Likewise.
5517 (xgene1_load_pair): Likewise.
5518 (xgene1_store_pair): Likewise.
5519 (xgene1_fp_load1): Likewise.
5520 (xgene1_load1): Likewise.
5521 (xgene1_store1): Likewise.
5522 (xgene1_move): Likewise.
5523 (xgene1_alu): Likewise.
5524 (xgene1_simd): Likewise.
5525 (xgene1_bfm): Likewise.
5526 (xgene1_neon_load1): Likewise.
5527 (xgene1_neon_store1): Likewise.
5528 (xgene1_neon_logic): Likewise.
5529 (xgene1_neon_st1): Likewise.
5530 (xgene1_neon_ld1r): Likewise.
5531 (xgene1_alu_cond): Added.
5532 (xgene1_shift_reg): Likwise.
5533 (xgene1_bfx): Likewise.
5534 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
5536 2017-11-15 Jakub Jelinek <jakub@redhat.com>
5539 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
5541 (can_widen_mult_without_libcall): New function.
5542 (expand_mul_overflow): If only checking unsigned mul overflow,
5543 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
5544 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
5545 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
5546 (expand_DIVMOD): Formatting fix.
5547 * expmed.h (expand_mult): Add NO_LIBCALL argument.
5548 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
5549 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
5551 PR tree-optimization/82977
5552 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
5553 constructed temporary to strlen_to_stridx.put.
5555 2017-11-15 Martin Liska <mliska@suse.cz>
5557 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
5558 * configure: Regenerate.
5560 2017-11-15 Martin Liska <mliska@suse.cz>
5563 * config/sh/sh-mem.cc: Use proper probability for
5566 2017-11-14 Jeff Law <law@redhat.com>
5568 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
5569 the red zone for stack_clash_protection_final_dynamic_probe targets
5570 when the total dynamic stack size is zero bytes.
5572 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
5573 blocks is post order.
5575 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
5577 * dumpfile.h (TDF_COMPARE_DEBUG): New.
5578 * final.c (rest_of_clean_state): Set it for the
5579 -fcompare-debug dump.
5580 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
5581 class when TDF_COMPARE_DEBUG is set.
5583 * dwarf2out.c (gen_producer_string): Discard
5586 2017-11-15 Joseph Myers <joseph@codesourcery.com>
5589 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
5590 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
5591 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
5592 (__TGMATH_REAL_2_3): Remove macros.
5593 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
5594 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
5596 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
5598 (remquo): Define using __TGMATH_REAL_3.
5600 2017-11-14 Jeff Law <law@redhat.com>
5602 * vr-values.c: New file with contents extracted from tree-vrp.c.
5603 * Makefile.in (OBJS): Add vr-values.o
5604 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
5605 (set_value_range, set_and_canonicalize_value_range): Likewise.
5606 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
5607 (value_range_constant_singleton, symbolic_range_p): Likewise.
5608 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
5609 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
5610 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
5611 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
5612 (find_case_label_range, find_case_label_index): Likewise.
5613 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
5614 (range_int_cst_singleton_p, value_inside_range): Likewise.
5615 (get_single_symbol): Likewise.
5616 (switch_update): Move structure definition here.
5617 (to_remove_edges, to_update_switch_stmts): Provide externs.
5618 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
5619 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
5620 (vrp_val_is_min, set_value_range): Likewise.
5621 (set_and_canonicalize_value_range, copy_value_range): Likewise.
5622 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
5623 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
5624 (range_is_nonnull, range_int_cst_p): Likewwise.
5625 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
5626 (get_single_symbol, operand_less_p): Likewise
5627 (compare_values_warnv, compare_values): Likewise.
5628 (value_inside_range, value_range_constant_singleton): Likewise.
5629 (zero_nonzero_bitgs_from_vr): Likewise.
5630 (extract_range_from_binary_expr_1): Likewise.
5631 (overflow_comparison_p): Likewise.
5632 (to_remove_edges, to_update_switch_stmts): Likewise.
5633 (find_case_label-index, find_case_label_range): Likewise.
5634 (switch_update, set_value_range_to_nonnegative): Remove.
5635 (set_value_range_to_truthvalue): Likewise.
5636 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
5637 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
5638 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
5639 (find_case_label_ranges, test_for_singularity): Likewise.
5640 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
5641 (x_vr_values): Move to its remaining use site.
5643 2017-11-10 Jeff Law <law@redhat.com>
5645 * vr-values.h (VR_INITIALIZER): Move #define here.
5646 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
5647 * Makefile.in (OBJS): Add tree-evrp.o
5648 * tree-vrp.h (assert_info): Move structure definition here.
5649 (set_value_range_to_varying): Prototype.
5650 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
5651 (infer_value_range, register_edge_assert_for): Likewise.
5652 (stmt_interesting_for_vrp): Likewise.
5653 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
5654 (set_value_range_to_varying): No longer static.
5655 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
5656 (infer_value_range, register_edge_assert_for): Likewise.
5658 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
5660 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
5661 generate the XXBRD instruction.
5663 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
5664 (rs6000_builtin_type_compatible): Treat _Float128 and long double
5665 as being compatible if -mabi=ieeelongdouble.
5666 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
5667 to setup float128 built-ins with hardware support.
5668 (BU_FLOAT128_HW_2): Likewise.
5669 (BU_FLOAT128_HW_3): Likewise.
5670 (BU_FLOAT128_HW_VSX_1): Likewise.
5671 (BU_FLOAT128_HW_VSX_2): Likewise.
5672 (scalar_extract_expq): Change float128 built-in functions to
5673 accommodate having both KFmode and TFmode functions. Use the
5674 KFmode variant as the default.
5675 (scalar_extract_sigq): Likewise.
5676 (scalar_test_neg_qp): Likewise.
5677 (scalar_insert_exp_q): Likewise.
5678 (scalar_insert_exp_qp): Likewise.
5679 (scalar_test_data_class_qp): Likewise.
5680 (sqrtf128_round_to_odd): Delete processing the round to odd
5681 built-in functions as special built-in functions, and define them
5682 as float128 built-ins. Use the KFmode variant as the default.
5683 (truncf128_round_to_odd): Likewise.
5684 (addf128_round_to_odd): Likewise.
5685 (subf128_round_to_odd): Likewise.
5686 (mulf128_round_to_odd): Likewise.
5687 (divf128_round_to_odd): Likewise.
5688 (fmaf128_round_to_odd): Likewise.
5689 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
5690 support for KFmode and TFmode xststdcqp calls.
5691 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
5692 point, switch the built-in handlers for the get/set float128
5693 exponent, get float128 mantissa, float128 test built-ins, and the
5694 float128 round to odd built-in functions. Eliminate creating the
5695 float128 round to odd built-in functions as special built-ins.
5696 (rs6000_init_builtins): Eliminate special creation of the float128
5697 round to odd built-in functions.
5698 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
5699 function insns to support both TFmode and KFmode variants.
5700 (xsxsigqp_<mode>): Likewise.
5701 (xsiexpqpf_<mode>): Likewise.
5702 (xsiexpqp_<mode>): Likewise.
5703 (xststdcqp_<mode>): Likewise.
5704 (xststdcnegqp_<mode>): Likewise.
5705 (xststdcqp_<mode>): Likewise.
5707 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5709 * tree-ssa-threadupdate.c (compute_path_counts): Remove
5710 unused path_in_freq_ptr parameter.
5711 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
5713 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5715 * ipa-inline.c (edge_badness): Dump sreal frequency.
5716 (compute_inlined_call_time): Match natural implementaiton ...
5717 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
5718 forgotten division by CGRAPH_FREQ_BASE.
5720 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5722 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
5723 Solaris 11. Update comment.
5724 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
5726 * config/sol2.h (STARTFILE_SPEC): Likewise.
5727 * configure: Regenerate.
5729 2017-11-14 Carl Love <cel@us.ibm.com>
5731 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
5732 le_ and be_ prefixes to swap* variables. Remove
5733 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
5736 2017-11-14 Jason Merrill <jason@redhat.com>
5738 Support GTY((cache)) on hash_map.
5739 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
5740 (ggc_cache_remove): Override it instead of ggc_mx.
5741 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
5742 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
5743 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
5744 (simple_cache_map_traits): Override maybe_mx.
5745 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
5746 (hash_map): Friend gt_cleare_cache.
5747 (gt_cleare_cache): New.
5748 * tree.h (tree_cache_traits): New hash_map traits class.
5749 (tree_cache_map): New typedef.
5751 2017-11-14 Richard Biener <rguenther@suse.de>
5753 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
5754 paramter and handling.
5755 (cleanup_control_flow_bb): Likewise.
5756 (cleanup_control_flow_pre): New helper performing a DFS walk
5757 to call cleanup_control_flow_bb in PRE order.
5758 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
5759 via cleanup_control_flow_pre.
5761 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
5763 * config/aarch64/aarch64-simd.md
5764 (aarch64_simd_bsl<mode>_internal): Remove DImode.
5765 (*aarch64_simd_bsl<mode>_alt): Likewise.
5766 (aarch64_simd_bsldi_internal): New.
5767 (aarch64_simd_bsldi_alt): Likewise.
5769 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5771 * tracer.c (better_p): Do not compare frequencies.
5772 * reg-stack.c (better_edge): Likewise.
5773 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
5776 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5778 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
5779 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
5780 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
5781 * lto-streamer-in.c (input_function): Use update_max_bb_count.
5782 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
5783 * predict.c (maybe_hot_frequency_p): Inline to ...
5784 (maybe_hot_count_p): ... here; rewrite to counts.
5785 (counts_to_freqs): Rename to ...
5786 (update_max_bb_count): ... this one.
5787 (expensive_function_p): Use counts.
5788 (estimate_bb_frequencies): Update.
5789 (rebuild_frequencies): Update.
5790 * predict.h (counts_to_freqs): Rename to ...
5791 (update_max_bb_count): ... this one.
5792 * profile.c (compute_branch_probabilities): Add debug info
5793 * tree-inline.c (expand_call_inline): Update debug info.
5794 (optimize_inline_calls): Use update_max_bb_count..
5795 (tree_function_versioning): Use update_max_bb_count..
5796 * value-prof.c (gimple_value_profile_transformations):
5797 Do not use update_max_bb_count.
5799 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5801 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
5802 always use frequencies.
5804 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5806 * bb-reorder.c: Remove frequencies from comments.
5807 (better_edge_p): Use profile counts.
5808 (find_traces): Dump profile counts.
5809 (rotate_loop): Use profile counts.
5810 (find_traces_1_round): Likewise.
5811 (connect_better_edge_p): Use counts instead of probabilities for
5813 (copy_bb_p): Drop early check for non-0 frequency.
5814 (sanitize_hot_paths): Update comments.
5816 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5818 * ipa-split.c (struct split_point): Add count.
5819 (consider_split): Do not compute incoming frequency; compute incoming
5820 count and store it to split_point.
5821 (split_function): Set count of the call to split part correctly.
5823 2017-11-13 Carl Love <cel@us.ibm.com>
5825 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
5827 2017-11-13 Tom Tromey <tom@tromey.com>
5829 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
5831 2017-11-13 Carl Love <cel@us.ibm.com>
5833 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
5834 Add support for builtins:
5835 unsigned int vec_first_{,miss}_match_{,or_eos}index,
5836 vector {un,}signed {char,int,short},
5837 vector {un,}signed {char,int,short}) arguments.
5838 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
5839 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
5840 Add BU_P9V_AV_2 expansions for the builtins.
5841 * config/rs6000/altivec.h (vec_first_match_index,
5842 vec_first_mismatch_index, vec_first_match_or_eos_index,
5843 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
5844 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
5845 new extern declaration.
5846 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
5847 * config/rs6000/vsx.md (first_match_index_<mode>,
5848 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
5849 first_mismatch_or_eos_index_<mode>): Add define expand.
5850 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
5851 * doc/extend.texi: Update the built-in documenation file for the new
5854 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
5856 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
5857 into the min/max operations for _Float<N> and _Float<N>X types.
5859 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
5862 * dwarf2out.c (do_eh_frame): New static variable.
5863 (dwarf2out_begin_prologue): Set it.
5864 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
5866 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5868 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
5870 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
5872 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
5875 * debug.h (dwarf2out_do_eh_frame): Declare.
5876 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
5877 (dwarf2out_do_frame): Use it.
5878 (dwarf2out_do_cfi_asm): Likewise.
5879 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
5880 (dwarf2out_assembly_start): Likewise.
5881 (dwarf2out_begin_prologue): Fix comment.
5882 * toplev.c (compile_file): Always call dwarf2out_frame_finish
5883 if the target needs either debug or unwind DWARF2 info.
5884 * lto-opts.c (lto_write_options): Do not save -fexceptions,
5885 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
5886 -ftrapping-math, -ftrapv and -fwrapv.
5888 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5890 * cgraph.c (cgraph_edge::sreal_frequency): New function.
5891 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
5892 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
5893 (estimate_edge_size_and_time): Likewise.
5894 (ipa_merge_fn_summary_after_inlining): Likewise.
5895 * ipa-inline.c (cgraph_freq_base_rec): Remove.
5896 (compute_uninlined_call_time): Use sreal_frequency.
5897 (compute_inlined_call_time): Likewise.
5898 (ipa_inline): Do not initialize cgraph_freq_base_rec.
5899 * profile-count.c: Include sreal.h.
5900 (profile_count::to_sreal_scale): New.
5901 * profile-count.h: Forward declare sreal.
5902 (profile_count::to_sreal_scale): Declare.
5904 2017-11-13 Nathan Sidwell <nathan@acm.org>
5906 * diagnostic.c (maybe_line_and_column): New.
5907 (diagnostic_get_location_text): Use it.
5908 (diagnostic_report_current_module): Likewise.
5909 (test_diagnostic_get_location_text): Add tests.
5911 2017-11-13 Luis Machado <luis.machado@linaro.org>
5913 * doc/md.texi (Specifying processor pipeline description): Fix
5914 incorrect latency for the div instruction example.
5916 2017-11-13 Jakub Jelinek <jakub@redhat.com>
5918 PR tree-optimization/78821
5919 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
5920 that bit_not_p is the same.
5921 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
5922 (split_group): Count precisely bit_not_p bits in each statement.
5923 (invert_op): New function.
5924 (imm_store_chain_info::output_merged_store): Use invert_op to
5925 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
5926 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
5928 2017-11-13 Martin Liska <mliska@suse.cz>
5930 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
5931 (struct source_info): Likewise.
5932 (add_branch_counts): Likewise.
5933 (add_line_counts): Likewise.
5934 (function_summary): Likewise.
5935 (output_intermediate_line): Likewise.
5936 (generate_results): Likewise.
5938 2017-11-13 Martin Liska <mliska@suse.cz>
5940 * gcov.c (struct block_info): Remove typedef for block_t.
5941 (struct line_info): Likewise.
5942 (line_info::has_block): Likewise.
5943 (EXIT_BLOCK): Likewise.
5944 (unblock): Likewise.
5945 (circuit): Likewise.
5946 (get_cycles_count): Likewise.
5947 (process_file): Likewise.
5948 (read_graph_file): Likewise.
5949 (solve_flow_graph): Likewise.
5950 (find_exception_blocks): Likewise.
5951 (add_line_counts): Likewise.
5952 (accumulate_line_info): Likewise.
5953 (output_line_details): Likewise.
5955 2017-11-13 Martin Liska <mliska@suse.cz>
5957 * gcov.c (struct arc_info): Remove typedef for arc_t.
5958 (struct line_info): Likewise.
5959 (add_branch_counts): Likewise.
5960 (output_branch_count): Likewise.
5961 (function_info::~function_info): Likewise.
5962 (circuit): Likewise.
5963 (output_intermediate_line): Likewise.
5964 (read_graph_file): Likewise.
5965 (solve_flow_graph): Likewise.
5966 (find_exception_blocks): Likewise.
5967 (add_line_counts): Likewise.
5968 (accumulate_line_info): Likewise.
5969 (output_line_details): Likewise.
5970 (output_function_details): Likewise.
5972 2017-11-13 Martin Liska <mliska@suse.cz>
5974 * gcov.c (struct function_info): Remove typedef for function_t.
5975 (struct source_info): Likewise.
5976 (source_info::get_functions_at_location): Likewise.
5977 (solve_flow_graph): Likewise.
5978 (find_exception_blocks): Likewise.
5979 (add_line_counts): Likewise.
5980 (output_intermediate_file): Likewise.
5981 (process_file): Likewise.
5982 (generate_results): Likewise.
5983 (release_structures): Likewise.
5984 (read_graph_file): Likewise.
5985 (read_count_file): Likewise.
5986 (accumulate_line_counts): Likewise.
5987 (output_lines): Likewise.
5989 2017-11-13 Martin Liska <mliska@suse.cz>
5991 * gcov.c (function_info::function_info): Remove num_counts
5992 and add vector<gcov_type>.
5993 (function_info::~function_info): Use the vector.
5994 (process_file): Likewise.
5995 (read_graph_file): Likewise.
5996 (read_count_file): Likewise.
5997 (solve_flow_graph): Likewise.
5999 2017-11-13 Martin Liska <mliska@suse.cz>
6001 * gcov.c (function_info::is_artificial): New function.
6002 (process_file): Erase all artificial early.
6003 (generate_results): Skip as all artificial are already
6006 2017-11-13 Martin Liska <mliska@suse.cz>
6008 * gcov.c (read_graph_file): Store to global vector of functions.
6009 (read_count_file): Iterate the vector.
6010 (process_file): Likewise.
6011 (generate_results): Likewise.
6012 (release_structures): Likewise.
6014 2017-11-13 Jakub Jelinek <jakub@redhat.com>
6016 PR tree-optimization/82954
6017 * gimple-ssa-store-merging.c
6018 (imm_store_chain_info::coalesce_immediate_stores): If
6019 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
6021 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
6023 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
6024 Upddate call to ENDIAN_LANE_N.
6025 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
6026 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
6027 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
6028 and use aarch64_endian_lane_rtx.
6029 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
6031 2017-11-12 Tom de Vries <tom@codesourcery.com>
6033 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
6035 2017-11-12 Tom de Vries <tom@codesourcery.com>
6037 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
6039 2017-11-12 Tom de Vries <tom@codesourcery.com>
6041 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
6043 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
6044 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
6045 * defaults.h (ASM_OUTPUT_LABELREF): Same.
6047 2017-11-11 Martin Sebor <msebor@redhat.com>
6050 * doc/extend.texi (attribute nonstring): Remove spurious argument.
6053 * prefic.c (translate_name): Replace strncpy with memcpy to
6054 avoid -Wstringop-truncation.
6056 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6058 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
6060 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6062 * predict.c (maybe_hot_frequency_p): Do not use cfun.
6064 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6066 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
6069 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
6071 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
6072 enable generating XXBRH if the value is in a vector register.
6073 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
6074 value is in a vector register.
6075 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
6076 register to register bswap64's instead of doing the GPR sequence
6077 used on previous machines.
6078 (bswapdi2_xxbrd): New insn.
6079 (bswapdi2_reg): Disallow on ISA 3.0.
6080 (register to register bswap64 splitter): Do not split the insn on
6081 ISA 3.0 systems that use XXBRD.
6083 2017-11-10 Martin Sebor <msebor@redhat.com>
6086 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
6088 (find_subframework_file): Same.
6090 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6092 * auto-profile.c (afdo_indirect_call): Drop frequency.
6093 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
6094 (cgraph_node::create_edge): Drop frequency argument.
6095 (cgraph_node::create_indirect_edge): Drop frequency argument.
6096 (cgraph_edge::make_speculative): Drop frequency arguments.
6097 (cgraph_edge::resolve_speculation): Do not update frequencies
6098 (cgraph_edge::dump_edge_flags): Do not dump frequency.
6099 (cgraph_node::dump): Check consistency in IPA mode.
6100 (cgraph_edge::maybe_hot_p): Use IPA counter.
6101 (cgraph_edge::verify_count_and_frequency): Rename to ...
6102 (cgraph_edge::verify_count): ... this one; drop frequency checking.
6103 (cgraph_node::verify_node): Update.
6104 * cgraph.h (struct cgraph_edge): Drop frequency.
6105 (cgraph_edge::frequency): New function.
6106 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
6108 (cgraph_edge::rebuild_edges): Likewise.
6109 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
6110 (duplicate_thunk_for_node): Do not pass frequency.
6111 (cgraph_node::create_clone): Scale only counts.
6112 (cgraph_node::create_virtual_clone): Do not pass frequency.
6113 (cgraph_node::create_edge_including_clones): Do not pass frequency.
6114 (cgraph_node::create_version_clone): Do not pass frequency.
6115 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
6116 (cgraph_node::expand_thunk): Do not pass frequency.
6117 (cgraph_node::create_wrapper): Do not pass frequency.
6118 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
6120 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
6121 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
6122 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
6123 (gather_caller_stats): Use frequency function.
6124 (ipcp_cloning_candidate_p): Use frequency function.
6125 (ipcp_propagate_stage): Use frequency function.
6126 (get_info_about_necessary_edges): Use frequency function.
6127 (update_profiling_info): Update only IPA profile.
6128 (update_specialized_profile): Use frequency functoin.
6129 (perhaps_add_new_callers): Update only IPA profile.
6130 * ipa-devirt.c (ipa_devirt): Use IPA profile.
6131 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
6132 (dump_ipa_call_summary): Use frequency function.
6133 (estimate_edge_size_and_time): Use frequency function.
6134 (ipa_merge_fn_summary_after_inlining): Use frequency function.
6135 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
6136 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
6137 (update_noncloned_counts): ... ths one; scale counts only.
6138 (clone_inlined_nodes): Do not scale frequency.
6139 (inline_call): Do not pass frequency.
6140 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
6141 (compute_inlined_call_time): Use IPA profile.
6142 (want_inline_small_function_p): Use IPA profile.
6143 (want_inline_self_recursive_call_p): Use IPA profile.
6144 (edge_badness): Use IPA profile.
6145 (lookup_recursive_calls): Use IPA profile.
6146 (recursive_inlining): Do not pass frequency.
6147 (resolve_noninline_speculation): Do not update frequency.
6148 (inline_small_functions): Collect max of IPA profile.
6149 (dump_overall_stats): Dump IPA porfile.
6150 (dump_inline_stats): Dump IPA porfile.
6151 (ipa_inline): Collect IPA stats.
6152 * ipa-inline.h (clone_inlined_nodes): Update prototype.
6153 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
6154 (ipa_propagate_frequency): Use frequency function.
6155 (ipa_profile): Cleanup.
6156 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
6157 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
6158 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
6159 (input_node): Do not stream frequency.
6160 (input_edge): Do not stream frequency.
6161 (merge_profile_summaries): Scale only IPA profiles.
6162 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
6163 * predict.c (drop_profile): Do not recompute frequency.
6164 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
6165 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
6166 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
6167 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
6168 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
6169 * tree-inline.c (copy_bb): Do not scale frequency.
6170 (expand_call_inline): Do not scale frequency.
6171 (tree_function_versioning): Do not scale frequency.
6172 * ubsan.c (ubsan_create_edge): Do not pass frequency.
6174 2017-11-10 Julia Koval <julia.koval@intel.com>
6176 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
6177 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
6178 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
6179 (_mm256_maskz_gf2p8affine_epi64_epi8)
6180 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
6181 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
6182 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
6183 (__builtin_ia32_vgf2p8affineqb_v32qi)
6184 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
6185 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
6187 2017-11-10 Tamar Christina <tamar.christina@arm.com>
6191 (arm_option_override): Refactor.
6192 (arm_option_reconfigure_globals): New.
6193 (arm_options_perform_arch_sanity_checks): New.
6194 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
6196 (arm_options_perform_arch_sanity_checks): Likewise
6198 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
6200 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
6201 (power9-qpmul): New.
6202 * rs6000/rs6000.md ("type" attr): Add qmul.
6203 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
6204 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
6205 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
6207 2017-11-10 Martin Sebor <msebor@redhat.com>
6210 * builtins.c (compute_objsize): Handle arrays that
6211 compute_builtin_object_size likes to fail for. Make extern.
6212 * builtins.h (compute_objsize): Declare.
6213 (check_strncpy_sizes): New function.
6214 (expand_builtin_strncpy): Call check_strncpy_sizes.
6215 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
6216 -Wstringop-truncation.
6217 (gimple_fold_builtin_strncat): Same.
6218 * gimple.c (gimple_build_call_from_tree): Set call location.
6219 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
6220 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
6221 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
6222 (handle_builtin_strlen): Use strlen_to_stridx.
6223 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
6225 Use strlen_to_stridx.
6226 (pass_strlen::execute): Release strlen_to_stridx.
6227 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
6228 (-Wstringop-truncation): Document new option.
6230 2017-11-10 Martin Liska <mliska@suse.cz>
6232 PR gcov-profile/82702
6233 * gcov.c (main): Handle intermediate files in a different
6235 (get_gcov_intermediate_filename): New function.
6236 (output_gcov_file): Remove support of intermediate files.
6237 (generate_results): Allocate intermediate file.
6238 (release_structures): Clean-up properly fn_end.
6239 (output_intermediate_file): Start iterating with line 1.
6241 2017-11-10 Jakub Jelinek <jakub@redhat.com>
6243 PR tree-optimization/82929
6244 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
6245 ops_swapped_p non-static data member.
6246 (store_immediate_info::store_immediate_info): Clear it.
6247 (imm_store_chain_info::coalesce_immediate_stores): If swapping
6248 ops set ops_swapped_p.
6249 (count_multiple_uses): Handle ops_swapped_p.
6251 2017-11-10 Martin Liska <mliska@suse.cz>
6253 * coverage.c (coverage_init): Stream information about
6254 support of has_unexecuted_blocks.
6255 * doc/gcov.texi: Document that.
6256 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
6257 * gcov.c (read_graph_file): Likewise.
6258 (output_line_beginning): Fix a small issue with
6261 2017-11-10 Bin Cheng <bin.cheng@arm.com>
6263 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
6264 reference of trivial component.
6266 2017-11-10 Jakub Jelinek <jakub@redhat.com>
6269 * gimple-ssa-store-merging.c
6270 (pass_store_merging::terminate_all_aliasing_chains): For
6271 gimple_store_p stmts also call refs_output_dependent_p.
6273 PR rtl-optimization/82913
6274 * compare-elim.c (try_merge_compare): Punt if def_insn is not
6277 2017-11-09 Jeff Law <law@redhat.com>
6279 * vr-values.h: New file with vr_values class.
6280 * tree-vrp.c: Include vr-values.h
6281 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
6282 data objects into the vr_values class.
6283 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
6284 (get_value_range): Make it a member function within vr_values class.
6285 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
6286 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
6287 (op_with_constant_singleton_value_range): Likewise.
6288 (extract_range_for_var_from_comparison_expr): Likewise.
6289 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
6290 (extract_range_from_binary_expr): Likewise.
6291 (extract_range_from_unary_expr): Likewise.
6292 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
6293 (check_for_binary_op_overflow, extract_range_basic): Likewise.
6294 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
6295 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
6296 (compare_name_with_value, compare_names): Likewise.
6297 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
6298 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
6299 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
6300 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
6301 (extract_range_from_phi_node): Likewise.
6302 (simplify_truth_ops_using_ranges): Likewise.
6303 (simplify_div_or_mod_using_ranges): Likewise.
6304 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
6305 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
6306 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
6307 (simplify_float_conversion_using_ranges): Likewise.
6308 (simplify_internal_call_using_ranges): Likewise.
6309 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
6310 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
6311 poitner via x_vr_values for calls into gimple folder.
6312 (vrp_initialize_lattice): Make this the vr_values ctor.
6313 (vrp_free_lattice): Make this the vr_values dtor.
6314 (set_vr_value): New function.
6315 (class vrp_prop): Add vr_values data member. Add various member
6316 functions as well as member functions that delegate to vr_values.
6317 (check_array_ref): Make a member function within vrp_prop class.
6318 (search_for_addr_array, vrp_initialize): Likewise.
6319 (vrp_finalize): Likewise. Revamp to avoid direct access to
6320 vr_value, values_propagated, etc.
6321 (check_array_bounds): Extract vrp_prop class instance pointer from
6322 walk info structure. Use it to call member functions.
6323 (check_all_array_refs): Make a member function within vrp_prop class.
6324 Smuggle class instance pointer via walk info structure.
6325 (x_vr_values): New local static.
6326 (vrp_valueize): Use x_vr_values to get class instance.
6327 (vr_valueize_1): Likewise.
6328 (class vrp_folder): Add vr_values data member. Add various member
6329 functions as well as member functions that delegate to vr_values.
6330 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
6331 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
6332 class instance from vr_values. Use it to call member functions.
6333 (vrp_dom_walker): Add vr_values data member.
6334 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
6335 instance via x_vr_values.
6336 (identify_jump_threads): Accept vr_values as argument. Store
6337 it into the walker structure.
6338 (evrp_dom_walker): Add vr_values class data member. Add various
6340 (evrp_dom_walker::try_find_new_range): Use vr_values data
6341 member to access the memory allocator.
6342 (evrp_dom_walker::before_dom_children): Store vr_values class
6343 instance into the vrp_folder class.
6344 (evrp_dom_walker::push_value_range): Rework to avoid direct
6345 access to num_vr_values and vr_value.
6346 (evrp_dom_walker::pop_value_range): Likewise.
6347 (execute_early_vrp): Remove call to vrp_initialize_lattice.
6348 Use vr_values to get to dump_all_value_ranges member function.
6349 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
6350 and simplify_cond_using_ranges_2 via vrp_prop class instance.
6351 Pass vr_values class instance down to identify_jump_threads.
6352 Remove call to vrp_free_lattice.
6353 (debug_all_value_ranges): Remove.
6355 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
6356 (vrp_folder): Likewise.
6358 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
6359 Get it from the existing bitmap instead.
6360 (vrp_intersect_ranges_1): Likewise.
6362 2017-11-09 Jakub Jelinek <jakub@redhat.com>
6364 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
6366 (store_immediate_info::store_immediate_info): Add bitnotp argument,
6367 set bit_not_p to it.
6368 (imm_store_chain_info::coalesce_immediate_stores): Break group
6369 if bit_not_p is different.
6370 (count_multiple_uses, split_group,
6371 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
6372 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
6373 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
6374 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
6375 to store_immediate_info ctor.
6377 2017-11-09 Jim Wilson <jimw@sifive.com>
6379 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
6380 (scan_prog_file): Likewise.
6382 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
6384 * bb-reorder.c (max_entry_frequency): Remove.
6385 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
6386 connect_traces, push_to_next_round_p): Remove prototypes.
6387 (find_traces_1_round): Use counts only.
6388 (push_to_next_round_p): Likewise.
6389 (find_traces): Likewise.
6390 (rotate_loop): Likewise.
6391 (find_traces_1_round): Likewise.
6392 (connect_traces): Likewise.
6393 (edge_order): Likewise.
6395 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
6397 * config/arm/arm.c (output_return_instruction): Add comments to
6398 indicate requirement for cmse_nonsecure_entry return to account
6399 for the size of clearing instruction output here.
6400 (thumb_exit): Likewise.
6401 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
6402 return in hardfloat mode.
6404 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
6406 * config/rs6000/rs6000.c (machine_function): Add a bool,
6407 "toc_is_wrapped_separately".
6408 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
6409 if it wasn't explicitly set or unset, we are optimizing for speed, and
6410 doing separate shrink-wrapping.
6411 (rs6000_get_separate_components): Enable the TOC component if
6412 saving the TOC register in the prologue.
6413 (rs6000_components_for_bb): Handle the TOC component.
6414 (rs6000_emit_prologue_components): Store the TOC register where needed.
6415 (rs6000_set_handled_components): Mark TOC as handled, if handled.
6416 (rs6000_emit_prologue): Don't save the TOC if that is already done.
6418 2017-11-09 Martin Jambor <mjambor@suse.cz>
6420 * ipa-param-manipulation.c: New file.
6421 * ipa-param-manipulation.h: Likewise.
6422 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
6423 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
6424 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
6425 (ipa_parm_adjustment): Likewise.
6426 (ipa_parm_adjustment_vec): Likewise.
6427 (ipa_get_vector_of_formal_parms): Moved declaration to
6428 ipa-param-manipulation.h.
6429 (ipa_get_vector_of_formal_parm_types): Likewise.
6430 (ipa_modify_formal_parameters): Likewise.
6431 (ipa_modify_call_arguments): Likewise.
6432 (ipa_combine_adjustments): Likewise.
6433 (ipa_dump_param_adjustments): Likewise.
6434 (ipa_modify_expr): Likewise.
6435 (ipa_get_adjustment_candidate): Likewise.
6436 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
6437 ipa-param-manipulation.c.
6438 (ipa_get_vector_of_formal_parm_types): Likewise.
6439 (ipa_modify_formal_parameters): Likewise.
6440 (ipa_modify_call_arguments): Likewise.
6441 (ipa_modify_expr): Likewise.
6442 (get_ssa_base_param): Likewise.
6443 (ipa_get_adjustment_candidate): Likewise.
6444 (index_in_adjustments_multiple_times_p): Likewise.
6445 (ipa_combine_adjustments): Likewise.
6446 (ipa_dump_param_adjustments): Likewise.
6447 * tree-sra.c: Also include ipa-param-manipulation.h
6448 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
6451 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6452 Alan Hayward <alan.hayward@arm.com>
6453 David Sherwood <david.sherwood@arm.com>
6455 * doc/sourcebuild.texi (vect_masked_store): Document.
6457 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6458 Alan Hayward <alan.hayward@arm.com>
6459 David Sherwood <david.sherwood@arm.com>
6461 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
6463 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6464 Alan Hayward <alan.hayward@arm.com>
6465 David Sherwood <david.sherwood@arm.com>
6467 * doc/sourcebuild.texi (vect_variable_length): Document.
6469 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6470 Alan Hayward <alan.hayward@arm.com>
6471 David Sherwood <david.sherwood@arm.com>
6473 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
6475 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6476 Alan Hayward <alan.hayward@arm.com>
6477 David Sherwood <david.sherwood@arm.com>
6479 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
6481 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6482 Alan Hayward <alan.hayward@arm.com>
6483 David Sherwood <david.sherwood@arm.com>
6485 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
6486 previously undocumented selectors.
6487 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
6489 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6491 * doc/rtl.texi (const_vector): Say that elements can be
6492 const_wide_ints too.
6493 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
6494 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
6495 (gen_vec_duplicate): Use it instead of CONSTANT_P.
6496 * optabs.c (expand_vector_broadcast): Likewise.
6498 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6499 Alan Hayward <alan.hayward@arm.com>
6500 David Sherwood <david.sherwood@arm.com>
6502 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
6503 scaled index even if the unscaled address was invalid.
6504 Don't increase the complexity of using a scale in that case.
6506 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6507 Alan Hayward <alan.hayward@arm.com>
6508 David Sherwood <david.sherwood@arm.com>
6510 * doc/rtl.texi: Rewrite the subreg rules so that they partition
6511 the inner register into REGMODE_NATURAL_SIZE bytes rather than
6512 UNITS_PER_WORD bytes.
6513 * emit-rtl.c (validate_subreg): Divide subregs into blocks
6514 based on REGMODE_NATURAL_SIZE of the inner mode.
6515 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
6516 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
6517 * expmed.c (lowpart_bit_field_p): Divide the value up into
6518 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
6519 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
6520 whether something is likely to occupy more than one register.
6522 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
6525 * ipa-inline-transform.c (update_noncloned_frequencies): Use
6526 profile_count::adjust_for_ipa_scaling.
6527 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
6528 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
6530 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
6532 2017-11-09 Jakub Jelinek <jakub@redhat.com>
6534 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
6535 (split_group): Add total_orig and total_new arguments, estimate the
6536 number of statements related to the store group without store merging
6537 and with store merging.
6538 (imm_store_chain_info::output_merged_store): Adjust split_group
6539 callers, punt if estimated number of statements with store merging
6540 is not smaller than estimated number of statements without it.
6542 (handled_load): Remove has_single_use checks.
6543 (pass_store_merging::process_store): Likewise.
6545 2017-11-09 Richard Biener <rguenther@suse.de>
6547 PR tree-optimization/82902
6548 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
6550 2017-11-09 Martin Liska <mliska@suse.cz>
6553 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
6556 2017-11-09 Martin Liska <mliska@suse.cz>
6558 PR tree-optimization/82669
6559 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
6561 2017-11-09 Martin Liska <mliska@suse.cz>
6563 PR gcov-profile/48463
6564 * coverage.c (coverage_begin_function): Output also end locus
6565 of a function and information whether the function is
6567 * gcov-dump.c (tag_function): Parse and print the information.
6568 * gcov.c (INCLUDE_MAP): Add include.
6569 (INCLUDE_SET): Likewise.
6570 (struct line_info): Move earlier in the source file because
6571 of vector<line_info> in function_info structure.
6572 (line_info::line_info): Likewise.
6573 (line_info::has_block): Likewise.
6574 (struct source_info): Add new member index.
6575 (source_info::get_functions_at_location): New function.
6576 (function_info::group_line_p): New function.
6577 (output_intermediate_line): New function.
6578 (output_intermediate_file): Use the mentioned function.
6579 (struct function_start): New.
6580 (struct function_start_pair_hash): Likewise.
6581 (process_file): Add code that identifies group functions.
6582 Assign lines either to global or function scope.
6583 (generate_results): Skip artificial functions.
6584 (find_source): Assign index for each source file.
6585 (read_graph_file): Read new flag artificial and end_line.
6586 (add_line_counts): Assign it either to global of function scope.
6587 (accumulate_line_counts): Isolate core of the function to
6588 accumulate_line_info and call it for both function and global
6590 (accumulate_line_info): New function.
6591 (output_line_beginning): Fix GNU coding style.
6592 (print_source_line): New function.
6593 (output_line_details): Likewise.
6594 (output_function_details): Likewise.
6595 (output_lines): Iterate both source (global) scope and function
6597 (struct function_line_start_cmp): New class.
6598 * doc/gcov.texi: Reflect changes in documentation.
6600 2017-11-09 Jakub Jelinek <jakub@redhat.com>
6603 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
6604 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
6605 and similarly for not instead of neg.
6607 2017-11-08 Andi Kleen <ak@linux.intel.com>
6609 * config/i386/i386.opt: Add -mforce-indirect-call.
6610 * config/i386/predicates.md: Check for flag_force_indirect_call.
6611 * doc/invoke.texi: Document -mforce-indirect-call
6613 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
6615 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
6617 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
6618 config/riscv/riscv.c (predict.h): New include.
6619 (riscv_slow_unaligned_access_p): No longer static.
6620 (riscv_block_move_straight): Add require.
6621 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
6623 2017-11-08 Jakub Jelinek <jakub@redhat.com>
6626 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
6627 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
6628 nonimmediate_operand predicate for operand 1 instead of
6631 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6633 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
6635 * config/aarch64/constraints.md (Uml): New constraint.
6636 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
6639 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6641 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
6642 of two vec_duplicates into a vec_concat.
6644 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6646 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
6647 Simplify vec_merge of vec_duplicate and vec_concat.
6648 * config/aarch64/constraints.md (Utq): New constraint.
6649 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
6652 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6654 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
6655 Simplify vec_merge of vec_duplicate and const_vector.
6656 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
6658 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
6659 mode iterator. Update predicate on operand 1 to
6660 handle non-const_vec constants. Delete constraints.
6661 (*aarch64_combinez_be<mode>): Likewise for operand 2.
6663 2017-11-08 Jakub Jelinek <jakub@redhat.com>
6665 PR tree-optimization/78821
6666 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
6668 (store_operand_info::store_operand_info): Initialize it to false.
6669 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
6670 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
6671 store in the group, and if chain_info is non-NULL, to ignore altogether
6673 (compatible_load_p): Fail if bit_not_p does not match.
6674 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
6675 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
6676 (pass_store_merging::process_store): Adjust
6677 terminate_all_aliasing_chains calls to pass NULL in all current spots,
6678 call terminate_all_aliasing_chains newly when adding a store into
6679 a chain with non-NULL chain_info.
6681 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
6683 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
6685 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
6687 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
6689 (aarch64_layout_frame): Initialise emit_frame_chain.
6690 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
6691 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
6693 2017-11-08 Martin Liska <mliska@suse.cz>
6695 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
6696 of gimple_call_internal_p.
6698 2017-11-07 Tom de Vries <tom@codesourcery.com>
6700 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
6702 2017-11-07 Tom de Vries <tom@codesourcery.com>
6704 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
6707 2017-11-08 Martin Liska <mliska@suse.cz>
6710 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
6712 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
6714 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
6716 2017-11-07 Jakub Jelinek <jakub@redhat.com>
6719 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
6720 (*cmp<mode>_ccz_1): New insn with $k alternative.
6723 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
6724 RTX_COMM_COMPARE as commutative as well.
6725 (ix86_binary_operator_ok): Formatting fix.
6726 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
6727 *<code><mode>3<mask_name><round_saeonly_name>,
6728 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
6729 *<s>mul<mode>3_highpart<mask_name>,
6730 *vec_widen_umult_even_v16si<mask_name>,
6731 *vec_widen_umult_even_v8si<mask_name>,
6732 *vec_widen_umult_even_v4si<mask_name>,
6733 *vec_widen_smult_even_v16si<mask_name>,
6734 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
6735 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
6736 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
6737 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
6738 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
6739 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
6740 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
6741 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
6742 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
6743 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
6744 ix86_binary_operator_ok. Formatting fixes.
6745 (*<plusminus_insn><mode>3<mask_name><round_name>,
6746 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
6749 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
6751 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
6752 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
6753 eq and ne if TARGET_ISEL.
6754 (cmp): New code_iterator.
6755 (UNS, UNSU_, UNSIK): New code_attrs.
6756 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
6757 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
6759 ("eq<mode>3"): ... this.
6760 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
6762 ("ne<mode>3"): ... this.
6764 2017-11-07 Julia Koval <julia.koval@intel.com>
6767 * common/config/i386/i386-common.c
6768 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
6769 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
6770 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
6771 * config/i386/i386.opt: Ditto.
6772 * config/i386/i386.c (ix86_target_string): Ditto.
6773 (ix86_option_override_internal): Ditto.
6774 (ix86_init_mpx_builtins): Move MPX to args2.
6775 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
6776 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
6777 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
6778 __builtin_ia32_vgf2p8affineinvqb_v32qi,
6779 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
6780 __builtin_ia32_vgf2p8affineinvqb_v16qi,
6781 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
6783 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
6786 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
6787 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
6788 (zero-extendsidi peephole2): Remove peephole.
6790 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
6793 * stor-layout.c: Include attribs.h.
6794 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
6795 explicit lookup of "aligned" attribute.
6797 2017-11-07 Andrew Waterman <andrew@sifive.com>
6799 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
6800 (riscv_expand_block_move): Likewise.
6801 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
6803 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
6804 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
6805 * config/riscv/riscv.c (riscv_block_move_straight): New function.
6806 (riscv_adjust_block_mem): Likewise.
6807 (riscv_block_move_loop): Likewise.
6808 (riscv_expand_block_move): Likewise.
6809 * config/riscv/riscv.md (movmemsi): New pattern.
6811 2017-11-07 Michael Clark <michaeljclark@mac.com>
6813 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
6814 (MUSL_DYNAMIC_LINKER): Likewise.
6816 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
6818 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
6819 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
6822 2017-11-07 Richard Biener <rguenther@suse.de>
6824 * match.pd: Fix build.
6826 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
6827 Jackson Woodruff <jackson.woodruff@arm.com>
6829 PR tree-optimization/71026
6830 * match.pd: Canonicalize negate in division.
6832 2017-11-07 Sudakshina Das <sudi.das@arm.com>
6835 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
6837 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
6839 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
6840 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
6842 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
6844 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
6845 non-scalar integral types.
6846 * match.pd (negate_expr_p): Handle MINUS_EXPR.
6847 (-(A-B), -(~A)): New transformations.
6849 2017-11-07 Tom de Vries <tom@codesourcery.com>
6851 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
6852 semicolon after "do {} while (0)".
6853 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6854 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6855 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6856 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6857 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6859 2017-11-07 Tom de Vries <tom@codesourcery.com>
6861 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
6862 after "do {} while (0)".
6863 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6864 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6865 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6866 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6867 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6869 2017-11-07 Tom de Vries <tom@codesourcery.com>
6872 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
6874 (arm_rtx_costs_internal): Add missing semicolon after
6875 HANDLE_NARROW_SHIFT_ARITH call.
6877 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
6879 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
6880 disable isel if it was not set explicitly.
6882 2017-11-06 James Bowman <james.bowman@ftdichip.com>
6884 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
6885 (add_type_attribute) likewise.
6887 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
6889 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
6890 of struct ix86_frame.
6891 (ix86_initial_elimination_offset): Likewise.
6892 (ix86_expand_split_stack_prologue): Likewise.
6894 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
6896 * tree-vrp.h (enum value_range_type): Update stale comment.
6898 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
6899 Alan Hayward <alan.hayward@arm.com>
6900 David Sherwood <david.sherwood@arm.com>
6902 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
6903 (aarch64_expand_vec_perm_const): Take the number of units too.
6904 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
6905 (aarch64_expand_vec_perm_const): Likewise.
6906 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
6907 (vec_perm<mode>): Update accordingly.
6909 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
6910 Alan Hayward <alan.hayward@arm.com>
6911 David Sherwood <david.sherwood@arm.com>
6913 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
6914 Take the number of units too.
6915 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
6916 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
6917 but check for a vector mode before rather than after the call.
6918 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
6919 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
6920 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
6921 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
6922 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
6923 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
6924 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
6925 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
6926 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
6927 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
6928 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
6929 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
6930 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
6931 (aarch64_sqdmull2_n<mode>): Update accordingly.
6933 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
6934 Alan Hayward <alan.hayward@arm.com>
6935 David Sherwood <david.sherwood@arm.com>
6937 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
6938 the number of units too.
6939 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
6940 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
6941 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
6942 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
6943 (vec_store_lanesxi<mode>): Update accordingly.
6945 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
6946 Alan Hayward <alan.hayward@arm.com>
6947 David Sherwood <david.sherwood@arm.com>
6949 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
6950 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
6951 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
6952 of units rather than the mode.
6953 * config/aarch64/iterators.md (nunits): New mode attribute.
6954 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
6955 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
6956 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
6957 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
6958 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
6959 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
6960 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
6961 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
6962 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
6963 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
6964 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
6965 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
6966 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
6967 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
6968 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
6969 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
6970 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
6971 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
6972 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
6973 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
6974 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
6975 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
6976 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
6977 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
6978 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
6979 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
6980 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
6981 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
6982 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
6983 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
6984 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
6985 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
6986 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
6987 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
6988 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
6989 (aarch64_simd_vec_setv2di): Likewise.
6991 2017-11-06 Carl Love <cel@us.ibm.com>
6993 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
6995 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
6996 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
6998 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
7000 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
7001 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
7002 Add power 8 macro expansions.
7003 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
7004 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
7005 power 8 instructions. (VSX_XXBR): Add iterator.
7007 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
7009 * config/arm/arm.md (predicable_short_it): Change default to "no",
7010 improve documentation, remove uses that are identical to the default.
7011 (enabled_for_depr_it): Rename to enabled_for_short_it.
7012 * gcc/config/arm/arm-fixed.md (predicable_short_it):
7013 Remove default uses.
7014 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
7015 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
7016 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
7017 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
7019 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
7022 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
7023 float128 helper macros, which are no longer used after deleting
7024 the old 'q' built-in functions, and moving the round to odd
7025 built-in functions to being special built-in functions.
7026 (BU_FLOAT128_2): Likewise.
7027 (BU_FLOAT128_1_HW): Likewise.
7028 (BU_FLOAT128_2_HW): Likewise.
7029 (BU_FLOAT128_3_HW): Likewise.
7030 (FABSQ): Delete old 'q' built-in functions.
7031 (COPYSIGNQ): Likewise.
7032 (SQRTF128_ODD): Move round to odd built-in functions to be
7033 special built-in functions, so that we can handle
7034 -mabi=ieeelongdouble.
7035 (TRUNCF128_ODD): Likewise.
7036 (ADDF128_ODD): Likewise.
7037 (SUBF128_ODD): Likewise.
7038 (MULF128_ODD): Likewise.
7039 (DIVF128_ODD): Likewise.
7040 (FMAF128_ODD): Likewise.
7041 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
7042 built-in names to 'f128'.
7043 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
7044 old 'q' built-in functions, as the machine independent code for
7045 'f128' built-in functions handles this.
7046 (rs6000_expand_builtin): Add expansion for float128 round to odd
7047 functions, keying off on -mabi=ieeelongdouble of whether to use
7048 the KFmode or TFmode variant.
7049 (rs6000_init_builtins): Initialize the _Float128 round to odd
7051 * doc/extend.texi (PowerPC Built-in Functions): Document the old
7052 _Float128 'q' built-in functions are now mapped into the new
7053 'f128' built-in functions.
7055 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
7057 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
7058 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
7060 2017-11-06 David Malcolm <dmalcolm@redhat.com>
7063 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
7064 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
7065 * toplev.c: Include "ipa-fnsummary.h".
7066 (toplev::finalize): Call ipa_fnsummary_c_finalize.
7068 2017-11-06 Jakub Jelinek <jakub@redhat.com>
7070 PR tree-optimization/82838
7071 * gimple-ssa-store-merging.c
7072 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
7073 on a separate gimple_seq which is then appended to seq.
7075 2017-11-06 Jeff Law <law@redhat.com>
7078 * config/i386/i386.c (PROBE_INTERVAL): Remove.
7079 (get_probe_interval): New functions.
7080 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
7081 (ix86_adjust_stack_and_probe): Likewise.
7082 (output_adjust_stack_and_probe): Likewise.
7083 (ix86_emit_probe_stack_range): Likewise.
7084 (ix86_expand_prologue): Likewise.
7086 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7088 PR tree-optimization/82816
7089 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
7090 if the modes of the two types are the same.
7091 (convert_plusminus_to_widen): Likewise.
7093 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7095 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
7097 (usadv16qi): New define_expand.
7098 (usadv8hi): New define_expand.
7100 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
7103 * ipa-inline-transform.c (update_noncloned_frequencies): Always
7105 (inline_transform): Likewise.
7106 * predict.c (counts_to_freqs): Remove useless conditional.
7107 * profile-count.h (profile_count::apply_scale): Move sanity check.
7108 * tree-inline.c (copy_bb): Always scale.
7109 (copy_cfg_body): Likewise.
7111 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
7114 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
7116 (*cmp_ite0): Add enabled_for_depr_it attribute.
7117 (*cmp_ite1): Likewise.
7119 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
7121 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
7124 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7126 * tree-vrp.c (vrp_int_const_binop): Return true on success and
7127 return the value by pointer.
7128 (extract_range_from_multiplicative_op_1): Update accordingly.
7129 Return as soon as an operation fails.
7131 2017-11-05 Tom de Vries <tom@codesourcery.com>
7134 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
7135 (DEF_SANITIZER_BUILTIN): ... here.
7136 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
7137 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
7139 2017-11-05 Tom de Vries <tom@codesourcery.com>
7142 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
7144 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
7145 ASM_OUTPUT_BEFORE_CASE_LABEL call.
7146 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
7148 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
7149 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
7150 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
7152 2017-11-05 Tom de Vries <tom@codesourcery.com>
7155 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
7158 2017-11-04 Michael Clark <michaeljclark@mac.com>
7160 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
7161 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
7163 (*addsi3_extended): Likewise.
7164 (*addsi3_extended2): Likewise.
7165 (<optab>si3): Likewise.
7166 (<optab>di3): Likewise.
7167 (<optab><mode>3): Likewise.
7168 (<*optabe>si3_internal): Likewise.
7169 (zero_extendqi<SUPERQI:mode>2): Likewise.
7170 (*add<mode>hi3): Likewise.
7171 (*xor<mode>hi3): Likewise.
7172 (<optab>di3): Likewise.
7173 (*<optab>si3_extend): Likewise.
7174 (*sge<u>_<X:mode><GPR:mode>): Likewise.
7175 (*slt<u>_<X:mode><GPR:mode>): Likewise.
7176 (*sle<u>_<X:mode><GPR:mode>): Likewise.
7178 2017-11-04 Andrew Waterman <andrew@sifive.com>
7180 * config/riscv/riscv.c (riscv_option_override): Conditionally set
7181 TARGET_STRICT_ALIGN based upon -mtune argument.
7183 2017-11-04 Andrew Waterman <andrew@sifive.com>
7185 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
7187 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
7189 * config/i386/i386.c (choose_basereg): Use optional scratch
7190 register and add assertion.
7191 (x86_emit_outlined_ms2sysv_save): Use scratch register when
7192 needed, and don't allocate stack.
7193 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
7194 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
7195 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
7197 2017-11-03 Jeff Law <law@redhat.com>
7199 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
7200 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
7201 to probe at the start of a noreturn function.
7203 2017-11-03 Jakub Jelinek <jakub@redhat.com>
7205 PR tree-optimization/78821
7206 * gimple-ssa-store-merging.c: Update the file comment.
7207 (MAX_STORE_ALIAS_CHECKS): Define.
7208 (struct store_operand_info): New type.
7209 (store_operand_info::store_operand_info): New constructor.
7210 (struct store_immediate_info): Add rhs_code and ops data members.
7211 (store_immediate_info::store_immediate_info): Add rhscode, op0r
7212 and op1r arguments to the ctor, initialize corresponding data members.
7213 (struct merged_store_group): Add load_align_base and load_align
7215 (merged_store_group::merged_store_group): Initialize them.
7216 (merged_store_group::do_merge): Update them.
7217 (merged_store_group::apply_stores): Pick the constant for
7218 encode_tree_to_bitpos from one of the two operands, or skip
7219 encode_tree_to_bitpos if neither operand is a constant.
7220 (class pass_store_merging): Add process_store method decl. Remove
7221 bool argument from terminate_all_aliasing_chains method decl.
7222 (pass_store_merging::terminate_all_aliasing_chains): Remove
7223 var_offset_p argument and corresponding handling.
7224 (stmts_may_clobber_ref_p): New function.
7225 (compatible_load_p): New function.
7226 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
7227 if there is overlap and rhs_code is not INTEGER_CST. For
7228 non-overlapping stores terminate group if rhs is not mergeable.
7229 (get_alias_type_for_stmts): Change first argument from
7230 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
7231 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
7232 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
7234 (get_location_for_stmts): Change first argument from
7235 auto_vec<gimple *> & to vec<gimple *> &.
7236 (struct split_store): Remove orig_stmts data member, add orig_stores.
7237 (split_store::split_store): Create orig_stores rather than orig_stmts.
7238 (find_constituent_stmts): Renamed to ...
7239 (find_constituent_stores): ... this. Change second argument from
7240 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
7241 to info structures rather than the statements.
7242 (split_group): Rename ALLOW_UNALIGNED argument to
7243 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
7244 it. Adjust find_constituent_stores caller.
7245 (imm_store_chain_info::output_merged_store): Handle rhs_code other
7246 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
7247 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
7248 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
7249 (mem_valid_for_store_merging): New function.
7250 (handled_load): New function.
7251 (pass_store_merging::process_store): New method.
7252 (pass_store_merging::execute): Use process_store method. Adjust
7253 terminate_all_aliasing_chains caller.
7255 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
7257 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
7258 Return true for more constants, symbols and label references.
7259 (aarch64_valid_floating_const): Remove unused function.
7261 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
7264 * config/aarch64/aarch64.c (aarch64_layout_frame):
7265 Undo forcing of LR at bottom of frame.
7267 2017-11-03 Jeff Law <law@redhat.com>
7270 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
7271 for int_registers_saved.
7273 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
7274 extracted from tree-ssa-dom.c.
7275 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
7276 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
7277 (record_equivalences_from_incoming_edge): Add additional argument
7278 to single_pred_edge_ignoring_loop_edges call.
7279 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
7280 (uncprop_dom_walker::before_dom_children): Add additional argument
7281 to single_pred_edge_ignoring_loop_edges call.
7282 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
7283 single_pred_edge_ignoring_loop_edges rather than open coding.
7284 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
7286 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
7288 * match.pd (-(-A)): Rewrite.
7290 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
7292 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
7293 (rs6000_emit_int_cmove): New declaration.
7294 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
7295 (rs6000_emit_sISEL): Delete.
7296 (rs6000_emit_int_cmove): Make non-static.
7297 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
7298 instead of rs6000_emit_sISEL.
7300 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
7302 * asan.c (create_cond_insert_point): Maintain profile.
7303 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
7305 * basic-block.h (struct basic_block_def): Remove frequency.
7306 (EDGE_FREQUENCY): Use to_frequency
7307 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
7309 (find_traces): Update to use to_frequency.
7310 (find_traces_1_round): Likewise; use only IPA counts.
7311 (bb_to_key): Likewise.
7312 (connect_traces): Use IPA counts only.
7313 (copy_bb_p): Update to use to_frequency.
7314 (fix_up_crossing_landing_pad): Likewise.
7315 (sanitize_hot_paths): Likewise.
7316 * bt-load.c (basic_block_freq): Likewise.
7317 * cfg.c (init_flow): Set count_max to uninitialized.
7318 (check_bb_profile): Remove frequencies; check counts.
7319 (dump_bb_info): Do not dump frequencies.
7320 (update_bb_profile_for_threading): Update counts only.
7321 (scale_bbs_frequencies_int): Likewise.
7322 (MAX_SAFE_MULTIPLIER): Remove.
7323 (scale_bbs_frequencies_gcov_type): Update counts only.
7324 (scale_bbs_frequencies_profile_count): Update counts only.
7325 (scale_bbs_frequencies): Update counts only.
7326 * cfg.h (struct control_flow_graph): Add count-max.
7327 (update_bb_profile_for_threading): Update prototype.
7328 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
7329 (find_many_sub_basic_blocks): Likewise.
7330 * cfgcleanup.c (try_forward_edges): Likewise.
7331 (try_crossjump_to_edge): Likewise.
7332 * cfgexpand.c (expand_gimple_cond): Likewise.
7333 (expand_gimple_tailcall): Likewise.
7334 (construct_init_block): Likewise.
7335 (construct_exit_block): Likewise.
7336 * cfghooks.c (verify_flow_info): Check consistency of counts.
7337 (dump_bb_for_graph): Do not dump frequencies.
7338 (split_block_1): Do not update frequencies.
7339 (split_edge): Do not update frequencies.
7340 (make_forwarder_block): Do not update frequencies.
7341 (duplicate_block): Do not update frequencies.
7342 (account_profile_record): Do not update frequencies.
7343 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
7344 for global heuristics.
7345 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
7346 (expected_loop_iterations_unbounded): Use counts only.
7347 * cfgloopmanip.c (scale_loop_profile): Simplify.
7348 (create_empty_loop_on_edge): Simplify
7350 (duplicate_loop_to_header_edge): Simplify
7351 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
7352 (update_br_prob_note): Take care of removing note when profile
7354 (relink_block_chain): Do not dump frequency.
7355 (rtl_account_profile_record): Use to_frequency.
7356 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
7357 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
7358 (cgraph_update_edges_for_call_stmt_node): Likewise.
7359 (cgraph_edge::verify_count_and_frequency): Update.
7360 (cgraph_node::verify_node): Temporarily disable frequency verification.
7361 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
7362 to_cgraph_frequency.
7363 (cgraph_edge::rebuild_edges): Convert to ipa counts.
7364 * cgraphunit.c (init_lowered_empty_function): Do not initialize
7366 (cgraph_node::expand_thunk): Update profile.
7367 * except.c (dw2_build_landing_pads): Do not update frequency.
7368 * final.c (compute_alignments): Use to_frequency.
7369 (dump_basic_block_info): Do not dump frequency.
7370 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
7371 (dump_gimple_bb_header): Do not dump frequency.
7372 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
7374 * gimple-streamer-in.c (input_bb): Do not stream frequency.
7375 * gimple-streamer-out.c (output_bb): Do not stream frequency.
7376 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
7377 (init_before_recovery): Do not update frequency.
7378 (sched_create_recovery_edges): Do not update frequency.
7379 * hsa-gen.c (convert_switch_statements): Do not update frequency.
7380 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
7381 (ipa_cp_c_finalize): Set max_count to uninitialized.
7382 * ipa-fnsummary.c (get_minimal_bb): Use counts.
7383 (param_change_prob): Use counts.
7384 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
7386 * ipa-split.c (consider_split): Use to_frequency.
7387 (split_function): Use to_frequency.
7388 * ira-build.c (loop_compare_func): Likewise.
7389 (mark_loops_for_removal): Likewise.
7390 (mark_all_loops_for_removal): Likewise.
7391 * loop-doloop.c (doloop_modify): Do not update frequency.
7392 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
7394 * lto-streamer-in.c (input_function): Update count_max.
7395 * omp-expand.c (expand_omp_taskreg): Update count_max.
7396 * omp-simd-clone.c (simd_clone_adjust): Update profile.
7397 * predict.c (maybe_hot_frequency_p): Use to_frequency.
7398 (maybe_hot_count_p): Use ipa counts only.
7399 (maybe_hot_bb_p): Simplify.
7400 (maybe_hot_edge_p): Simplify.
7401 (probably_never_executed): Do not take frequency argument.
7402 (probably_never_executed_bb_p): Do not pass frequency.
7403 (probably_never_executed_edge_p): Likewise.
7404 (combine_predictions_for_bb): Check that profile is nonzero.
7405 (propagate_freq): Do not set frequency.
7406 (drop_profile): Simplify.
7407 (counts_to_freqs): Simplify.
7408 (expensive_function_p): Use to_frequency.
7409 (propagate_unlikely_bbs_forward): Simplify.
7410 (determine_unlikely_bbs): Simplify.
7411 (estimate_bb_frequencies): Add hack to silence graphite issues.
7412 (compute_function_frequency): Use ipa counts.
7413 (pass_profile::execute): Update.
7414 (rebuild_frequencies): Use counts only.
7415 (force_edge_cold): Use counts only.
7416 * profile-count.c (profile_count::dump): Dump new count types.
7417 (profile_count::differs_from_p): Check compatiblity.
7418 (profile_count::to_frequency): New function.
7419 (profile_count::to_cgraph_frequency): New function.
7420 * profile-count.h (struct function): Declare.
7421 (enum profile_quality): Add profile_guessed_local and
7422 profile_guessed_global0.
7423 (class profile_proability): Decrease number of bits to 29;
7424 update from_reg_br_prob_note and to_reg_br_prob_note.
7425 (class profile_count: Update comment; decrease number of bits
7426 to 61. Check compatibility.
7427 (profile_count::compatible_p): New private member function.
7428 (profile_count::ipa_p): New member function.
7429 (profile_count::operator<): Handle global zero correctly.
7430 (profile_count::operator>): Handle global zero correctly.
7431 (profile_count::operator<=): Handle global zero correctly.
7432 (profile_count::operator>=): Handle global zero correctly.
7433 (profile_count::nonzero_p): New member function.
7434 (profile_count::force_nonzero): New member function.
7435 (profile_count::max): New member function.
7436 (profile_count::apply_scale): Handle IPA scalling.
7437 (profile_count::guessed_local): New member function.
7438 (profile_count::global0): New member function.
7439 (profile_count::ipa): New member function.
7440 (profile_count::to_frequency): Declare.
7441 (profile_count::to_cgraph_frequency): Declare.
7442 * profile.c (OVERLAP_BASE): Delete.
7443 (compute_frequency_overlap): Delete.
7444 (compute_branch_probabilities): Do not use compute_frequency_overlap.
7445 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
7446 * sched-ebb.c (rank): Use counts only.
7447 * shrink-wrap.c (handle_simple_exit): Use counts only.
7448 (try_shrink_wrapping): Use counts only.
7449 (place_prologue_for_one_component): Use counts only.
7450 * tracer.c (find_best_predecessor): Use to_frequency.
7451 (find_trace): Use to_frequency.
7452 (tail_duplicate): Use to_frequency.
7453 * trans-mem.c (expand_transaction): Do not update frequency.
7454 * tree-call-cdce.c: Do not update frequency.
7455 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
7456 (gimple_merge_blocks): Likewise.
7457 (gimple_split_edge): Likewise.
7458 (gimple_duplicate_sese_region): Likewise.
7459 (gimple_duplicate_sese_tail): Likewise.
7460 (move_sese_region_to_fn): Likewise.
7461 (gimple_account_profile_record): Likewise.
7462 (insert_cond_bb): Likewise.
7463 * tree-complex.c (expand_complex_div_wide): Likewise.
7464 * tree-eh.c (lower_resx): Update profile.
7465 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
7467 (initialize_cfun): Do not initialize frequencies
7468 (freqs_to_counts): Delete.
7469 (copy_cfg_body): Ignore count parameter.
7470 (copy_body): Update.
7471 (expand_call_inline): Update count_max.
7472 (optimize_inline_calls): Update count_max.
7473 (tree_function_versioning): Update count_max.
7474 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
7475 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
7477 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
7478 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
7479 (try_peel_loop): Likewise.
7480 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
7482 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
7483 (tree_transform_and_unroll_loop): Do not use frequencies
7484 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
7485 Use reliable prediction only.
7486 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
7487 * tree-ssa-sink.c (select_best_block): Use to_frequency.
7488 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
7489 probability scaling.
7490 * tree-ssa-threadupdate.c (create_block_for_threading): Do
7491 not update frequency
7492 (any_remaining_duplicated_blocks): Likewise.
7493 (update_profile): Likewise.
7494 (estimated_freqs_path): Delete.
7495 (freqs_to_counts_path): Delete.
7496 (clear_counts_path): Delete.
7497 (ssa_fix_duplicate_block_edges): Likewise.
7498 (duplicate_thread_path): Likewise.
7499 * tree-switch-conversion.c (gen_inbound_check): Use counts.
7500 * tree-tailcall.c (decrease_profile): Do not update frequency.
7501 (eliminate_tail_call): Likewise.
7502 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
7503 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
7504 (optimize_mask_stores): Likewise.
7505 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
7506 * ubsan.c (ubsan_expand_null_ifn): Update profile.
7507 (ubsan_expand_ptr_ifn): Update profile.
7508 * value-prof.c (gimple_ic): Simplify.
7509 * value-prof.h (gimple_ic): Update prototype.
7510 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
7511 * ipa-inline.c (compute_uninlined_call_time): Be sure that
7513 (want_inline_self_recursive_call_p): Likewise.
7514 (resolve_noninline_speculation): Only cummulate defined counts.
7515 (inline_small_functions): Use nonzero_p.
7516 (ipa_inline): Do not access freed node.
7518 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
7520 * config/aarch64/aarch64.c (aarch64_override_options_internal):
7521 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
7523 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
7525 * config/riscv/riscv.c (riscv_legitimize_move): Handle
7526 non-legitimate address.
7528 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
7530 * config/rs6000/rs6000.md (*lt0_disi): Delete.
7531 (*lt0_<mode>di, *lt0_<mode>si): New.
7533 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
7535 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
7536 TARGET_PAIRED_FLOAT.
7538 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
7539 Jim Wilson <jim.wilson@linaro.org>
7541 * config/aarch64/aarch64-cores.def (saphira): New CPU.
7542 * config/aarch64/aarch64-tune.md: Regenerated.
7543 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
7544 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
7546 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
7548 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
7549 (arc_expand_prologue): Restore blink for millicode.
7550 * config/arc/linux.h (LINK_EH_SPEC): Defined.
7552 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
7555 * config/i386/i386.c (ix86_vector_duplicate_value): Use
7556 gen_vec_duplicate after forcing the scalar into a register.
7558 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
7560 * combine (try_combine): Print the insns input to try_combine to the
7563 2017-11-02 Steve Ellcey <sellcey@cavium.com>
7566 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
7567 Remove second argument from aarch64_process_target_attr call.
7568 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
7570 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
7572 (aarch64_handle_attr_arch): Remove second argument.
7573 (aarch64_handle_attr_cpu): Ditto.
7574 (aarch64_handle_attr_tune): Ditto.
7575 (aarch64_handle_attr_isa_flags): Ditto.
7576 (aarch64_process_one_target_attr): Ditto.
7577 (aarch64_process_target_attr): Ditto.
7578 (aarch64_option_valid_attribute_p): Remove second argument.
7579 on aarch64_process_target_attr call.
7581 2017-11-02 David Malcolm <dmalcolm@redhat.com>
7583 * diagnostic.c: Include "selftest-diagnostic.h".
7584 (selftest::assert_location_text): New function.
7585 (selftest::test_diagnostic_get_location_text): New function.
7586 (selftest::diagnostic_c_tests): Call it.
7588 2017-11-02 David Malcolm <dmalcolm@redhat.com>
7590 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
7591 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
7592 (class selftest::test_diagnostic_context): Move to...
7593 * selftest-diagnostic.c: New file.
7594 * selftest-diagnostic.h: New file.
7596 2017-11-02 James Bowman <james.bowman@ftdichip.com>
7598 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
7599 offset range for FT32B.
7600 * config/ft32/ft32.h: option "mcompress" enables relaxation.
7601 * config/ft32/ft32.md: Add TARGET_NOPM.
7602 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
7603 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
7605 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
7607 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
7609 2017-11-02 Jeff Law <law@redhat.com>
7611 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
7612 virtual keyword on FINAL OVERRIDE members.
7614 * tree-ssa-propagate.h (ssa_propagation_engine): Group
7615 virtuals together. Add virtual destructor.
7616 (substitute_and_fold_engine): Similarly.
7618 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
7620 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
7622 2017-11-02 Richard Biener <rguenther@suse.de>
7624 PR tree-optimization/82795
7625 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
7627 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7629 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
7631 (gcc_GAS_CHECK_FEATURE): Remove.
7632 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
7633 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
7634 * configure: Regenerate.
7636 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
7638 * config/arc/arc.c (hwloop_optimize): Account for empty
7641 2017-11-02 Richard Biener <rguenther@suse.de>
7644 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
7645 Truncate ARRAY_REF index and element size.
7647 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
7649 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
7651 2017-11-01 Jeff Law <law@redhat.com>
7653 * tree-ssa-ccp.c (ccp_folder): New class derived from
7654 substitute_and_fold_engine.
7655 (ccp_folder::get_value): New member function.
7656 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
7657 (ccp_fold_stmt): Remove prototype.
7658 (ccp_finalize): Call substitute_and_fold from the ccp_class.
7659 * tree-ssa-copy.c (copy_folder): New class derived from
7660 substitute_and_fold_engine.
7661 (copy_folder::get_value): Renamed from get_value.
7662 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
7663 * tree-vrp.c (vrp_folder): New class derived from
7664 substitute_and_fold_engine.
7665 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
7666 (vrp_folder::get_value): New member function.
7667 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
7668 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
7669 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
7670 provide a class interface to folder/substitute routines.
7671 (ssa_prop_fold_stmt_fn): Remove typedef.
7672 (ssa_prop_get_value_fn): Likewise.
7673 (subsitute_and_fold): Remove prototype.
7674 (replace_uses_in): Likewise.
7675 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
7676 Renamed from replace_uses_in. Call the virtual member function
7677 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
7678 (substitute_and_fold_dom_walker): Remove initialization of
7679 data member entries for calbacks. Add substitute_and_fold_engine
7680 member and initialize it.
7681 (substitute_and_fold_dom_walker::before_dom_children0: Use the
7682 member functions for get_value, replace_phi_args_in c
7683 replace_uses_in, and fold_stmt calls.
7684 (substitute_and_fold_engine::substitute_and_fold): Renamed from
7685 substitute_and_fold. Remove assert. Update ctor call.
7687 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
7688 (ssa_prop_visit_phi_fn): Likewise.
7689 (class ssa_propagation_engine): New class to provide an interface
7691 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
7693 (ssa_prop_visit_phi): Likewise.
7694 (ssa_propagation_engine::simulate_stmt): Moved into class.
7695 Call visit_phi/visit_stmt from the class rather than via
7696 file scoped static variables.
7697 (ssa_propagation_engine::simulate_block): Moved into class.
7698 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
7699 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
7700 set file scoped statics for the visit_stmt/visit_phi callbacks.
7701 * tree-complex.c (complex_propagate): New class derived from
7702 ssa_propagation_engine.
7703 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
7704 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
7705 (tree_lower_complex): Call ssa_propagate via the complex_propagate
7707 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
7708 ssa_propagation_engine.
7709 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
7710 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
7711 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
7712 * tree-ssa-copy.c (copy_prop): New class derived from
7713 ssa_propagation_engine.
7714 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
7715 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
7716 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
7717 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
7718 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
7719 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
7720 (execute_vrp): Call ssa_propagate from the vrp_prop class.
7722 2017-11-01 Jakub Jelinek <jakub@redhat.com>
7724 PR rtl-optimization/82778
7725 PR rtl-optimization/82597
7726 * compare-elim.c (struct comparison): Add in_a_setter field.
7727 (find_comparison_dom_walker::before_dom_children): Remove killed
7728 bitmap and df_simulate_find_defs call, instead walk the defs.
7729 Compute last_setter and initialize in_a_setter. Merge definitions
7730 with first initialization for a few variables.
7731 (try_validate_parallel): Use insn_invalid_p instead of
7732 recog_memoized. Return insn rather than just the pattern.
7733 (try_merge_compare): Fix up comment. Don't uselessly test if
7734 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
7736 (execute_compare_elim_after_reload): Remove df_chain_add_problem
7739 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7740 Alan Hayward <alan.hayward@arm.com>
7741 David Sherwood <david.sherwood@arm.com>
7743 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
7744 aarch64_hard_regno_nregs to get the number of registers
7747 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7748 Alan Hayward <alan.hayward@arm.com>
7749 David Sherwood <david.sherwood@arm.com>
7751 * config/aarch64/constraints.md (Upl): Rename to...
7753 * config/aarch64/aarch64.md
7754 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
7757 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7758 Alan Hayward <alan.hayward@arm.com>
7759 David Sherwood <david.sherwood@arm.com>
7761 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
7762 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
7765 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7766 Alan Hayward <alan.hayward@arm.com>
7767 David Sherwood <david.sherwood@arm.com>
7769 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
7770 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
7771 (aarch64_evpc_dup): Generate rtl direcly, rather than using
7773 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
7774 of a single element.
7775 * config/aarch64/iterators.md: Add a comment above the permute
7776 unspecs to say that they are generated directly by
7777 aarch64_expand_vec_perm_const.
7778 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
7780 2017-11-01 Nathan Sidwell <nathan@acm.org>
7782 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
7784 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
7786 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
7787 and medany code models, and describe what they do.
7789 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7791 Revert accidental duplicate:
7793 * combine.c (can_change_dest_mode): Reject changes in
7794 REGMODE_NATURAL_SIZE.
7796 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
7798 PR rtl-optimization/64682
7799 PR rtl-optimization/69567
7800 PR rtl-optimization/69737
7801 PR rtl-optimization/82683
7802 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
7803 register mentioned in the note, drop the note, unless it came from I3,
7804 in which case it should go to I3 again.
7806 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7808 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
7809 and return false if not.
7810 (clear_bytes_written_by, live_bytes_read): Update accordingly.
7812 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7814 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
7815 range is known to be empty.
7817 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7818 Alan Hayward <alan.hayward@arm.com>
7819 David Sherwood <david.sherwood@arm.com>
7821 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
7822 and CONST_VECTOR_NUNITS instead of computing the number of units from
7823 the byte sizes of the vector and element.
7824 (simplify_binary_operation_1): Likewise.
7825 (simplify_const_binary_operation): Likewise.
7826 (simplify_ternary_operation): Likewise.
7828 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7829 Alan Hayward <alan.hayward@arm.com>
7830 David Sherwood <david.sherwood@arm.com>
7832 * var-tracking.c (INT_MEM_OFFSET): Replace with...
7833 (int_mem_offset): ...this new function.
7834 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
7835 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
7836 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
7839 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7840 Alan Hayward <alan.hayward@arm.com>
7841 David Sherwood <david.sherwood@arm.com>
7843 * lower-subreg.c (interesting_mode_p): New function.
7844 (compute_costs, find_decomposable_subregs, decompose_register)
7845 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
7846 (resolve_clobber, dump_choices): Use it.
7848 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7849 Alan Hayward <alan.hayward@arm.com>
7850 David Sherwood <david.sherwood@arm.com>
7852 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
7854 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7855 Alan Hayward <alan.hayward@arm.com>
7856 David Sherwood <david.sherwood@arm.com>
7858 * alias.c (find_base_value, find_base_term): Only process integer
7859 truncations. Check the precision rather than the size.
7861 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7862 Alan Hayward <alan.hayward@arm.com>
7863 David Sherwood <david.sherwood@arm.com>
7865 * machmode.h (is_narrower_int_mode): New function
7866 * optabs.c (expand_float, expand_fix): Use it.
7867 * dwarf2out.c (rotate_loc_descriptor): Likewise.
7869 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7870 Alan Hayward <alan.hayward@arm.com>
7871 David Sherwood <david.sherwood@arm.com>
7873 * rtl.h (narrower_subreg_mode): New function.
7874 * ira-color.c (update_costs_from_allocno): Use it.
7876 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7877 Alan Hayward <alan.hayward@arm.com>
7878 David Sherwood <david.sherwood@arm.com>
7880 * optabs-query.h (convert_optab_p): New function, split out from...
7881 (convert_optab_handler): ...here.
7882 (widening_optab_handler): Delete.
7883 (find_widening_optab_handler): Remove permit_non_widening parameter.
7884 (find_widening_optab_handler_and_mode): Likewise. Provide an
7885 override that operates on mode class wrappers.
7886 * optabs-query.c (widening_optab_handler): Delete.
7887 (find_widening_optab_handler_and_mode): Remove permit_non_widening
7888 parameter. Assert that the two modes are the same class and that
7889 the "from" mode is narrower than the "to" mode. Use
7890 convert_optab_handler instead of widening_optab_handler.
7891 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
7892 instead of widening_optab_handler.
7893 * expr.c (expand_expr_real_2): Update calls to
7894 find_widening_optab_handler.
7895 * optabs.c (expand_widen_pattern_expr): Likewise.
7896 (expand_binop_directly): Take the insn_code as a parameter.
7897 (expand_binop): Only call find_widening_optab_handler for
7898 conversion optabs; use optab_handler otherwise. Update calls
7899 to find_widening_optab_handler and expand_binop_directly.
7900 Use convert_optab_handler instead of widening_optab_handler.
7901 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
7902 find_widening_optab_handler and use scalar_mode rather than
7904 (convert_plusminus_to_widen): Likewise.
7906 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7907 Alan Hayward <alan.hayward@arm.com>
7908 David Sherwood <david.sherwood@arm.com>
7910 * machmode.h (fixed_size_mode): New class.
7911 * rtl.h (get_pool_mode): Return fixed_size_mode.
7912 * gengtype.c (main): Add fixed_size_mode.
7913 * target.def (get_raw_result_mode): Return a fixed_size_mode.
7914 (get_raw_arg_mode): Likewise.
7915 * doc/tm.texi: Regenerate.
7916 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
7917 * targhooks.c (default_get_reg_raw_mode): Likewise.
7918 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
7919 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
7920 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
7921 (msp430_get_raw_result_mode): Likewise.
7922 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
7923 * dbxout.c (dbxout_parms): Require fixed-size modes.
7924 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
7925 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
7926 * omp-low.c (lower_oacc_reductions): Likewise.
7927 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
7928 (simplify_subreg): Update accordingly.
7929 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
7930 (force_const_mem): Update accordingly. Return NULL_RTX for modes
7931 that aren't fixed-size.
7932 (get_pool_mode): Return a fixed_size_mode.
7933 (output_constant_pool_2): Take a fixed_size_mode.
7935 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7936 Alan Hayward <alan.hayward@arm.com>
7937 David Sherwood <david.sherwood@arm.com>
7939 * doc/rtl.texi (vec_series): Document.
7940 (const): Say that the operand can be a vec_series.
7941 * rtl.def (VEC_SERIES): New rtx code.
7942 * rtl.h (const_vec_series_p_1): Declare.
7943 (const_vec_series_p): New function.
7944 * emit-rtl.h (gen_const_vec_series): Declare.
7945 (gen_vec_series): Likewise.
7946 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
7947 (gen_vec_series): Likewise.
7948 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
7949 * simplify-rtx.c (simplify_unary_operation): Handle negations
7951 (simplify_binary_operation_series): New function.
7952 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
7953 (test_vector_ops_series): New function.
7954 (test_vector_ops): Call it.
7955 * config/powerpcspe/altivec.md (altivec_lvsl): Use
7956 gen_const_vec_series.
7957 (altivec_lvsr): Likewise.
7958 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
7960 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7961 Alan Hayward <alan.hayward@arm.com>
7962 David Sherwood <david.sherwood@arm.com>
7964 * doc/rtl.texi (const): Update description of address constants.
7965 Say that vector constants are allowed too.
7966 * common.md (E, F): Use CONSTANT_P instead of checking for
7968 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
7969 checking for CONST_VECTOR.
7970 * expmed.c (make_tree): Use build_vector_from_val for a CONST
7972 * expr.c (expand_expr_real_2): Check for vector modes instead
7973 of checking for CONST_VECTOR.
7974 * rtl.h (const_vec_p): New function.
7975 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
7976 (unwrap_const_vec_duplicate): Handle them here too.
7978 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7979 David Malcolm <dmalcolm@redhat.com>
7980 Alan Hayward <alan.hayward@arm.com>
7981 David Sherwood <david.sherwood@arm.com>
7983 * rtl.h (vec_duplicate_p): New function.
7984 * selftest-rtl.c (assert_rtx_eq_at): New function.
7985 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
7986 (assert_rtx_eq_at): Declare.
7987 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
7988 * selftest-run-tests.c (selftest::run_tests): Call it.
7989 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
7990 (simplify_unary_operation_1): Recursively handle vector duplicates.
7991 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
7993 (simplify_subreg): Handle subregs of vector duplicates.
7994 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
7995 (selftest::simplify_rtx_c_tests): New functions.
7997 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7998 Alan Hayward <alan.hayward@arm.com>
7999 David Sherwood <david.sherwood@arm.com>
8001 * emit-rtl.h (gen_const_vec_duplicate): Declare.
8002 (gen_vec_duplicate): Likewise.
8003 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
8005 (gen_const_vector): ...here.
8006 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
8007 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
8008 whose elements are all equal.
8009 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
8010 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
8011 (simplify_relational_operation): Likewise.
8012 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
8014 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
8015 (aarch64_expand_vector_init): Likewise.
8016 * config/arm/arm.c (neon_vdup_constant): Likewise.
8017 (neon_expand_vector_init): Likewise.
8018 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
8019 (arm_block_set_unaligned_vect): Likewise.
8020 (arm_block_set_aligned_vect): Likewise.
8021 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
8022 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
8023 (expand_vec_perm_even_odd_pack): Likewise.
8024 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
8025 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
8026 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
8027 gen_const_vec_duplicate.
8028 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
8029 * config/mips/mips.c (mips_gen_const_int_vector): Use
8030 gen_const_vec_duplicate.
8031 (mips_expand_vector_init): Use CONST0_RTX.
8032 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
8033 (define_split): Use gen_const_vec_duplicate.
8034 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
8035 (define_split): Use gen_const_vec_duplicate.
8036 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
8037 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
8038 * config/spu/spu.c (spu_const): Likewise.
8040 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8041 Alan Hayward <alan.hayward@arm.com>
8042 David Sherwood <david.sherwood@arm.com>
8044 * combine.c (can_change_dest_mode): Reject changes in
8045 REGMODE_NATURAL_SIZE.
8047 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
8049 * configure.ac (--enable-libssp): New.
8050 (gcc_cv_libc_provides_ssp): Check for explicit setting before
8051 trying to determine target-specific default. Adjust indentation.
8052 * configure: Regenerated.
8053 * doc/install.texi (Configuration): Expand --disable-libssp
8056 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
8058 config/i386/i386.c (ix86_expand_epilogue): Correct stack
8061 2017-10-31 Martin Jambor <mjambor@suse.cz>
8064 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
8066 2017-10-31 David Malcolm <dmalcolm@redhat.com>
8068 * auto-profile.c (autofdo_source_profile::read): Use
8069 UNKNOWN_LOCATION rather than 0.
8070 * diagnostic-core.h (warning_at_rich_loc): Rename to...
8071 (warning_at): ...this overload.
8072 (warning_at_rich_loc_n): Rename to...
8073 (warning_n): ...this overload.
8074 (error_at_rich_loc): Rename to...
8075 (error_at): ...this overload.
8076 (pedwarn_at_rich_loc): Rename to...
8077 (pedwarn): ...this overload.
8078 (permerror_at_rich_loc): Rename to...
8079 (permerror): ...this overload.
8080 (inform_at_rich_loc): Rename to...
8081 (inform): ...this overload.
8082 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
8083 (diagnostic_n_impl_richloc): Rename to...
8084 (diagnostic_n_impl): ...this rich_location *-based decl.
8085 (inform_at_rich_loc): Rename to...
8086 (inform): ...this, and add an assertion.
8087 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
8088 (warning_at_rich_loc): Rename to...
8089 (warning_at): ...this, and add an assertion.
8090 (warning_at_rich_loc_n): Rename to...
8091 (warning_n): ...this, and add an assertion.
8092 (warning_n): Update location_t-based implementation for removal of
8093 location_t-based diagnostic_n_impl.
8094 (pedwarn_at_rich_loc): Rename to...
8095 (pedwarn): ...this, and add an assertion.
8096 (permerror_at_rich_loc): Rename to...
8097 (permerror): ...this, and add an assertion.
8098 (error_n): Update for removal of location_t-based diagnostic_n_impl.
8099 (error_at_rich_loc): Rename to...
8100 (error_at): ...this, and add an assertion.
8101 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
8102 (driver::do_spec_on_infiles): Likewise.
8103 * substring-locations.c (format_warning_va): Update for renaming
8104 of inform_at_rich_loc.
8106 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
8108 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
8109 _Float<N>X built-in functions so that the variant without the
8110 "__builtin_" prefix is only enabled for the GNU C and Objective C
8111 languages when they are in non-strict ANSI/ISO mode.
8112 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
8113 * target.def (floatn_builtin_p): Add a target hook to control
8114 whether _Float<N> and _Float<N>X built-in functions without the
8115 "__builtin_" prefix are enabled, and return true for C and
8116 Objective C in the default hook. Include langhooks.h in
8118 * targhooks.h (default_floatn_builtin_p): Likewise.
8119 * targhooks.c (default_floatn_builtin_p): Likewise.
8120 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
8121 floatn_builtin_p target hook.
8122 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
8124 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
8125 Eric Botcazou <ebotcazou@adacore.com>
8127 PR rtl-optimization/81803
8128 * lra-constraints.c (curr_insn_transform): Also reload the whole
8129 register for a strict subreg no wider than a word if this is for
8130 a WORD_REGISTER_OPERATIONS target.
8132 2017-10-31 Jason Merrill <jason@redhat.com>
8134 * gdbinit.in: Skip over inlines from timevar.h.
8136 2017-10-31 Martin Liska <mliska@suse.cz>
8138 * doc/gcov.texi: Document new option.
8139 * gcov.c (print_usage): Likewise print it.
8140 (process_args): Support the argument.
8141 (format_count): New function.
8142 (format_gcov): Use the function.
8144 2017-10-31 Martin Liska <mliska@suse.cz>
8146 * gcov.c (struct name_map): do not use typedef.
8147 Define operator== and operator<.
8148 (name_search): Remove.
8149 (name_sort): Remove.
8150 (main): Do not allocate names.
8151 (process_file): Add vertical space.
8152 (generate_results): Use std::find.
8153 (release_structures): Do not release memory.
8154 (find_source): Use std::find.
8156 2017-10-31 Martin Liska <mliska@suse.cz>
8158 * gcov.c (struct line_info): Remove it's typedef.
8159 (line_info::line_info): Add proper ctor.
8160 (line_info::has_block): Do not use a typedef.
8161 (struct source_info): Do not use typedef.
8162 (circuit): Likewise.
8163 (get_cycles_count): Likewise.
8164 (output_intermediate_file): Iterate via vector iterator.
8165 (add_line_counts): Use std::vector methods.
8166 (accumulate_line_counts): Likewise.
8167 (output_lines): Likewise.
8169 2017-10-31 Martin Liska <mliska@suse.cz>
8171 * gcov.c (struct source_info): Remove typedef.
8172 (source_info::source_info): Add proper ctor.
8173 (accumulate_line_counts): Use struct, not it's typedef.
8174 (output_gcov_file): Likewise.
8175 (output_lines): Likewise.
8176 (main): Do not allocate an array.
8177 (output_intermediate_file): Use size of vector container.
8178 (process_file): Resize the vector.
8179 (generate_results): Do not preallocate, use newly added vector
8181 (release_structures): Do not release sources.
8182 (find_source): Use vector methods.
8183 (add_line_counts): Do not use typedef.
8185 2017-10-31 Martin Liska <mliska@suse.cz>
8187 * doc/gcov.texi: Document that.
8188 * gcov.c (add_line_counts): Mark lines with a non-executed
8190 (output_line_beginning): Handle such lines.
8191 (output_lines): Pass new argument.
8192 (output_intermediate_file): Print it in intermediate format.
8194 2017-10-31 Martin Liska <mliska@suse.cz>
8196 * color-macros.h: New file.
8197 * diagnostic-color.c: Factor out color related to macros to
8199 * doc/gcov.texi: Document -k option.
8200 * gcov.c (INCLUDE_STRING): Include string.h.
8201 (print_usage): Add -k option.
8202 (process_args): Parse it.
8203 (pad_count_string): New function.
8204 (output_line_beginning): Likewise.
8205 (DEFAULT_LINE_START): New macro.
8206 (output_lines): Support color output.
8208 2017-10-31 Martin Liska <mliska@suse.cz>
8210 PR gcov-profile/82633
8211 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
8212 their interaction with GCOV infrastructure.
8213 * configure.ac: Add -fkeep-{inline,static}-functions to
8215 * configure: Regenerate.
8217 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
8220 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
8222 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
8225 * config/rs6000/rs6000.md (allocate_stack): Force update interval
8226 into a register if it does not fit into an immediate offset field.
8228 2017-10-31 Olivier Hainque <hainque@adacore.com>
8230 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
8232 2017-10-31 Julia Koval <julia.koval@intel.com>
8234 * config.gcc: Add gfniintrin.h.
8235 * config/i386/gfniintrin.h: New.
8236 * config/i386/i386-builtin-types.def
8237 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
8238 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
8239 __builtin_ia32_vgf2p8affineinvqb_v32qi,
8240 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
8241 __builtin_ia32_vgf2p8affineinvqb_v16qi,
8242 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
8243 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
8244 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
8245 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
8246 V64QI_FTYPE_V64QI_V64QI_INT): New types.
8247 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
8248 * config/i386/immintrin.h: Include gfniintrin.h.
8249 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
8251 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
8253 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
8255 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
8257 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
8258 (arm_ashldi3_1bit): Remove pattern.
8259 (ashrdi3): Remove shift by 1 expansion.
8260 (arm_ashrdi3_1bit): Remove pattern.
8261 (lshrdi3): Remove shift by 1 expansion.
8262 (arm_lshrdi3_1bit): Remove pattern.
8263 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
8264 cost of ashldi3 by 1.
8265 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
8266 (<shift>di3_neon): Likewise.
8268 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
8270 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
8271 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
8272 and (*aarch64_simd_mov<VQ:mode>).
8273 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
8274 pattern alternative.
8275 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
8276 attributes to match pattern alternative.
8278 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
8280 * config.gcc (powerpc*-*-*): Add emmintrin.h.
8281 * config/rs6000/emmintrin.h: New file.
8282 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
8284 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
8286 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
8287 * (movdi_vfp_cortexa8): Remove pattern.
8289 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8291 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
8293 (Specific, alpha*-dec-osf5.1): Remove.
8294 (Specific, mips-sgi-irix5): Remove.
8295 (Specific, mips-sgi-irix6): Remove.
8297 2017-10-30 Jakub Jelinek <jakub@redhat.com>
8300 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
8301 arguments to clear_bit_region_be.
8303 2017-10-30 Jim Wilson <wilson@tuliptree.org>
8305 * gimplify.c: Include memmodel.h.
8307 2017-10-30 Martin Jambor <mjambor@suse.cz>
8309 * omp-grid.c (grid_attempt_target_gridification): Also insert a
8310 condition whether loop should be executed at all.
8312 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
8314 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
8315 gimple folding of vec_madd() intrinsics.
8316 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
8317 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
8318 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
8320 2017-10-30 Richard Biener <rguenther@suse.de>
8322 PR tree-optimization/82762
8324 2017-10-23 Richard Biener <rguenther@suse.de>
8326 PR tree-optimization/82129
8328 2017-08-01 Richard Biener <rguenther@suse.de>
8330 PR tree-optimization/81181
8331 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
8332 (compute_antic): ... end of iteration here.
8334 2017-10-30 Joseph Myers <joseph@codesourcery.com>
8336 * doc/invoke.texi (C Dialect Options): Document -std=c17,
8337 -std=iso9899:2017 and -std=gnu17.
8338 * doc/standards.texi (C Language): Document C17 support.
8339 * doc/cpp.texi (Overview): Mention -std=c17.
8340 (Standard Predefined Macros): Document C11 and C17 values of
8341 __STDC_VERSION__. Do not refer to C99 support as incomplete.
8342 * doc/extend.texi (Inline): Do not list individual options for
8343 standards newer than C99.
8344 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
8346 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
8349 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
8351 * asan.c (asan_finish_file): Align asan globals array by shadow
8354 2017-10-30 Jakub Jelinek <jakub@redhat.com>
8357 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
8358 (struct store_immediate_info): Add bitregion_start and bitregion_end
8360 (store_immediate_info::store_immediate_info): Add brs and bre
8361 arguments and initialize bitregion_{start,end} from those.
8362 (struct merged_store_group): Add bitregion_start, bitregion_end,
8363 align_base and mask fields. Drop unnecessary struct keyword from
8364 struct store_immediate_info. Add do_merge method.
8365 (clear_bit_region_be): Use memset instead of loop storing zeros.
8366 (merged_store_group::do_merge): New method.
8367 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
8368 stores as long as the surrounding bitregions have no gaps.
8369 (merged_store_group::merge_overlapping): Use do_merge.
8370 (merged_store_group::apply_stores): Test that bitregion_{start,end}
8371 is byte aligned, rather than requiring that start and width are
8372 byte aligned. Drop unnecessary struct keyword from
8373 struct store_immediate_info. Allocate and populate also mask array.
8374 Make start of the arrays relative to bitregion_start rather than
8375 start and size them according to bitregion_{end,start} difference.
8376 (struct imm_store_chain_info): Drop unnecessary struct keyword from
8377 struct store_immediate_info.
8378 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
8379 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
8380 struct keyword from struct store_immediate_info.
8381 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
8382 between stores as long as the surrounding bitregions have no gaps.
8384 (struct split_store): Add orig non-static data member.
8385 (split_store::split_store): Initialize orig to false.
8386 (find_constituent_stmts): Return store_immediate_info *, non-NULL
8387 if there is exactly a single original stmt. Change stmts argument
8388 to pointer from reference, if NULL, don't push anything to it. Add
8389 first argument, use it to optimize skipping over orig stmts that
8390 are known to be before bitpos already. Simplify.
8391 (split_group): Return unsigned int count how many stores are or
8392 would be needed rather than a bool. Add allow_unaligned argument.
8393 Change split_stores argument from reference to pointer, if NULL,
8394 only do a dry run computing how many stores would be produced.
8395 Rewritten algorithm to use both alignment and misalign if
8396 !allow_unaligned and handle bitfield stores with gaps.
8397 (imm_store_chain_info::output_merged_store): Set start_byte_pos
8398 from bitregion_start instead of start. Compute allow_unaligned
8399 here, if true, do 2 split_group dry runs to compute which one
8400 produces fewer stores and prefer aligned if equal. Punt if
8401 new count is bigger or equal than original before emitting any
8402 statements, rather than during that. Remove no longer needed
8403 new_ssa_names tracking. Replace num_stmts with
8404 split_stores.length (). Use 32-bit stack allocated entries
8405 in split_stores auto_vec. Try to reuse original store lhs/rhs1
8406 if possible. Handle bitfields with gaps.
8407 (pass_store_merging::execute): Ignore bitsize == 0 stores.
8408 Compute bitregion_{start,end} for the stores and construct
8409 store_immediate_info with that. Formatting fixes.
8411 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
8414 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
8415 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
8417 2017-10-29 Jim Wilson <wilson@tuliptree.org>
8419 * gimplify.c: Include tm_p.h.
8421 * common.opt (gcoff): Re-add as ignored option.
8422 (gcoff1, gcoff2, gcoff3): Likewise.
8424 * Makefile.in (OBJS): Delete sdbout.o.
8425 (GTFILES): Delete $(srcdir)/sdbout.c.
8426 * debug.h: Delete sdb_debug_hooks.
8427 * final.c: Delete sdbout.h include.
8428 (final_scan_insn): Delete SDB_DEBUG check.
8429 (rest_of_clean_state): Likewise.
8430 * output.h: Delete sdb_begin_function_line.
8433 * toplev.c: Delete sdbout.h include.
8434 (process_options): Delete SDB_DEBUG check.
8435 * tree-core.h (tree_type_common): Delete pointer field of
8437 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
8438 TYPE_SYMTAB_POINTER.
8439 * tree.h (TYPE_SYMTAB_POINTER): Delete.
8440 (TYPE_SYMTAB_IS_POINTER): Delete.
8441 (TYPE_SYMTAB_IS_DIE): Renumber.
8442 * xcoffout.c: Refer to former sdbout.c file.
8443 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
8445 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
8446 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
8447 (Debugging Options): Delete -gcoff.
8448 (-gstabs): Delete SDB reference.
8450 (-gcoff@var{level}): Delete.
8451 * doc/passes.texi (Debugging information output): Delete SDB and
8452 sdbout.c references.
8453 * doc/tm.texi: Regenerate.
8454 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
8455 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
8457 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
8458 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
8459 (SDB_DEBUGGING_INFO): Delete.
8460 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
8461 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
8462 * target.def (output_source_filename): Delete COFF reference.
8464 * common.opt (gcoff): Delete.
8465 (gxcoff+): Update Negative chain.
8466 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
8468 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
8469 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
8470 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
8472 (expand_function_start): Change sdb reference to past tense.
8473 (expand_function_end): Change sdb reference to past tense.
8474 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
8475 * opts.c (debug_type_names): Delete coff entry.
8476 (common_handle_option): Delete OPT_gcoff case.
8477 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
8479 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
8480 * config/cris/cris.h: Delete SDB reference in comment.
8481 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
8482 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
8483 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
8484 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
8486 (ix86_expand_prologue): Likewise.
8487 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
8488 * config/ia64/ia64.h: Likewise.
8489 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
8490 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
8492 * config/mmix/mmix.h: Likewise.
8493 * config/nds32/nds32.c: Likewise.
8494 * config/stormy/storym16.h: Likewise.
8495 * config/visium/visium.h: Likewise.
8496 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
8498 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
8500 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
8501 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
8502 FRAME_POINTER_REGNUM point at high end of local var area.
8504 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
8506 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
8507 Move comment around. Do not reset best_edge for a copiable
8508 destination if the copy would cause a partition change.
8509 (better_edge_p): Remove redundant check.
8511 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
8513 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
8515 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
8517 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
8518 for math functions that have _Float<N> and _Float<N>X variants.
8519 (mathfn_built_in_2): Add support for math functions that have
8520 _Float<N> and _Float<N>X variants.
8521 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
8522 (expand_builtin_mathfn_ternary): Add support for fma with
8523 _Float<N> and _Float<N>X variants.
8524 (expand_builtin): Likewise.
8525 (fold_builtin_3): Likewise.
8526 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
8527 create math function _Float<N> and _Float<N>X variants as external
8529 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
8530 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
8531 the __builtin_ prefix and if not strict ansi, without the prefix.
8532 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
8533 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
8534 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
8535 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
8536 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
8537 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
8538 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
8539 function signatures for fma _Float<N> and _Float<N>X variants.
8540 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
8541 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
8542 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
8543 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
8544 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
8545 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
8546 * gencfn-macros.c (print_case_cfn): Add support for math functions
8547 that have _Float<N> and _Float<N>X variants.
8548 (print_define_operator_list): Likewise.
8549 (fltfn_suffixes): Likewise.
8551 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
8552 for math functions that have _Float<N> and _Float<N>X variants.
8553 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
8554 and _Float<N>X variants.
8555 (COPYSIGN): Likewise.
8558 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
8559 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
8561 (integer_valued_read_call_p): Likewise.
8562 * fold-const-call.c (fold_const_call_ss): Likewise.
8563 (fold_const_call_sss): Add support for copysign, fmin, and fmax
8564 _Float<N> and _Float<N>X variants.
8565 (fold_const_call_ssss): Add support for fma _Float<N> and
8566 _Float<N>X variants.
8567 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
8568 support for copysign and fma _Float<N> and _Float<N>X variants.
8569 (backprop::process_builtin_call_use): Likewise.
8570 * tree-call-cdce.c (can_test_argument_range); Add support for
8571 sqrt _Float<N> and _Float<N>X variants.
8572 (edom_only_function): Likewise.
8573 (get_no_error_domain): Likewise.
8574 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
8575 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
8576 copysign _Float<N> and _Float<N>X variants.
8577 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
8578 handled by machine independent code.
8579 (FMAF128): Likewise.
8580 * doc/cpp.texi (Common Predefined Macros): Document defining
8581 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
8582 fma _Float<N> and _Float<N>X variants.
8584 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
8587 * config/i386/i386-modes.def (CCFPU): Remove definition.
8588 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
8589 (ix86_cc_modes_compatible): Ditto.
8590 (ix86_expand_carry_flag_compare): Ditto.
8591 (ix86_expand_int_movcc): Ditto.
8592 (ix86_expand_int_addcc): Ditto.
8593 (ix86_reverse_condition): Ditto.
8594 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
8595 Return true/false for unordered/ordered fp comparisons.
8596 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
8597 (ix86_prepare_fp_compare_args): Update for rename.
8598 (ix86_expand_fp_compare): Update for rename. Generate unordered
8599 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
8600 (ix86_expand_sse_compare_and_jump): Ditto.
8601 * config/i386/predicates.md (fcmov_comparison_operator):
8602 Remove CCFPU mode handling.
8603 (ix86_comparison_operator): Ditto.
8604 (ix86_carry_flag_operator): Ditto.
8605 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
8606 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
8607 (*cmpu<mode>_cc_i387): Ditto.
8608 (FPCMP): Remove mode iterator.
8609 (unord): Remove mode attribute.
8610 (unord_subst): New define_subst transformation
8611 (unord): New define_subst attribute.
8613 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
8614 (*cmpi<unord>xf_i387): Ditto.
8615 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
8616 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
8617 using unord_subst transformation.
8618 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
8619 (round_saeonly): Also handle CCFP mode.
8620 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
8621 Remove UNSPEC_SAHF unspec handling.
8623 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
8625 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
8627 2017-10-27 Jeff Law <law@redhat.com>
8629 * gimple-ssa-sprintf.c: Include domwalk.h.
8630 (class sprintf_dom_walker): New class, derived from dom_walker.
8631 (sprintf_dom_walker::before_dom_children): New function.
8632 (struct call_info): Moved into sprintf_dom_walker class
8633 (compute_formath_length, handle_gimple_call): Likewise.
8634 (sprintf_length::execute): Call the dominator walker rather
8635 than walking the statements.
8637 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
8638 gimple statement locations.
8639 (check_array_bounds): Corresponding changes. Get the statement's
8640 location directly from wi->stmt.
8642 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
8645 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
8647 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
8649 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
8650 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
8652 2017-10-27 Jakub Jelinek <jakub@redhat.com>
8655 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
8656 * config/i386/i386.c (maybe_get_pool_constant): Removed.
8657 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
8658 maybe_get_pool_constant.
8659 * config/i386/predicates.md (zero_extended_scalar_load_operand):
8662 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8664 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
8665 2.26 caveat. Update gas and gld versions.
8666 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
8669 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
8671 * cgraph.h (set_malloc_flag): Declare.
8672 * cgraph.c (set_malloc_flag_1): New function.
8673 (set_malloc_flag): Likewise.
8674 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
8675 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
8677 (read_ipa_call_summary): Add support for reading is_return_callee.
8678 (write_ipa_call_summary): Stream is_return_callee.
8679 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
8680 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
8681 ipa-prop.h, ipa-fnsummary.h.
8682 (pure_const_names): Change to static.
8683 (malloc_state_e): Define.
8684 (malloc_state_names): Define.
8685 (funct_state_d): Add field malloc_state.
8686 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
8687 (check_retval_uses): New function.
8688 (malloc_candidate_p): Likewise.
8689 (analyze_function): Add support for malloc attribute.
8690 (pure_const_write_summary): Stream malloc_state.
8691 (pure_const_read_summary): Add support for reading malloc_state.
8692 (dump_malloc_lattice): New function.
8693 (propagate_malloc): New function.
8694 (warn_function_malloc): New function.
8695 (ipa_pure_const::execute): Call propagate_malloc and
8696 ipa_free_fn_summary.
8697 (pass_local_pure_const::execute): Add support for malloc attribute.
8698 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
8699 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
8701 2017-10-27 Martin Liska <mliska@suse.cz>
8703 PR gcov-profile/82457
8704 * doc/invoke.texi: Document that one needs a non-strict ISO mode
8705 for fork-like functions to be properly instrumented.
8707 2017-10-27 Richard Biener <rguenther@suse.de>
8710 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
8711 info when we redirected EH.
8713 2017-10-26 Michael Collison <michael.collison@arm.com>
8715 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
8717 (<optab>_trunchf<GPI:mode>2: New pattern.
8718 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
8719 * config/aarch64/iterators.md (wv): New mode attribute.
8720 (vf, VF): New mode attributes.
8721 (vgp, VGP): New mode attributes.
8722 (s): Update attribute with SImode and DImode prefixes.
8724 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
8726 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
8727 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
8728 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
8729 (nios2_option_overide): Initialize it. Don't allow R0-relative
8730 addressing with PIC.
8731 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
8732 (nios2_symbolic_constant_p): Likewise.
8733 (nios2_legitimate_address_p): Likewise.
8734 (nios2_r0rel_section_name_p): New.
8735 (nios2_symbol_ref_in_r0rel_data_p): New.
8736 (nios2_emit_move_sequence): Handle r0rel_constant_p.
8737 (r0rel_constant_p): New.
8738 (nios2_print_operand_address): Handle r0rel_constant_p.
8739 (nios2_cdx_narrow_form_p): Likewise.
8740 * config/nios2/nios2.opt (mr0rel-sec=): New option.
8741 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
8742 (Nios II Options): Document -mr0rel-sec.
8744 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
8746 * config/nios2/nios2.c: Include xregex.h.
8747 (nios2_gprel_sec_regex): New.
8748 (nios2_option_overide): Initialize it. Don't allow GP-relative
8749 addressing with PIC.
8750 (nios2_small_section_name_p): Check for regex match.
8751 * config/nios2/nios2.opt (mgprel-sec=): New option.
8752 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
8753 (Nios II Options): Document -mgprel-sec.
8755 2017-10-26 Jim Wilson <wilson@tuliptree.org>
8757 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
8759 2017-10-26 Tom de Vries <tom@codesourcery.com>
8761 PR tree-optimization/82707
8762 * gimple.c (gimple_copy): Fix unsharing of
8763 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
8765 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
8767 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
8768 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
8769 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
8770 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
8771 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
8772 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
8773 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
8774 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
8775 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
8776 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
8777 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
8778 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
8779 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
8780 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
8781 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
8782 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
8783 _mm512_mask_cmpunord_ps_mask): New intrinsics.
8785 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
8787 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
8789 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
8790 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
8791 warning to rs6000.c. Remove the Undocumented flag, since it has
8793 (-mabi=ibmlongdouble): Likewise.
8794 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
8795 already set, set the default format for long double.
8796 (rs6000_debug_reg_global): Print whether long double is IBM or
8798 (rs6000_option_override_internal): Rework setting long double
8799 format. Only warn if the user is changing the long double default
8800 and they did not use -Wno-psabi.
8801 * doc/invoke.texi (PowerPC options): Update the documentation for
8802 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
8804 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
8805 Alan Hayward <alan.hayward@arm.com>
8806 David Sherwood <david.sherwood@arm.com>
8808 * rtl.h (wider_subreg_mode): New function.
8809 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
8810 rather than an unsigned int *.
8811 * ira-color.c (regno_max_ref_width): Replace with...
8812 (regno_max_ref_mode): ...this new variable.
8813 (coalesced_pseudo_reg_slot_compare): Update accordingly.
8814 Use wider_subreg_mode.
8815 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
8816 rather than an unsigned int *.
8817 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
8818 (process_alt_operands): Likewise.
8819 (invariant_p): Likewise.
8820 * lra-spills.c (assign_mem_slot): Likewise.
8821 (add_pseudo_to_slot): Likewise.
8822 * lra.c (collect_non_operand_hard_regs): Likewise.
8823 (add_regs_to_insn_regno_info): Likewise.
8824 * reload1.c (regno_max_ref_width): Replace with...
8825 (regno_max_ref_mode): ...this new variable.
8826 (reload): Update accordingly. Update call to
8827 ira_sort_regnos_for_alter_reg.
8828 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
8829 (init_eliminable_invariants): Update to use regno_max_ref_mode.
8830 (scan_paradoxical_subregs): Likewise.
8832 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
8834 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
8835 (aarch64_frame): Add emit_frame_chain boolean.
8836 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
8837 Move eh_return case to aarch64_layout_frame.
8838 (aarch64_layout_frame): Initialize emit_frame_chain.
8839 (aarch64_expand_prologue): Use emit_frame_chain.
8841 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
8843 * config/aarch64/aarch64.c (aarch64_layout_frame):
8844 Ensure LR is always stored at the bottom of the callee-saves.
8845 Remove rarely used frame layout which saves callee-saves at top of
8846 frame, so the store of LR can be used as a valid probe in all cases.
8848 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
8850 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
8851 Improve unaligned TImode/TFmode base/offset split.
8853 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
8854 Alan Hayward <alan.hayward@arm.com>
8855 David Sherwood <david.sherwood@arm.com>
8857 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
8858 * combine.c (find_single_use_1): Likewise.
8859 (expand_field_assignment): Likewise.
8860 (move_deaths): Likewise.
8861 * lra-constraints.c (simplify_operand_subreg): Likewise.
8862 (curr_insn_transform): Likewise.
8863 * lra.c (collect_non_operand_hard_regs): Likewise.
8864 (add_regs_to_insn_regno_info): Likewise.
8865 * rtlanal.c (reg_referenced_p): Likewise.
8866 (covers_regno_no_parallel_p): Likewise.
8868 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
8870 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
8871 Don't print any bits outside the precision of the value.
8872 * wide-int.cc (test_printing): Add some new tests.
8874 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8876 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
8877 supports -xbrace_comment option.
8878 * configure: Regenerate.
8879 * config.in: Regenerate.
8880 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
8881 (ASM_CPU_SPEC): Use it.
8883 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
8885 * target.def (static_rtx_alignment): New hook.
8886 * targhooks.h (default_static_rtx_alignment): Declare.
8887 * targhooks.c (default_static_rtx_alignment): New function.
8888 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
8889 * doc/tm.texi: Regenerate.
8890 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
8891 instead of targetm.constant_alignment. Remove call to
8893 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
8894 (cris_preferred_mininum_alignment): New function, split out from...
8895 (cris_constant_alignment): ...here.
8896 (cris_static_rtx_alignment): New function.
8897 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
8899 (ix86_constant_alignment): ...here.
8900 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
8901 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
8902 (mmix_static_rtx_alignment): New function.
8903 * config/spu/spu.c (spu_static_rtx_alignment): New function.
8904 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
8906 2017-10-26 Tamar Christina <tamar.christina@arm.com>
8909 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
8910 Add flag_trapping_math and flag_fp_int_builtin_inexact.
8912 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
8914 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
8915 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
8916 mark as a sign-extending load.
8917 (local_pic_load_u): Define.
8919 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
8922 * fold-const.c (operand_equal_for_comparison_p): Also return true
8923 if ARG0 is a simple variant of ARG1 with narrower precision.
8924 (fold_ternary_loc): Always pass unstripped operands to the predicate.
8926 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
8928 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
8930 * i386.h (processor_costs): Add gather_static, gather_per_elt,
8931 scatter_static, scatter_per_elt.
8932 * x86-tune-costs.h: Add new cost entries.
8934 2017-10-25 Richard Biener <rguenther@suse.de>
8936 * tree-ssa-sccvn.h (vn_eliminate): Declare.
8937 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
8938 class pass_fre): Move to ...
8939 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
8940 class pass_fre): ... here and adjust for statistics.
8942 2017-10-25 Jakub Jelinek <jakub@redhat.com>
8945 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
8946 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
8947 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
8948 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
8951 2017-10-25 Richard Biener <rguenther@suse.de>
8953 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
8954 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
8955 eliminate_push_avail, eliminate_insert): Move inside...
8956 (class eliminate_dom_walker): ... this class in preparation
8958 (fini_eliminate): Remove by merging with ...
8959 (eliminate): ... this function. Adjust for class changes.
8960 (pass_pre::execute): Remove fini_eliminate call.
8961 (pass_fre::execute): Likewise.
8963 2017-10-24 Jakub Jelinek <jakub@redhat.com>
8966 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
8967 (VPERMI2, VPERMI2I): New mode iterators.
8968 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
8969 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
8971 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
8972 mode iterator. Remove 3 old define_insn patterns.
8973 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
8974 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
8975 VPERMI2 mode iterator, remove the other two expanders.
8976 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
8977 to use VPERMI2 mode iterator, add another alternative for vpermi2*
8978 instructions, remove the other two patterns.
8979 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
8980 mode iterator, remove the other two patterns.
8981 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
8982 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
8983 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
8984 and adjust argument order accordingly.
8985 (ix86_expand_vec_perm): Adjust caller.
8986 (expand_vec_perm_1): Likewise.
8987 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
8988 (expand_vec_perm_vpermt2_vpshub2): ... this.
8989 (ix86_expand_vec_perm_const_1): Adjust caller.
8990 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
8993 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
8994 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
8995 (vec_shl_<mode>): Remove unused expander.
8996 (avx512bw_<shift_insn><mode>3): New define_insn.
8997 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
8998 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
9000 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
9003 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
9006 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
9008 PR rtl-optimization/82396
9009 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
9010 (autopref_multipass_init): Simplify initialization.
9011 (autopref_rank_data): Simplify sort order.
9012 * gcc/sched-int.h (autopref_multipass_data_): Remove
9013 multi_mem_insn_p, min_offset and max_offset.
9015 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
9018 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
9019 Check special value of flag_omit_frame_pointer.
9020 (aarch64_can_eliminate): Likewise.
9021 (aarch64_override_options_after_change_1): Simplify handling of
9022 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
9024 2017-10-24 Richard Biener <rguenther@suse.de>
9026 PR tree-optimization/82697
9027 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
9028 zero for conditional load and unconditional store.
9030 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
9032 * doc/install.texi: Document bootstrap-cet.
9034 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
9037 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
9038 ENDBR instruction at function entrance if function is only
9041 2017-10-24 Jakub Jelinek <jakub@redhat.com>
9044 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
9045 patterns to better describe from which operation the CF is computed.
9046 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
9047 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
9048 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
9049 is 0, use _0 suffixed expanders instead of emitting a comparison
9052 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
9054 * config/i386/i386.md(*movsf_internal, *movdf_internal):
9055 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
9057 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
9060 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
9061 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
9062 * loop-iv.c (iv_get_reaching_def): Likewise.
9063 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
9064 variable is promoted and the partition contains undefined values.
9066 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9068 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
9070 (nios2_address_cost): Define.
9071 (nios2_legitimize_address): Recognize (exp + constant) directly.
9072 (TARGET_ADDRESS_COST): Define.
9074 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9076 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
9077 (nios2_symbolic_memory_operand_p): Declare.
9078 (nios2_split_large_constant): Declare.
9079 (nios2_split_symbolic_memory_operand): Declare.
9080 * config/nios2/nios2.c: Adjust includes.
9081 (nios2_symbolic_constant_allowed): New.
9082 (nios2_symbolic_constant_p): New.
9083 (nios2_plus_symbolic_constant_p): New.
9084 (nios2_valid_addr_expr_p): Recognize addresses involving
9086 (nios2_legitimate_address_p): Likewise, also LO_SUM.
9087 (nios2_symbolic_memory_operand_p): New.
9088 (nios2_large_constant_p): New.
9089 (nios2_split_large_constant): New.
9090 (nios2_split_plus_large_constant): New.
9091 (nios2_split_symbolic_memory_operand): New.
9092 (nios2_legitimize_address): Code refactoring. Handle addresses
9093 involving symbolic constants.
9094 (nios2_emit_move_sequence): Likewise.
9095 (nios2_print_operand): Improve error output.
9096 (nios2_print_operand_address): Handle LO_SUM.
9097 (nios2_cdx_narrow_form_p): Likewise.
9098 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
9099 operands involving symbolic constants.
9100 (movhi_internal, movsi_internal): Likewise.
9101 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
9102 (extendhisi2, extendqi<mode>2): Likewise.
9104 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9106 * tree-pass.h (PROP_rtl_split_insns): Define.
9107 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
9109 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9111 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
9113 2017-10-23 Jakub Jelinek <jakub@redhat.com>
9116 * target.def (const_not_ok_for_debug_p): Default to
9117 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
9118 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
9119 * targhooks.c (default_const_not_ok_for_debug_p): New function.
9120 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
9121 which targetm.const_not_ok_for_debug_p returned true.
9122 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
9124 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
9126 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
9127 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
9128 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
9130 (ix86_const_not_ok_for_debug_p): New function.
9131 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
9132 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
9134 2017-10-23 David Malcolm <dmalcolm@redhat.com>
9137 * system.h: Conditionally include "unique-ptr.h" if
9138 INCLUDE_UNIQUE_PTR is defined.
9139 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
9140 of defining INCLUDE_UNIQUE_PTR before including "system.h".
9142 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
9144 * config/rl78/rl78.md: New define_expand "subdi3".
9146 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
9149 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
9150 DF_REF_INSN if DF_REF_INSN_INFO is false.
9152 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
9154 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
9155 xmm_move instead of sse_move.
9156 (sse_store_index): New function.
9157 (ix86_register_move_cost): Be more sensible about mismatch stall;
9158 model AVX moves correctly; make difference between sse->integer and
9160 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
9161 moves; make difference between SSE and AVX.
9162 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
9163 and zmm_move. Increase size of sse load and store tables;
9164 add unaligned load and store tables; add ssemmx_to_integer.
9165 * x86-tune-costs.h: Update all entries according to real
9166 move latencies from Agner Fog's manual and chip documentation.
9168 2017-10-23 Jakub Jelinek <jakub@redhat.com>
9171 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
9172 * config/i386/constraints.md (Wf): New constraint.
9173 * config/i386/i386.md (UNSPEC_SBB): New unspec.
9174 (cmp<dwi>_doubleword): Removed.
9175 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
9176 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
9177 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
9178 expand with cmp<dwi>_doubleword. For LTU and GEU use
9179 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
9181 * common.opt (gcolumn-info): Enable by default.
9182 * doc/invoke.texi (gcolumn-info): Document new default.
9184 2017-10-23 Richard Biener <rguenther@suse.de>
9186 PR tree-optimization/82672
9187 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
9188 Fold the stmt if we propagated into it.
9190 2017-10-23 Richard Biener <rguenther@suse.de>
9192 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
9193 (bitmap_remove_expr_from_set): ... this. All callers call this
9194 for non-constant values.
9195 (bitmap_set_subtract): Rename to...
9196 (bitmap_set_subtract_expressions): ... this. Adjust and
9198 (bitmap_set_contains_value): Remove superfluous check.
9199 (bitmap_set_replace_value): Inline into single caller ...
9200 (bitmap_value_replace_in_set): ... here and simplify.
9201 (dependent_clean): Merge into ...
9202 (clean): ... this using an overload. Adjust.
9203 (prune_clobbered_mems): Adjust.
9204 (compute_antic_aux): Likewise.
9205 (compute_partial_antic_aux): Likewise.
9207 2017-10-23 Richard Biener <rguenther@suse.de>
9209 PR tree-optimization/82129
9211 2017-08-01 Richard Biener <rguenther@suse.de>
9213 PR tree-optimization/81181
9214 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
9215 (compute_antic): ... end of iteration here.
9217 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9219 * target.def (starting_frame_offset): New hook.
9220 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
9221 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
9222 * doc/tm.texi.in: Regenerate.
9223 * hooks.h (hook_hwi_void_0): Declare.
9224 * hooks.c (hook_hwi_void_0): New function.
9225 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
9226 STARTING_FRAME_OFFSET.
9227 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
9228 * reload1.c (reload): Likewise.
9229 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
9230 instead of STARTING_FRAME_OFFSET.
9231 * function.c (try_fit_stack_local): Likewise.
9232 (assign_stack_local_1): Likewise
9233 (instantiate_virtual_regs): Likewise.
9234 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
9235 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
9236 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
9237 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
9238 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
9239 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
9240 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
9241 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
9242 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
9243 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
9244 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
9245 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
9246 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
9247 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
9248 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
9249 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
9250 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
9251 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
9252 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
9253 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
9254 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
9255 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
9256 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
9257 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
9258 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
9259 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
9260 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
9261 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
9262 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
9263 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
9264 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
9265 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
9266 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
9267 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
9268 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
9269 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
9270 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
9271 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
9272 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
9273 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
9274 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
9275 * config/avr/avr.c (avr_starting_frame_offset): Make static and
9276 return a HOST_WIDE_INT.
9277 (avr_builtin_setjmp_frame_value): Use it instead of
9278 STARTING_FRAME_OFFSET.
9279 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9280 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
9281 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
9283 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9284 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
9285 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
9286 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9287 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
9288 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
9289 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9290 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
9291 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
9292 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9293 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
9294 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
9296 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9297 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
9298 * config/mips/mips.c (mips_compute_frame_info): Refer to
9299 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
9300 (mips_starting_frame_offset): New function.
9301 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9302 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
9303 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
9304 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
9305 and return a HOST_WIDE_INT.
9306 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9307 (mmix_initial_elimination_offset): Refer to
9308 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
9309 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
9310 * config/pa/pa.c (pa_starting_frame_offset): New function.
9311 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
9312 (pa_expand_prologue): Likewise.
9313 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9314 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
9315 !FRAME_GROWS_DOWNWARD handling to...
9316 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9317 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
9318 !FRAME_GROWS_DOWNWARD handling to...
9319 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9320 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
9321 !FRAME_GROWS_DOWNWARD handling to...
9322 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9323 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
9325 (rs6000_starting_frame_offset): New function.
9326 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
9327 !FRAME_GROWS_DOWNWARD handling to...
9328 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9329 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
9330 !FRAME_GROWS_DOWNWARD handling to...
9331 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9332 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
9333 !FRAME_GROWS_DOWNWARD handling to...
9334 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9335 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
9336 (rs6000_starting_frame_offset): New function.
9337 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
9338 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
9339 * config/vax/vax.c (vax_starting_frame_offset): New function.
9340 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
9341 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9342 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
9343 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
9344 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9345 * system.h (STARTING_FRAME_OFFSET): Poison.
9347 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9349 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
9350 SCALAR_TYPE_MODE instead of TYPE_MODE.
9352 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9353 Alan Hayward <alan.hayward@arm.com>
9354 David Sherwood <david.sherwood@arm.com>
9356 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
9358 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9359 Alan Hayward <alan.hayward@arm.com>
9360 David Sherwood <david.sherwood@arm.com>
9362 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
9364 2017-10-23 Richard Biener <rguenther@suse.de>
9366 PR tree-optimization/82129
9367 * tree-ssa-pre.c (bitmap_set_and): Remove.
9368 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
9369 canonicalizing expressions in the set to those with lowest
9370 ID rather than taking that from the first edge.
9372 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9374 * combine.c (rtx_equal_for_field_assignment_p): Use
9375 byte_lowpart_offset.
9377 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9378 Alan Hayward <alan.hayward@arm.com>
9379 David Sherwood <david.sherwood@arm.com>
9381 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
9382 to a SUBREG_PROMOTED_VAR.
9384 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9385 Alan Hayward <alan.hayward@arm.com>
9386 David Sherwood <david.sherwood@arm.com>
9388 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
9389 (expand_debug_source_expr): Likewise.
9390 * combine.c (combine_simplify_rtx): Likewise.
9391 * cse.c (fold_rtx): Likewise.
9392 * optabs.c (expand_float): Likewise.
9393 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9394 (simplify_binary_operation_1): Likewise.
9396 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9397 Alan Hayward <alan.hayward@arm.com>
9398 David Sherwood <david.sherwood@arm.com>
9400 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
9401 (record_promoted_value): Likewise.
9402 * expr.c (expand_expr_real_2): Likewise.
9403 * ree.c (update_reg_equal_equiv_notes): Likewise.
9404 (combine_set_extension): Likewise.
9405 * rtlanal.c (low_bitmask_len): Likewise.
9406 * simplify-rtx.c (neg_const_int): Likewise.
9407 (simplify_binary_operation_1): Likewise.
9409 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9410 Alan Hayward <alan.hayward@arm.com>
9411 David Sherwood <david.sherwood@arm.com>
9413 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
9414 * regcprop.c (maybe_mode_change): Likewise.
9415 * reload1.c (alter_reg): Likewise.
9417 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9419 * inchash.h (inchash::hash::add_wide_int): New function.
9420 * lto-streamer-out.c (hash_tree): Use it.
9422 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9424 * inchash.h (inchash::hash::add_wide_int): Rename to...
9425 (inchash::hash::add_hwi): ...this.
9426 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
9427 (polymorphic_call_target_hasher::hash): Likewise.
9428 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
9429 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
9430 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
9431 * lto-streamer-out.c (hash_tree): Likewise.
9432 * optc-save-gen.awk: Likewise.
9433 * tree.c (add_expr): Likewise.
9435 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
9438 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
9439 for ordered inequality comparisons even with TARGET_IEEE_FP.
9441 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
9444 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
9445 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
9446 Expand with cmp<dwi>_doubleword.
9448 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
9450 * extend.texi: Add x86 specific to 'nocf_check' attribute.
9451 List CET intrinsics.
9452 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
9453 specific to -fcf-protection option.
9455 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
9457 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
9458 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
9459 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
9460 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
9461 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
9462 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
9463 (extra_objs): Add cet.o for Linux/x86 targets.
9464 (tmake_file): Add i386/t-cet for Linux/x86 targets.
9465 * config/i386/cet.c: New file.
9466 * config/i386/cetintrin.h: Likewise.
9467 * config/i386/t-cet: Likewise.
9468 * config/i386/cpuid.h (bit_SHSTK): New.
9469 (bit_IBT): Likewise.
9470 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
9471 pass IBT and SHSTK bits.
9472 * config/i386/i386-builtin-types.def
9473 (VOID_FTYPE_UNSIGNED_PVOID): New.
9474 (VOID_FTYPE_UINT64_PVOID): Likewise.
9475 * config/i386/i386-builtin.def: Add CET intrinsics.
9476 * config/i386/i386-c.c (ix86_target_macros_internal): Add
9477 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
9478 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
9479 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
9481 * config/i386/i386.c (rest_of_insert_endbranch): New.
9482 (pass_data_insert_endbranch): Likewise.
9483 (pass_insert_endbranch): Likewise.
9484 (make_pass_insert_endbranch): Likewise.
9485 (ix86_notrack_prefixed_insn_p): Likewise.
9486 (ix86_target_string): Add -mibt, -mshstk flags.
9487 (ix86_option_override_internal): Add flag_cf_protection
9489 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
9490 (ix86_print_operand): Add 'notrack' prefix output.
9491 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
9492 (ix86_expand_builtin): Expand CET intrinsics.
9493 (x86_output_mi_thunk): Add 'endbranch' instruction.
9494 * config/i386/i386.h (TARGET_IBT): New.
9495 (TARGET_IBT_P): Likewise.
9496 (TARGET_SHSTK): Likewise.
9497 (TARGET_SHSTK_P): Likewise.
9498 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
9499 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
9500 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
9501 (builtin_setjmp_setup): New pattern.
9502 (builtin_longjmp): Likewise.
9503 (rdssp<mode>): Likewise.
9504 (incssp<mode>): Likewise.
9505 (saveprevssp): Likewise.
9506 (rstorssp): Likewise.
9507 (wrss<mode>): Likewise.
9508 (wruss<mode>): Likewise.
9509 (setssbsy): Likewise.
9510 (clrssbsy): Likewise.
9511 (nop_endbr): Likewise.
9512 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
9514 * config/i386/immintrin.h: Include <cetintrin.h>.
9515 * config/i386/linux-common.h
9516 (file_end_indicate_exec_stack_and_cet): New prototype.
9517 (TARGET_ASM_FILE_END): New.
9519 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
9521 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
9522 latencies instead of having separate table; make difference between
9523 integer and float costs.
9524 * i386.h (processor_costs): Remove scalar_stmt_cost,
9525 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
9526 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
9528 * x86-tune-costs.h: Remove entries which has been removed in
9529 procesor_costs from all tables; make cond_taken_branch_cost
9530 and cond_not_taken_branch_cost COST_N_INSNS based.
9532 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
9534 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
9536 2017-10-20 Jakub Jelinek <jakub@redhat.com>
9538 * config/i386/i386.md (isa): Remove fma_avx512f.
9539 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
9540 <avx512>_fmadd_<mode>_mask3<round_name>,
9541 <avx512>_fmsub_<mode>_mask<round_name>,
9542 <avx512>_fmsub_<mode>_mask3<round_name>,
9543 <avx512>_fnmadd_<mode>_mask<round_name>,
9544 <avx512>_fnmadd_<mode>_mask3<round_name>,
9545 <avx512>_fnmsub_<mode>_mask<round_name>,
9546 <avx512>_fnmsub_<mode>_mask3<round_name>,
9547 <avx512>_fmaddsub_<mode>_mask<round_name>,
9548 <avx512>_fmaddsub_<mode>_mask3<round_name>,
9549 <avx512>_fmsubadd_<mode>_mask<round_name>,
9550 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
9551 (*vec_widen_umult_even_v16si<mask_name>,
9552 *vec_widen_smult_even_v16si<mask_name>): Likewise.
9553 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
9555 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
9557 * extend.texi: Add 'nocf_check' documentation.
9558 * gimple.texi: Add second parameter to
9559 gimple_build_call_from_tree.
9560 * invoke.texi: Add -fcf-protection documentation.
9561 * rtl.texi: Add REG_CALL_NOTRACK documenation.
9563 2017-10-20 Richard Biener <rguenther@suse.de>
9565 PR tree-optimization/82473
9566 * tree-vect-loop.c (vectorizable_reduction): Properly get at
9567 the largest input type.
9569 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
9571 * c-attribs.c (handle_nocf_check_attribute): New function.
9572 (c_common_attribute_table): Add 'nocf_check' handling.
9573 * gimple-parser.c: Add second argument NULL to
9574 gimple_build_call_from_tree.
9575 * attrib.c (comp_type_attributes): Check nocf_check attribute.
9576 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
9578 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
9579 * common.opt: Add fcf-protection flag.
9580 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
9581 * flag-types.h: Add enum cf_protection_level.
9582 * gimple.c (gimple_build_call_from_tree): Add second parameter.
9583 Add 'nocf_check' attribute propagation to gimple call.
9584 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
9585 (gimple_build_call_from_tree): Update prototype.
9586 (gimple_call_nocf_check_p): New function.
9587 (gimple_call_set_nocf_check): Likewise.
9588 * gimplify.c: Add second argument to gimple_build_call_from_tree.
9589 * ipa-icf.c: Add nocf_check attribute in statement hash.
9590 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
9591 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
9592 * toplev.c (process_options): Add flag_cf_protection handling.
9594 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
9596 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
9598 2017-10-20 Richard Biener <rguenther@suse.de>
9600 PR tree-optimization/82603
9601 * tree-if-conv.c (predicate_mem_writes): Make sure to only
9602 remove false predicated stores.
9604 2017-10-20 Richard Biener <rguenther@suse.de>
9606 * graphite-isl-ast-to-gimple.c
9607 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
9608 Remove return value and simplify, dump copied stmt after lhs
9610 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
9611 Reduce dump verbosity.
9612 (gsi_insert_earliest): Likewise.
9613 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
9614 * graphite.c (print_global_statistics): Adjust dumping.
9615 (print_graphite_scop_statistics): Likewise.
9616 (print_graphite_statistics): Do not dump loops here.
9617 (graphite_transform_loops): But here.
9619 2017-10-20 Nicolas Roche <roche@adacore.com>
9621 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
9622 * configure: Regenerate.
9624 2017-10-20 Jakub Jelinek <jakub@redhat.com>
9627 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
9628 functions when optimizing replace GIMPLE_RETURN stmts with
9629 calls to __builtin_unreachable ().
9632 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
9633 for -fsanitize=thread link of executables.
9634 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
9635 link of executables.
9638 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
9640 (<shift_insn><mode>3<mask_name>): Change the last of the 3
9641 define_insns for logical vector shifts to use VI248_AVX512BW
9642 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
9643 condition, useless isa and prefix attributes. Change the first
9644 2 of these define_insns to ...
9645 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
9646 define_insn for avx512vl.
9647 (<shift_insn><mode>3): ... and this, new define_insn without
9648 masking for non-avx512vl.
9651 * config/i386/sse.md (*andnot<mode>3,
9652 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
9653 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
9654 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
9655 not applied use empty suffix even for TARGET_AVX512VL.
9656 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
9657 is applied, supply evex,evex or evex,evex,evex instead of just
9660 2017-10-20 Julia Koval <julia.koval@intel.com>
9662 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
9663 (OPTION_MASK_ISA_GFNI_UNSET): New.
9664 (ix86_handle_option): Handle OPT_mgfni.
9665 * config/i386/cpuid.h (bit_GFNI): New.
9666 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
9667 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
9668 * config/i386/i386.c (ix86_target_string): Add -mgfni.
9669 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
9670 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
9671 * config/i386/i386.opt: Add mgfni.
9673 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
9675 * config/msp430/msp430.c (msp430_option_override): Disable
9676 -fdelete-null-pointer-checks.
9677 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
9679 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
9681 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
9682 of x87 and SSE instructions.
9684 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
9686 * asan.c (create_cond_insert_point): Do not update edge count.
9687 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
9688 (afdo_propagate_circuit): Likewise.
9689 (afdo_calculate_branch_prob): Likewise.
9690 (afdo_annotate_cfg): Likewise.
9691 * basic-block.h (struct edge_def): Remove count.
9692 (edge_def::count): New accessor.
9693 * bb-reorder.c (rotate_loop): Update.
9694 (find_traces_1_round): Update.
9695 (connect_traces): Update.
9696 (sanitize_hot_paths): Update.
9697 * cfg.c (unchecked_make_edge): Update.
9698 (make_single_succ_edge): Update.
9699 (check_bb_profile): Update.
9700 (dump_edge_info): Update.
9701 (update_bb_profile_for_threading): Update.
9702 (scale_bbs_frequencies_int): Update.
9703 (scale_bbs_frequencies_gcov_type): Update.
9704 (scale_bbs_frequencies_profile_count): Update.
9705 (scale_bbs_frequencies): Update.
9706 * cfganal.c (connect_infinite_loops_to_exit): Update.
9707 * cfgbuild.c (compute_outgoing_frequencies): Update.
9708 (find_many_sub_basic_blocks): Update.
9709 * cfgcleanup.c (try_forward_edges): Update.
9710 (try_crossjump_to_edge): Update
9711 * cfgexpand.c (expand_gimple_cond): Update
9712 (expand_gimple_tailcall): Update
9713 (construct_exit_block): Update
9714 * cfghooks.c (verify_flow_info): Update
9715 (redirect_edge_succ_nodup): Update
9716 (split_edge): Update
9717 (make_forwarder_block): Update
9718 (duplicate_block): Update
9719 (account_profile_record): Update
9720 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
9721 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
9722 * cfgloopmanip.c (scale_loop_profile): Update.
9724 (lv_adjust_loop_entry_edge): Update.
9725 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
9726 (force_nonfallthru_and_redirect): Update.
9727 (purge_dead_edges): Update.
9728 (rtl_flow_call_edges_add): Update.
9729 * cgraphunit.c (init_lowered_empty_function): Update.
9730 (cgraph_node::expand_thunk): Update.
9731 * gimple-pretty-print.c (dump_probability): Update.
9732 (dump_edge_probability): Update.
9733 * gimple-ssa-isolate-paths.c (isolate_path): Update.
9734 * haifa-sched.c (sched_create_recovery_edges): Update.
9735 * hsa-gen.c (convert_switch_statements): Update.
9736 * ifcvt.c (dead_or_predicable): Update.
9737 * ipa-inline-transform.c (inline_transform): Update.
9738 * ipa-split.c (split_function): Update.
9739 * ipa-utils.c (ipa_merge_profiles): Update.
9740 * loop-doloop.c (add_test): Update.
9741 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
9742 * lto-streamer-in.c (input_cfg): Update.
9743 (input_function): Update.
9744 * lto-streamer-out.c (output_cfg): Update.
9745 * modulo-sched.c (sms_schedule): Update.
9746 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
9747 * predict.c (maybe_hot_edge_p): Update.
9748 (unlikely_executed_edge_p): Update.
9749 (probably_never_executed_edge_p): Update.
9750 (dump_prediction): Update.
9751 (drop_profile): Update.
9752 (propagate_unlikely_bbs_forward): Update.
9753 (determine_unlikely_bbs): Update.
9754 (force_edge_cold): Update.
9755 * profile.c (compute_branch_probabilities): Update.
9756 * reg-stack.c (better_edge): Update.
9757 * shrink-wrap.c (handle_simple_exit): Update.
9758 * tracer.c (better_p): Update.
9759 * trans-mem.c (expand_transaction): Update.
9760 (split_bb_make_tm_edge): Update.
9761 * tree-call-cdce.c: Update.
9762 * tree-cfg.c (gimple_find_sub_bbs): Update.
9763 (gimple_split_edge): Update.
9764 (gimple_duplicate_sese_region): Update.
9765 (gimple_duplicate_sese_tail): Update.
9766 (gimple_flow_call_edges_add): Update.
9767 (insert_cond_bb): Update.
9768 (execute_fixup_cfg): Update.
9769 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
9770 * tree-complex.c (expand_complex_div_wide): Update.
9771 * tree-eh.c (lower_resx): Update.
9772 (unsplit_eh): Update.
9773 (cleanup_empty_eh_move_lp): Update.
9774 * tree-inline.c (copy_edges_for_bb): Update.
9775 (freqs_to_counts): Update.
9776 (copy_cfg_body): Update.
9777 * tree-ssa-dce.c (remove_dead_stmt): Update.
9778 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
9779 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
9780 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
9781 (unloop_loops): Update.
9782 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
9783 * tree-ssa-loop-split.c (connect_loops): Update.
9784 (split_loop): Update.
9785 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
9786 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
9787 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
9788 * tree-ssa-reassoc.c (branch_fixup): Update.
9789 * tree-ssa-tail-merge.c (replace_block_by): Update.
9790 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
9791 (compute_path_counts): Update.
9792 (update_profile): Update.
9793 (recompute_probabilities): Update.
9794 (update_joiner_offpath_counts): Update.
9795 (estimated_freqs_path): Update.
9796 (freqs_to_counts_path): Update.
9797 (clear_counts_path): Update.
9798 (ssa_fix_duplicate_block_edges): Update.
9799 (duplicate_thread_path): Update.
9800 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
9801 (case_bit_test_cmp): Update.
9802 (collect_switch_conv_info): Update.
9803 (gen_inbound_check): Update.
9804 (do_jump_if_equal): Update.
9805 (emit_cmp_and_jump_insns): Update.
9806 * tree-tailcall.c (decrease_profile): Update.
9807 (eliminate_tail_call): Update.
9808 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
9809 (vect_do_peeling): Update.
9810 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
9811 * ubsan.c (ubsan_expand_null_ifn): Update.
9812 (ubsan_expand_ptr_ifn): Update.
9813 * value-prof.c (gimple_divmod_fixed_value): Update.
9814 (gimple_mod_pow2): Update.
9815 (gimple_mod_subtract): Update.
9816 (gimple_ic): Update.
9817 (gimple_stringop_fixed_value): Update.
9819 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
9822 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
9824 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
9826 PR rtl-optimization/82395
9827 * ira-color.c (allocno_priority_compare_func): Fix comparison step
9828 based on non_spilled_static_chain_regno_p.
9830 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
9832 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
9833 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
9834 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
9836 2017-10-19 Martin Sebor <msebor@redhat.com>
9838 PR tree-optimization/82596
9839 * tree.c (array_at_struct_end_p): Handle STRING_CST.
9841 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
9843 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
9844 (get_mem_refs_of_builtin_call): Likewise.
9845 * builtins.c (expand_builtin_apply): Adjust call to
9846 allocate_dynamic_stack_space.
9847 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
9848 the third argument to allocate_dynamic_stack_space, otherwise -1.
9849 (expand_builtin): Deal with all alloca variants.
9850 (is_inexpensive_builtin): Likewise.
9851 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
9852 * calls.c (special_function_p): Deal with all alloca variants.
9853 (initialize_argument_information): Adjust call to
9854 allocate_dynamic_stack_space.
9855 (expand_call): Likewise.
9856 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
9857 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
9858 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
9859 use it for the stack usage computation.
9860 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
9861 * function.c (gimplify_parameters): Call build_alloca_call_expr.
9862 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
9863 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
9864 (in_loop_p): Remove first argument and useless check.
9865 (pass_walloca::execute): Remove useless test and adjust call to above.
9866 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
9867 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
9868 (gimplify_call_expr): Deal with all alloca variants.
9869 * hsa-gen.c (gen_hsa_alloca): Likewise.
9870 (gen_hsa_insns_for_call): Likewise.
9871 * ipa-pure-const.c (special_builtin_state): Likewise.
9872 * tree-chkp.c (chkp_build_returned_bound): Likewise.
9873 * tree-object-size.c (alloc_object_size): Likewise.
9874 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
9875 (call_may_clobber_ref_p_1): Likewise.
9876 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
9877 (ccp_fold_stmt): Likewise.
9878 (optimize_stack_restore): Likewise.
9879 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
9880 (mark_all_reaching_defs_necessary_1): Likewise.
9881 (propagate_necessity): Likewise.
9882 (eliminate_unnecessary_stmts): Likewise.
9883 * tree.c (build_common_builtin_nodes): Build
9884 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
9885 (build_alloca_call_expr): New function.
9886 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
9887 (CASE_BUILT_IN_ALLOCA): Likewise.
9888 (build_alloca_call_expr): Declare.
9889 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
9891 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
9894 * dwarf2out.c (new_die_raw): New static inline function.
9895 (new_die): Use it to create the DIE.
9896 (add_AT_external_die_ref): Likewise.
9897 (clone_die): Likewise.
9898 (clone_as_declaration): Likewise.
9899 (dwarf2out_vms_debug_main_pointer): Likewise.
9900 (base_type_die): Likewise. Remove early return for corner cases.
9901 Do not call add_pubtype on the DIE here.
9902 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
9903 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
9904 typedefs for base types with DW_AT_endianity. Make sure a DIE with
9905 native order exists for base types, attach the DIE manually and call
9906 add_pubtype on it. Do not equate a reverse order DIE to the type.
9908 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
9910 * config/arm/arm.c (align_ok_ldrd_strd): New function.
9911 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
9913 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
9915 2017-10-19 Jakub Jelinek <jakub@redhat.com>
9917 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
9918 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
9919 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
9920 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
9921 * opts.c (sanitizer_opts): Add builtin.
9922 * ubsan.c (instrument_builtin): New function.
9923 (pass_ubsan::execute): Call it.
9924 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
9925 * doc/invoke.texi: Document -fsanitize=builtin.
9927 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
9928 builtins, store max (log2 (align), 0) into uchar field instead of
9929 align into uptr field.
9930 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
9931 store uchar 0 field instead of uptr 0 field.
9932 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
9933 instead of passing one address of struct with 2 locations pass
9934 two addresses of structs with 1 location each.
9935 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
9936 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
9937 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
9938 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
9939 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
9940 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
9941 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
9942 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
9944 2017-10-19 Martin Liska <mliska@suse.cz>
9947 * file-find.c (remove_prefix): Remove.
9948 * file-find.h (remove_prefix): Likewise.
9949 * gcc-ar.c: Remove smartness of lookup.
9951 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
9953 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
9954 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
9955 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
9957 2017-10-19 Jakub Jelinek <jakub@redhat.com>
9960 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
9961 (setcc + and to xor + setcc): New peephole2.
9963 2017-10-19 Tom de Vries <tom@codesourcery.com>
9965 * doc/sourcebuild.texi (Test Directives, Variants of
9966 dg-require-support): Add dg-require-stack-size.
9968 2017-10-19 Martin Liska <mliska@suse.cz>
9971 * gimplify.c (gimplify_decl_expr): Do not instrument variables
9972 that have a large alignment.
9973 (gimplify_target_expr): Likewise.
9975 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
9977 PR rtl-optimization/82602
9978 * ira.c (rtx_moveable_p): Return false for volatile asm.
9980 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
9983 * config/i386/i386-modes.def (CCGZ): New CC mode.
9984 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
9985 * config/i386/predicates.md (ix86_comparison_operator):
9987 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
9988 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
9989 with double-word subtraction.
9990 (put_condition_code): Handle CCGZmode.
9992 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
9994 * wide-int.cc (debug (const wide_int &)): New.
9995 (debug (const wide_int *)): New.
9996 (debug (const widest_int &)): New.
9997 (debug (const widest_int *)): New.
9999 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
10001 PR middle-end/82556
10002 * lra-constraints.c (curr_insn_transform): Use non-input operand
10003 instead of output one for matched reload.
10005 2017-10-18 Bin Cheng <bin.cheng@arm.com>
10007 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
10008 (tree-ssa-loop-ivopts.h): New header file.
10009 (struct builtin_info): New fields.
10010 (classify_builtin_1): Compute and record base and offset parts for
10011 memset builtin partition by calling strip_offset.
10012 (offset_cmp, fuse_memset_builtins): New functions.
10013 (finalize_partitions): Fuse adjacent memset partitions by calling
10015 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
10016 Expose the interface.
10017 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
10019 2017-10-18 Bin Cheng <bin.cheng@arm.com>
10021 PR tree-optimization/82574
10022 * tree-loop-distribution.c (find_single_drs): New parameter. Check
10023 that data reference must be executed exactly once per iteration
10024 against the outermost loop in nest.
10025 (classify_partition): Update call to above function.
10027 2017-10-18 Richard Biener <rguenther@suse.de>
10029 PR tree-optimization/82591
10030 * graphite.c (graphite_transform_loops): Move code gen message
10032 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
10033 Here. Handle scop_to_isl_ast failing.
10034 (scop_to_isl_ast): Limit the number of ISL operations.
10036 2017-10-18 Richard Biener <rguenther@suse.de>
10038 * graphite-isl-ast-to-gimple.c
10039 (translate_isl_ast_to_gimple::set_rename): Simplify.
10040 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
10041 (graphite_copy_stmts_from_block): ... here.
10042 (copy_bb_and_scalar_dependences): Simplify.
10043 (add_parameters_to_ivs_params): Canonicalize.
10044 (generate_entry_out_of_ssa_copies): Simplify.
10045 * graphite-sese-to-poly.c (extract_affine_name): Simplify
10046 by passing in ISL dimension.
10047 (parameter_index_in_region_1): Rename to ...
10048 (parameter_index_in_region): ... this.
10049 (extract_affine): Adjust assert, pass down parameter index.
10050 (add_param_constraints): Use range-info when available.
10051 (build_scop_context): Adjust.
10052 * sese.c (new_sese_info): Adjust.
10053 (free_sese_info): Likewise.
10054 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
10055 Remove unused typedefs.
10056 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
10058 2017-10-18 Martin Liska <mliska@suse.cz>
10060 * combine.c (simplify_compare_const): Add gcc_fallthrough.
10062 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
10064 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
10065 (s390_sched_init): Do not reset s390_sched_state if we entered the
10066 current basic block via a fallthru edge and all others are unlikely.
10068 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
10070 * config/s390/s390.c (NUM_SIDES): New variable.
10071 (LONGRUNNING_THRESHOLD): New variable.
10072 (LATENCY_FACTOR): New variable.
10073 (s390_sched_score): Decrease score for long-running instructions on
10075 (s390_sched_variable_issue): Perform bookkeeping for long-running
10078 2017-10-18 Richard Biener <rguenther@suse.de>
10080 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
10081 Simplify with removal of the parameter rename map.
10082 (set_rename): Likewise.
10083 (should_copy_to_new_region): Likewise.
10084 (graphite_copy_stmts_from_block): Likewise.
10085 (copy_bb_and_scalar_dependences): Remove initialization of
10086 unused copied_bb_map.
10087 (copy_def): Remove.
10088 (copy_internal_parameters): Likewise.
10089 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
10090 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
10091 Use INTEGRAL_TYPE_P.
10092 (parameter_index_in_region_1): Rename to ...
10093 (assign_parameter_index_in_region): ... this. Assert we have
10094 a parameter we handle.
10095 (scan_tree_for_params): Adjust.
10096 * sese.h (parameter_rename_map_t): Remove.
10097 (struct sese_info_t): Remove unused parameter_rename_map and
10098 copied_bb_map members.
10099 * sese.c (new_sese_info): Adjust.
10100 (free_sese_info): Likewise.
10102 2017-10-18 Martin Liska <mliska@suse.cz>
10105 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
10106 on an abnormal edge.
10108 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
10110 * doc/invoke.texi (ffunction-sections and fdata-sections):
10113 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
10115 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
10116 the use statement can throw internally.
10118 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
10120 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
10121 any RTX present on the RHS of a SET.
10122 * compare-elim.c (try_eliminate_compare): Restore comment.
10124 2017-10-17 Jakub Jelinek <jakub@redhat.com>
10126 * langhooks.h (struct lang_hooks): Document that tree_size langhook
10127 may be also called on tcc_type nodes.
10128 * langhooks.c (lhd_tree_size): Likewise.
10130 2017-10-17 David Malcolm <dmalcolm@redhat.com>
10132 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
10133 format_warning_at_substring.
10134 (maybe_warn): Convert source_range * param to a location_t. Pass
10135 UNKNOWN_LOCATION rather than NULL to fmtwarn.
10136 (format_directive): Remove code to extract source_ranges and
10137 source_range * in favor of just a location_t.
10138 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
10140 * substring-locations.c (format_warning_va): Convert
10141 source_range * param to a location_t.
10142 (format_warning_at_substring): Likewise.
10143 * substring-locations.h (format_warning_va): Likewise.
10144 (format_warning_at_substring): Likewise.
10146 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
10148 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
10150 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
10151 and scatter/gather ops.
10153 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
10154 vec_gather_load and vec_scatter_store.
10155 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
10156 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
10157 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
10158 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
10159 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
10160 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
10162 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
10164 * reg-stack.c (compare_for_stack_reg): Add bool argument.
10165 Detect FTST instruction and handle its register pops. Only pop
10166 second operand if can_pop_second_op is true.
10167 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
10168 set can_pop_second_op to false in the compare_for_stack_reg call.
10170 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
10171 output_fp_compare for stack register operands.
10172 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
10173 instructions here. Do not emit stack register pops here. Assert
10174 that FCOMPP pops next to top stack register. Rewrite function.
10176 2017-10-17 Nathan Sidwell <nathan@acm.org>
10178 PR middle-end/82577
10179 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
10180 use DECL_ASSEMBLER_NAME_RAW.
10182 PR middle-end/82546
10183 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
10186 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
10187 Wilco Dijkstra <wilco.dijkstra@arm.com>
10189 * builtins.c (expand_builtin_update_setjmp_buf): Add a
10190 converstion to Pmode from the buf_addr.
10192 2017-10-17 Richard Biener <rguenther@suse.de>
10194 * graphite-dependences.c (scop_get_reads_and_writes): Change
10195 output parameters to references.
10197 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
10199 PR 71026/tree-optimization
10200 * fold-const.c (distribute_real_division): Removed.
10201 (fold_binary_loc): Remove calls to distribute_real_divison.
10203 2017-10-17 Richard Biener <rguenther@suse.de>
10205 * graphite-scop-detection.c
10206 (scop_detection::stmt_has_simple_data_refs_p): Always use
10207 the full nest as region.
10208 (try_generate_gimple_bb): Likewise.
10209 * sese.c (scalar_evolution_in_region): Simplify now that
10210 SCEV can handle instantiation in regions.
10211 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
10212 in the non-loop part of a function if requested.
10214 2017-10-17 Richard Biener <rguenther@suse.de>
10216 PR tree-optimization/82563
10217 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
10219 (graphite_regenerate_ast_isl): Call it.
10220 * graphite-scop-detection.c (build_scops): Remove entry edge split.
10222 2017-10-17 Jakub Jelinek <jakub@redhat.com>
10224 PR tree-optimization/82549
10225 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
10226 Formatting fixes. Instead of calling make_bit_field_ref with negative
10229 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
10231 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
10232 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
10233 _mm_maskz_reduce_ss): New.
10234 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
10235 __builtin_ia32_reducess_mask): Ditto..
10236 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
10237 * config/i386/sse.md (reduces<mode>): Renamed to ...
10238 (reduces<mode><mask_scalar_name>): ... this.
10239 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
10241 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
10242 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
10244 2017-10-16 David Malcolm <dmalcolm@redhat.com>
10246 * Makefile.in (OBJS): Add unique-ptr-tests.o.
10247 * selftest-run-tests.c (selftest::run_tests): Call
10248 selftest::unique_ptr_tests_cc_tests.
10249 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
10250 * unique-ptr-tests.cc: New file.
10252 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
10255 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
10257 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
10258 (make_hard_regno_born, make_hard_regno_dead): Update
10259 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
10261 2017-10-16 Jeff Law <law@redhat.com>
10263 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
10265 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
10267 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
10269 2017-10-16 Olivier Hainque <hainque@adacore.com>
10271 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
10272 with_cpu if we were configured for an e500v2 target cpu name.
10274 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
10276 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
10277 * doc/invoke.texi: Document +nodsp as a valid extension for
10280 2017-10-16 Martin Liska <mliska@suse.cz>
10282 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
10283 (test_set_range): Likewise.
10284 (test_range_functions): Rename to ...
10285 (test_bit_in_range): ... this.
10286 (sbitmap_c_tests): Add new test.
10288 2017-10-16 Tamar Christina <tamar.christina@arm.com>
10290 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
10292 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
10293 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
10295 2017-10-16 Tamar Christina <tamar.christina@arm.com>
10297 * config/aarch64/aarch64-builtins.c
10298 (aarch64_types_quadopu_lane_qualifiers): New.
10299 (TYPES_QUADOPU_LANE): New.
10300 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
10301 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
10302 (aarch64_<sur>dot_laneq<vsi2qi>): New.
10303 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
10304 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
10305 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
10306 (Vdottype, DOTPROD): New.
10307 (sur): Add SDOT and UDOT.
10309 2017-10-16 Tamar Christina <tamar.christina@arm.com>
10311 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
10312 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
10313 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
10314 Add TARGET_DOTPROD.
10315 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
10316 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
10317 Enable TARGET_DOTPROD.
10318 (cortex-a75.cortex-a55): Likewise.
10319 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
10321 2017-10-16 Tamar Christina <tamar.christina@arm.com>
10323 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
10324 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
10326 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
10328 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
10329 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
10330 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
10331 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
10332 * config/arm/types.md (neon_dot, neon_dot_q): New.
10333 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
10335 2017-10-16 Tamar Christina <tamar.christina@arm.com>
10337 * config/arm/arm.h (TARGET_DOTPROD): New.
10338 * config/arm/arm.c (arm_arch_dotprod): New.
10339 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
10340 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
10341 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
10342 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
10343 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
10344 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
10345 * doc/invoke.texi (armv8.2-a): Document dotprod
10347 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
10349 * i386.c (ix86_vec_cost): New function.
10350 (ix86_rtx_costs): Handle vector operations better.
10351 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
10352 * x86-tune-costs.h: Add new costs to all tables.
10354 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
10356 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
10358 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
10359 divsd, sqrtss and sqrtsd
10360 * x86-tune-costs.h: Add new entries to all costs.
10361 (znver1_cost): Fix to match real instruction latencies.
10363 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10364 Michael Collison <michael.collison@arm.com>
10366 * compare-elim.c: Include emit-rtl.h.
10367 (can_merge_compare_into_arith): New function.
10368 (try_validate_parallel): Likewise.
10369 (try_merge_compare): Likewise.
10370 (try_eliminate_compare): Call the above when no previous clobber
10372 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
10375 2017-10-14 Jakub Jelinek <jakub@redhat.com>
10377 PR middle-end/62263
10378 PR middle-end/82498
10379 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
10380 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
10382 PR middle-end/62263
10383 PR middle-end/82498
10384 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
10385 to be any operand_equal_p operands. For & (B - 1) require
10386 B to be power of 2. Recognize
10387 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
10389 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
10392 * optabs.c (expand_memory_blockage): Fix call of
10393 targetm.have_memory_blockage.
10395 2017-10-14 Jakub Jelinek <jakub@redhat.com>
10398 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
10399 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
10400 objects to extra_objs instead of overwriting it.
10402 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
10404 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
10405 Use any_fp_register_operand as operand[3] predicate. Simplify
10406 equality test for operands[2] and operands[4] memory location.
10407 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
10408 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
10409 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
10410 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
10411 any_fp_register_operand as operand[1] predicate. Simplify
10412 equality test for operands[0] and operands[3] memory location.
10413 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
10414 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
10415 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
10417 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
10419 * target-insns.def: Add memory_blockage.
10420 * optabs.c (expand_memory_blockage): New function.
10421 (expand_asm_memory_barrier): Rename ...
10422 (expand_asm_memory_blockage): ... to this.
10423 (expand_mem_thread_fence): Call expand_memory_blockage
10424 instead of expand_asm_memory_barrier.
10425 (expand_mem_singnal_fence): Ditto.
10426 (expand_atomic_load): Ditto.
10427 (expand_atomic_store): Ditto.
10428 * doc/md.texi (Standard Pattern Names For Generation):
10429 Document memory_blockage instruction pattern.
10431 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
10433 * config/rl78/rl78.c (rl78_emit_libcall): New function.
10434 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
10435 * config/rl78/rl78.md: New define_expand "adddi3".
10437 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
10439 * cfghooks.c (verify_flow_info): Disable check that all probabilities
10442 2017-10-13 Jeff Law <law@redhat.com>
10444 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
10446 2017-10-13 Jakub Jelinek <jakub@redhat.com>
10449 * internal-fn.c (expand_mul_overflow): If both operands have
10450 the same highpart of -1 or 0 and the topmost bit of lowpart
10451 is different, overflow is if res <= 0 rather than res < 0.
10453 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
10455 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
10456 TARGET_P9_VECTOR code for unaligned_load case.
10458 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
10460 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
10462 2017-10-13 Nathan Sidwell <nathan@acm.org>
10464 * tree-core.h (tree_contains_struct): Make bool.
10465 * tree.c (tree_contains_struct): Likewise.
10466 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
10467 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
10468 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
10469 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
10470 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
10472 2017-10-13 Richard Biener <rguenther@suse.de>
10474 * graphite-isl-ast-to-gimple.c
10475 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
10476 parameters and dominance check.
10477 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
10478 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
10479 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
10480 Do not update SSA form here or do intermediate IL verification.
10481 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
10482 (graphite_initialize): Remove check on the number of loops in
10483 the function and inline into graphite_transform_loops.
10484 (graphite_finalize): Inline into graphite_transform_loops.
10485 (graphite_transform_loops): Perform SSA update and IL verification
10487 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
10489 2017-10-13 Richard Biener <rguenther@suse.de>
10491 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
10492 graphite_expression_type_precision): Avoid global constructor
10494 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
10495 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
10496 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
10497 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
10498 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
10499 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
10501 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
10504 * config/i386/i386.h (ix86_red_zone_size): New.
10505 * config/i386/i386.md (push peephole2s): Replace
10506 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
10508 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
10509 Alan Hayward <alan.hayward@arm.com>
10510 David Sherwood <david.sherwood@arm.com>
10512 * combine.c (can_change_dest_mode): Reject changes in
10513 REGMODE_NATURAL_SIZE.
10515 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
10516 Alan Hayward <alan.hayward@arm.com>
10517 David Sherwood <david.sherwood@arm.com>
10519 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
10520 (expand_debug_source_expr): Likewise.
10521 * combine.c (combine_simplify_rtx): Likewise.
10522 * cse.c (fold_rtx): Likewise.
10523 * fwprop.c (canonicalize_address): Likewise.
10524 * targhooks.c (default_shift_truncation_mask): Likewise.
10526 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
10527 Alan Hayward <alan.hayward@arm.com>
10528 David Sherwood <david.sherwood@arm.com>
10530 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
10531 (widened_mode): Likewise.
10532 (expand_unop): Likewise.
10533 * ree.c (transform_ifelse): Likewise.
10534 (merge_def_and_ext): Likewise.
10535 (combine_reaching_defs): Likewise.
10536 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10538 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
10539 Alan Hayward <alan.hayward@arm.com>
10540 David Sherwood <david.sherwood@arm.com>
10542 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
10543 * combine.c (gen_lowpart_for_combine): Likewise.
10544 * dwarf2out.c (rtl_for_decl_location): Likewise.
10545 * final.c (alter_subreg): Likewise.
10546 * rtlhooks.c (gen_lowpart_general): Likewise.
10547 (gen_lowpart_if_possible): Likewise.
10549 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
10550 Alan Hayward <alan.hayward@arm.com>
10551 David Sherwood <david.sherwood@arm.com>
10553 * calls.c (expand_call): Use subreg_lowpart_offset.
10554 * cse.c (cse_insn): Likewise.
10555 * regcprop.c (copy_value): Likewise.
10556 (copyprop_hardreg_forward_1): Likewise.
10558 2017-10-13 Jakub Jelinek <jakub@redhat.com>
10561 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
10562 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
10563 =Q constraints to +Q and into insn condition add check
10564 that operands[0] and operands[1] are equal.
10565 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
10566 =Q constraints to +Q and into insn condition add check
10567 that operands[0] is equal to either operands[1] or operands[2].
10570 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
10571 instead of handling MINUS_EXPR twice (once for each argument),
10572 canonicalize operand order and handle just once, use rtype where
10573 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
10576 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
10577 any values of __C while still being pattern recognizable as a simple
10578 rotate instruction.
10580 2017-10-13 Richard Biener <rguenther@suse.de>
10582 PR tree-optimization/82451
10584 2017-10-02 Richard Biener <rguenther@suse.de>
10586 PR tree-optimization/82355
10587 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
10588 a mapping for the enclosing loop but avoid generating one for
10589 the loop tree root.
10590 (copy_bb_and_scalar_dependences): Remove premature codegen
10591 error on PHIs in blocks duplicated into multiple places.
10592 * graphite-scop-detection.c
10593 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
10594 in the region use it as loop and nest to analyze the DR in.
10595 (try_generate_gimple_bb): Likewise.
10596 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
10597 (add_loop_constraints): For blocks in a loop not in the region
10598 create a dimension with a single iteration.
10599 * sese.h (gbb_loop_at_index): Remove assert.
10601 * cfgloop.c (loop_preheader_edge): For the loop tree root
10602 return the single successor of the entry block.
10603 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
10604 Reset the SCEV hashtable and niters.
10605 * graphite-scop-detection.c
10606 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
10607 assert that we only have POLYNOMIAL_CHREC that vary in loops
10608 contained in the region.
10609 (scop_detection::graphite_can_represent_expr): Adjust.
10610 (scop_detection::stmt_has_simple_data_refs_p): For loops
10611 not in the region set loop to NULL. The nest is now the
10612 entry edge to the region.
10613 (try_generate_gimple_bb): Likewise.
10614 * sese.c (scalar_evolution_in_region): Adjust for
10615 instantiate_scev change.
10616 * tree-data-ref.h (graphite_find_data_references_in_stmt):
10617 Make nest parameter the edge into the region.
10618 (create_data_ref): Likewise.
10619 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
10620 entry edge into a region and adjust instantiate_scev calls.
10621 (create_data_ref): Likewise.
10622 (graphite_find_data_references_in_stmt): Likewise.
10623 (find_data_references_in_stmt): Pass the loop preheader edge
10624 from the nest argument.
10625 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
10626 parameter the edge into the region.
10627 (instantiate_parameters): Use the loop preheader edge as entry.
10628 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
10630 (get_instantiated_value_entry): Make instantiate_below parameter
10631 the edge into the region.
10632 (instantiate_scev_name): Likewise. Adjust dominance checks,
10633 when we cannot use loop-based instantiation instantiate by
10634 walking use-def chains.
10635 (instantiate_scev_poly): Adjust.
10636 (instantiate_scev_binary): Likewise.
10637 (instantiate_scev_convert): Likewise.
10638 (instantiate_scev_not): Likewise.
10639 (instantiate_array_ref): Remove.
10640 (instantiate_scev_3): Likewise.
10641 (instantiate_scev_2): Likewise.
10642 (instantiate_scev_1): Likewise.
10643 (instantiate_scev_r): Do not blindly handle N-operand trees.
10644 Do not instantiate array-refs. Handle all constants and invariants.
10645 (instantiate_scev): Make instantiate_below parameter
10646 the edge into the region.
10647 (resolve_mixers): Use the loop preheader edge for the region
10648 parameter to instantiate_scev_r.
10649 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
10651 2017-10-13 Richard Biener <rguenther@suse.de>
10653 PR tree-optimization/82525
10654 * graphite-isl-ast-to-gimple.c
10655 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
10657 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
10658 Fail code generation when we cannot represent the isl integer.
10659 (binary_op_to_tree): Elide modulo operations that are no-ops
10660 in the type we code generate. Remove now superfluous code
10663 2017-10-13 Richard Biener <rguenther@suse.de>
10665 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
10666 (scop_detection::harmful_loop_in_region): Remove premature
10667 IV type restriction.
10668 (scop_detection::graphite_can_represent_scev): We can handle
10669 pointer IVs just fine.
10671 2017-10-13 Alan Modra <amodra@gmail.com>
10673 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
10674 "Clobbers and Scratch Registers". Add paragraph on
10675 alternative to clobbers for scratch registers and OpenBLAS
10678 2017-10-13 Alan Modra <amodra@gmail.com>
10680 * doc/extend.texi (Clobbers): Correct vax example. Delete old
10681 example of a memory input for a string of known length. Move
10682 commentary out of table. Add a number of new examples
10683 covering array memory inputs.
10685 2017-10-12 Martin Liska <mliska@suse.cz>
10687 PR tree-optimization/82493
10688 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
10689 (test_range_functions): New function.
10690 (sbitmap_c_tests): Likewise.
10691 * selftest-run-tests.c (selftest::run_tests): Run new tests.
10692 * selftest.h (sbitmap_c_tests): New function.
10694 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
10696 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
10698 * config/rs6000/amo.h: Fix spacing issue.
10700 2017-10-12 Jakub Jelinek <jakub@redhat.com>
10703 * config/i386/i386.md (*ashl<mode>3_mask_1,
10704 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
10705 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
10708 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
10710 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
10711 (profile_probability): Set max_probability
10712 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
10714 * profile-count.c (profile_probability::differs_from_p): Do not
10715 rely on max_probaiblity == 10000
10717 2017-10-12 Jeff Law <law@redhat.com>
10719 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
10722 2017-10-12 Martin Sebor <msebor@redhat.com>
10726 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
10727 (handle_alias_pairs): Call it.
10728 * common.opt (-Wattribute-alias): New option.
10729 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
10730 * doc/invoke.texi (-Wattribute-alias): Document.
10732 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
10735 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
10737 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
10739 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
10740 (make_hard_regno_born, make_hard_regno_dead): Update
10741 bb_killed_pseudos and bb_gen_pseudos.
10743 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
10745 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
10747 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
10749 * config/alpha/alpha.c (alpha_split_conditional_move):
10750 Use std::swap instead of manually swapping.
10751 (alpha_stdarg_optimize_hook): Ditto.
10752 (alpha_canonicalize_comparison): Ditto.
10754 2017-10-12 Bin Cheng <bin.cheng@arm.com>
10756 * tree-loop-distribution.c (struct builtin_info): New struct.
10757 (struct partition): Refactor fields into struct builtin_info.
10758 (partition_free): Free struct builtin_info.
10759 (build_size_arg_loc, build_addr_arg_loc): Delete.
10760 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
10761 information from struct builtin_info.
10762 (find_single_drs): New function refactored from classify_partition.
10763 Also moved builtin validity checks to this function.
10764 (compute_access_range, alloc_builtin): New functions.
10765 (classify_builtin_st, classify_builtin_ldst): New functions.
10766 (classify_partition): Refactor code into functions find_single_drs,
10767 classify_builtin_st and classify_builtin_ldst.
10768 (distribute_loop): Don't do runtime alias check when distributing
10770 (find_seed_stmts_for_distribution): New function.
10771 (pass_loop_distribution::execute): Refactor code finding seed
10772 stmts into above function. Support distribution for the innermost
10773 two-level loop nest. Adjust dump information.
10775 2017-10-12 Bin Cheng <bin.cheng@arm.com>
10777 * tree-loop-distribution.c: Adjust the general comment.
10778 (NUM_PARTITION_THRESHOLD): New macro.
10779 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
10780 (classify_partition): Skip builtin pattern of loop nest's inner loop.
10781 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
10782 in call to build_partition_graph.
10783 (finalize_partitions): New parameter. Make loop distribution more
10784 conservative by fusing more partitions.
10785 (distribute_loop): Don't do runtime alias check in case of loop nest
10787 (find_seed_stmts_for_distribution): New function.
10788 (prepare_perfect_loop_nest): New function.
10789 (pass_loop_distribution::execute): Refactor code finding seed stmts
10790 and loop nest into above functions. Support loop nest distribution.
10791 Adjust dump information accordingly.
10793 2017-10-12 Bin Cheng <bin.cheng@arm.com>
10795 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
10796 and set PTYPE_SEQUENTIAL for merged partition.
10798 2017-10-12 Richard Biener <rguenther@suse.de>
10800 PR tree-optimization/69728
10802 2017-09-19 Richard Biener <rguenther@suse.de>
10804 PR tree-optimization/69728
10805 * graphite-sese-to-poly.c (schedule_error): New global.
10806 (add_loop_schedule): Handle empty domain by failing the
10808 (build_original_schedule): Handle schedule_error.
10810 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
10811 domain by returning an unchanged schedule.
10813 2017-10-12 Jakub Jelinek <jakub@redhat.com>
10815 * genrecog.c (validate_pattern): For VEC_SELECT verify that
10816 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
10818 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
10820 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
10823 2017-10-12 Jakub Jelinek <jakub@redhat.com>
10826 * expr.c (store_field): Don't optimize away bitsize == 0 store
10827 from CALL_EXPR with addressable return type.
10829 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
10831 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
10832 * config/rs6000/rs6000.md (sel): Delete mode attribute.
10833 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
10834 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
10835 TARGET_ISEL instead of TARGET_ISEL<sel>.
10837 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
10839 * config/rs6000/rs6000.c
10840 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
10842 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
10844 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
10846 (reg_or_cint_operand): Fix comment.
10847 (reg_or_zero_operand): New predicate.
10848 * config/rs6000/rs6000-protos.h (output_isel): Delete.
10849 * config/rs6000/rs6000.c (output_isel): Delete.
10850 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
10851 instead of reg_or_cint_operand. Output instruction directly (not via
10853 (isel_unsigned_<mode>): Ditto.
10854 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
10855 gpc_reg_operand. Add an instruction alternative for this. Output
10856 instruction directly.
10857 (*isel_reversed_unsigned_<mode>): Ditto.
10859 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
10861 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
10862 (TARGET_CANONICALIZE_COMPARISON): Define.
10864 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
10867 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
10868 Check whether the dest is REG before adding REG_EQUIV note.
10870 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
10873 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
10875 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
10876 (make_hard_regno_born, make_hard_regno_dead): Update
10877 bb_killed_pseudos and bb_gen_pseudos.
10879 2017-10-11 Nathan Sidwell <nathan@acm.org>
10881 * incpath.h (enum incpath_kind): Name enum, prefix values.
10882 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
10883 * incpath.c (heads, tails): Use INC_MAX.
10884 (add_env_var_paths, add_standard_paths): Use incpath_kind.
10885 (merge_include_chains, split_quote_chain,
10886 register_include_chains): Update incpath_kind names.
10887 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
10888 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
10890 (add_framework_path, darwin_register_objc_includes): Likewise.
10891 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
10893 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
10895 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
10896 Do not use float_operator operator predicate.
10897 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
10898 * config/i386/predicates.md (float_operator): Remove predicate.
10900 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
10902 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
10903 (*jccxf_i387): Ditto.
10904 (*jcc<mode>_i387): Ditto.
10905 (*jccu<mode>_i387): Ditto.
10906 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
10907 (*jcc_*_i387 splitters): Remove.
10908 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
10909 * config/i386/i386.c (ix86_split_fp_branch): Remove.
10910 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
10913 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
10915 * profile-count.h (slow_safe_scale_64bit): New function.
10916 (safe_scale_64bit): New inline.
10917 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
10918 * profile-count.c: Include wide-int.h
10919 (slow_safe_scale_64bit): New.
10921 2017-10-11 Nathan Sidwell <nathan@acm.org>
10923 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
10924 HAS_DECL_ASSEMBLER_NAME_P.
10925 * gimple-expr.c (gimple_decl_printable_name: Check
10926 HAS_DECL_ASSEMBLER_NAME_P too.
10927 * ipa-utils.h (type_in_anonymous_namespace_p): Check
10928 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
10929 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
10930 * passes.c (rest_of_decl_compilation): Check
10931 HAS_DECL_ASSEMBLER_NAME_P too.
10932 * recog.c (verify_changes): Likewise.
10933 * tree-pretty-print.c (dump_decl_name): Likewise.
10934 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
10936 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
10937 (SET_DECL_ASSEMBLER_NAME): Use it.
10938 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
10939 (COPY_DECL_ASSEMBLER_NAME): Likewise.
10940 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
10942 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
10944 * config.gcc (i386, x86_64): Add extra objects.
10945 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
10946 (ix86_min_insn_size): Declare.
10947 (ix86_issue_rate): Declare.
10948 (ix86_adjust_cost): Declare.
10949 (ia32_multipass_dfa_lookahead): Declare.
10950 (ix86_macro_fusion_p): Declare.
10951 (ix86_macro_fusion_pair_p): Declare.
10952 (ix86_bd_has_dispatch): Declare.
10953 (ix86_bd_do_dispatch): Declare.
10954 (ix86_core2i7_init_hooks): Declare.
10955 (ix86_atom_sched_reorder): Declare.
10956 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
10957 (COSTS_N_BYTES): Move to x86-tune-costs.h.
10958 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
10959 (rip_relative_addr_p): Rename to ...
10960 (ix86_rip_relative_addr_p): ... this one; export.
10961 (memory_address_length): Update.
10962 (ix86_issue_rate): Move to x86-tune-sched.c.
10963 (ix86_flags_dependent): Move to x86-tune-sched.c.
10964 (ix86_agi_dependent): Move to x86-tune-sched.c.
10965 (exact_dependency_1): Move to x86-tune-sched.c.
10966 (exact_store_load_dependency): Move to x86-tune-sched.c.
10967 (ix86_adjust_cost): Move to x86-tune-sched.c.
10968 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
10969 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
10970 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
10971 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
10972 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
10973 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
10974 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
10975 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
10976 (min_insn_size): Rename to ...
10977 (ix86_min_insn_size): ... this one; export.
10978 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
10979 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
10980 (core2i7_first_cycle_multipass_backtrack): Move to
10981 x86-tune-sched-core.c.
10982 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
10983 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
10984 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
10985 (ix86_avoid_jump_mispredicts): Update.
10986 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
10987 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
10988 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
10989 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
10990 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
10991 (MAX_INSN): Move to ix86-tune-sched-bd.c.
10992 (MAX_IMM): Move to ix86-tune-sched-bd.c.
10993 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
10994 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
10995 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
10996 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
10997 (MAX_STORE): Move to ix86-tune-sched-bd.c.
10998 (BIG): Move to ix86-tune-sched-bd.c.
10999 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
11000 (enum insn_path): Move to ix86-tune-sched-bd.c.
11001 (get_mem_group): Move to ix86-tune-sched-bd.c.
11002 (is_cmp): Move to ix86-tune-sched-bd.c.
11003 (dispatch_violation): Move to ix86-tune-sched-bd.c.
11004 (is_branch): Move to ix86-tune-sched-bd.c.
11005 (is_prefetch): Move to ix86-tune-sched-bd.c.
11006 (init_window): Move to ix86-tune-sched-bd.c.
11007 (allocate_window): Move to ix86-tune-sched-bd.c.
11008 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
11009 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
11010 (process_end_window): Move to ix86-tune-sched-bd.c.
11011 (allocate_next_window): Move to ix86-tune-sched-bd.c.
11012 (find_constant): Move to ix86-tune-sched-bd.c.
11013 (get_num_immediates): Move to ix86-tune-sched-bd.c.
11014 (has_immediate): Move to ix86-tune-sched-bd.c.
11015 (get_insn_path): Move to ix86-tune-sched-bd.c.
11016 (get_insn_group): Move to ix86-tune-sched-bd.c.
11017 (count_num_restricted): Move to ix86-tune-sched-bd.c.
11018 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
11019 (add_insn_window): Move to ix86-tune-sched-bd.c.
11020 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
11021 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
11022 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
11023 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
11024 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
11025 (do_dispatch): Move to ix86-tune-sched-bd.c.
11026 (has_dispatch): Move to ix86-tune-sched-bd.c.
11027 * config/i386/t-i386: Add new object files.
11028 * config/i386/x86-tune-costs.h: New file.
11029 * config/i386/x86-tune-sched-atom.c: New file.
11030 * config/i386/x86-tune-sched-bd.c: New file.
11031 * config/i386/x86-tune-sched-core.c: New file.
11032 * config/i386/x86-tune-sched.c: New file.
11034 2017-10-11 Liu Hao <lh_mouse@126.com>
11036 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
11037 the generic version below instead.
11038 (should_colorize): Recognize Windows consoles as terminals
11040 * pretty-print.c [__MINGW32__] (write_all): New function.
11041 [__MINGW32__] (find_esc_head): Likewise.
11042 [__MINGW32__] (find_esc_terminator): Likewise.
11043 [__MINGW32__] (eat_esc_sequence): Likewise.
11044 [__MINGW32__] (mingw_ansi_fputs): New function that handles
11046 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
11049 2017-10-11 Richard Biener <rguenther@suse.de>
11051 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
11052 Properly call analyze_scalar_evolution with the loop of the stmt.
11054 2017-10-11 Richard Biener <rguenther@suse.de>
11056 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
11057 * tree-core.h (tree_base): Add chrec_var union member.
11058 * tree.h (CHREC_VAR): Remove.
11059 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
11060 * tree-chrec.h (build_polynomial_chrec): Adjust.
11061 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
11062 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
11064 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
11066 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
11067 * match.pd: ... here.
11068 ((T) X == (T) Y): Relax condition.
11070 2017-10-11 Bin Cheng <bin.cheng@arm.com>
11072 PR tree-optimization/82472
11073 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
11075 (break_alias_scc_partitions): Update postorder number.
11077 2017-10-11 Martin Liska <mliska@suse.cz>
11080 * opts.c (parse_no_sanitize_attribute): Do not use error_value
11082 * opts.h (parse_no_sanitize_attribute): Remove last argument.
11084 2017-10-11 Martin Liska <mliska@suse.cz>
11086 * print-rtl.c (print_insn): Move declaration of idbuf
11087 to same scope as name.
11089 2017-10-11 Martin Liska <mliska@suse.cz>
11094 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
11095 volatile arguments.
11097 2017-10-11 Martin Liska <mliska@suse.cz>
11100 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
11101 volatile arguments.
11103 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
11105 * config.gcc (default_gnu_indirect_function): Default to yes for
11106 arm*-*-linux* with glibc.
11108 2017-10-11 Richard Biener <rguenther@suse.de>
11110 * tree-scalar-evolution.c (get_scalar_evolution): Handle
11111 default-defs and types we do not want to analyze.
11112 (interpret_loop_phi): Replace unreachable code with an assert.
11113 (compute_scalar_evolution_in_loop): Remove and inline ...
11114 (analyze_scalar_evolution_1): ... here, replacing condition with
11115 what makes the intent clearer. Remove handling of cases
11116 get_scalar_evolution now handles.
11118 2017-10-10 Jim Wilson <wilson@tuliptree.org>
11120 PR rtl-optimization/81434
11121 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
11122 comment for main loop. In sched_group_found if, also add checks for
11123 pass and min_cost_group.
11125 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
11127 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
11128 (rs6000_insn_cost): New function.
11129 * config/rs6000/rs6000.md (cost): New attribute.
11131 2017-10-10 Jakub Jelinek <jakub@redhat.com>
11132 H.J. Lu <hongjiu.lu@intel.com>
11136 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
11137 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
11138 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
11139 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
11140 to OPTION_MASK_ISA_AVX512VL - builtins that have both
11141 OPTION_MASK_ISA_MMX and some other bit set require both
11142 mmx and the ISAs without the mmx bit.
11143 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
11144 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
11145 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
11146 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
11147 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
11148 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
11149 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
11150 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
11151 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
11152 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
11153 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
11154 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
11155 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
11156 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
11157 Add OPTION_MASK_ISA_MMX.
11159 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
11161 * config.gcc (armv7*-*-freebsd*): New target.
11162 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
11164 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
11166 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
11167 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
11170 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
11172 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
11173 a template parameter.
11174 (WIDE_INT_REF_FOR): Update accordingly.
11175 * tree.h (wi::int_traits <const_tree>): Delete.
11176 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
11177 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
11178 (wi::tree_to_wide_ref): New typedef.
11179 (wi::to_wide): New function.
11180 * calls.c (get_size_range): Use wi::to_wide when operating on
11181 trees as wide_ints.
11182 * cgraph.c (cgraph_node::create_thunk): Likewise.
11183 * config/i386/i386.c (ix86_data_alignment): Likewise.
11184 (ix86_local_alignment): Likewise.
11185 * dbxout.c (stabstr_O): Likewise.
11186 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
11187 * expr.c (const_vector_from_tree): Likewise.
11188 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
11189 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
11190 (fold_negate_expr_1, int_const_binop_1, const_binop)
11191 (fold_convert_const_int_from_real, optimize_bit_field_compare)
11192 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
11193 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
11194 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
11195 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
11196 (fold_not_const, round_up_loc): Likewise.
11197 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
11198 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
11199 (alloca_call_type): Likewise.
11200 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
11201 * godump.c (go_output_typedef): Likewise.
11202 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
11203 * internal-fn.c (get_min_precision): Likewise.
11204 * ipa-cp.c (ipcp_store_vr_results): Likewise.
11205 * ipa-polymorphic-call.c
11206 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
11207 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
11208 (ipa_modify_call_arguments): Likewise.
11209 * match.pd: Likewise.
11210 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
11211 * print-tree.c (print_node_brief, print_node): Likewise.
11212 * stmt.c (expand_case): Likewise.
11213 * stor-layout.c (layout_type): Likewise.
11214 * tree-affine.c (tree_to_aff_combination): Likewise.
11215 * tree-cfg.c (group_case_labels_stmt): Likewise.
11216 * tree-data-ref.c (dr_analyze_indices): Likewise.
11217 (prune_runtime_alias_test_list): Likewise.
11218 * tree-dump.c (dequeue_and_dump): Likewise.
11219 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
11220 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
11221 * tree-pretty-print.c (dump_generic_node): Likewise.
11222 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
11223 (simple_iv_with_niters): Likewise.
11224 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
11225 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
11226 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
11227 * tree-ssa-loop-niter.c (split_to_var_and_offset)
11228 (refine_value_range_using_guard, number_of_iterations_ne_max)
11229 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
11230 (get_cst_init_from_scev, record_nonwrapping_iv)
11231 (scev_var_range_cant_overflow): Likewise.
11232 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
11233 * tree-ssa-pre.c (compute_avail): Likewise.
11234 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
11235 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
11236 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
11237 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
11238 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
11239 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
11240 (dump_case_nodes, try_switch_expansion): Likewise.
11241 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
11242 (vect_do_peeling): Likewise.
11243 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
11244 * tree-vect-stmts.c (vectorizable_load): Likewise.
11245 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
11246 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
11247 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
11248 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
11249 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
11250 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
11251 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
11252 (evrp_dom_walker::before_dom_children): Likewise.
11253 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
11254 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
11255 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
11256 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
11257 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
11258 (get_range_pos_neg): Likewise.
11259 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
11260 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
11261 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
11262 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
11263 * config/avr/avr.c (avr_fold_builtin): Likewise.
11264 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
11265 * config/msp430/msp430.c (msp430_attr): Likewise.
11266 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
11267 * config/powerpcspe/powerpcspe-c.c
11268 (altivec_resolve_overloaded_builtin): Likewise.
11269 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
11270 (rs6000_expand_ternop_builtin): Likewise.
11271 * config/rs6000/rs6000-c.c
11272 (altivec_resolve_overloaded_builtin): Likewise.
11273 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
11274 (rs6000_expand_ternop_builtin): Likewise.
11275 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
11277 2017-10-10 Bin Cheng <bin.cheng@arm.com>
11279 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
11280 when copying loop nest with only one inner loop.
11282 2017-10-10 Richard Biener <rguenther@suse.de>
11284 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
11285 blocks if SCEV is active.
11286 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
11288 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
11289 (scev_initialize): Assert we are not yet initialized.
11291 2017-10-10 Bin Cheng <bin.cheng@arm.com>
11293 * tree-loop-distribution.c (generate_loops_for_partition): Remove
11294 inner loop's exit stmt by making it always exit the loop, otherwise
11295 we would generate an infinite empty loop.
11297 2017-10-10 Bin Cheng <bin.cheng@arm.com>
11299 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
11300 renaming variables in new preheader if it's deleted.
11302 2017-10-10 Bin Cheng <bin.cheng@arm.com>
11304 * tree-loop-distribution.c (struct partition): Remove unused field
11305 loops of the structure.
11306 (partition_alloc, partition_free): Ditto.
11307 (build_rdg_partition_for_vertex): Ditto.
11309 2017-10-09 Jeff Law <law@redhat.com>
11311 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
11312 return type to match prototype and documentation.
11314 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
11316 * config/rs6000/rs6000.c (processor_costs): Move to ...
11317 * config/rs6000/rs6000.h: ... here.
11318 (rs6000_cost): Declare.
11320 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
11322 * except.c (setjmp_fn): New global variable.
11323 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
11324 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
11325 if DONT_USE_BUILTIN_SETJMP is defined.
11327 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
11329 * target.def (insn_cost): New hook.
11330 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
11331 * doc/tm.texi: Regenerate.
11332 * rtlanal.c (insn_cost): Use the new hook.
11334 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
11336 * combine.c (combine_validate_cost): Compute the new insn_cost,
11337 not just pattern_cost.
11338 (try_combine): Adjust comment.
11340 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
11342 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
11344 * combine.c (uid_insn_cost): Adjust comment.
11345 (combine_validate_cost): Adjust comment. Use pattern_cost instead
11347 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
11348 * dse.c (find_shift_sequence): Ditto.
11349 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
11350 (bb_valid_for_noce_process_p): Use pattern_cost.
11351 * rtl.h (insn_rtx_cost): Delete.
11352 (pattern_cost): New prototype.
11353 (insn_cost): New prototype.
11354 * rtlanal.c (insn_rtx_cost): Rename to...
11355 (pattern_cost): ... this.
11358 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
11360 * config/i386/i386.md (*jcc_2): Remove insn pattern.
11361 (*jcc<mode>_0_r_i387): Ditto.
11362 (*jccxf_r_i387): Ditto.
11363 (*jcc<mode>_r_i387): Ditto.
11364 (*jccu<mode>_r_i387): Ditto.
11365 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
11366 (*jcc): Rename from *jcc_1.
11368 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11370 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
11371 deferred rescans after the lvx/stvx recombination pre-pass.
11373 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
11375 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
11376 memory operation instruction support.
11377 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
11378 (rs6000-ibm-aix[789]*): Likewise.
11379 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
11380 Document new functions.
11382 2017-10-09 Richard Biener <rguenther@suse.de>
11384 PR tree-optimization/82397
11385 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
11386 equality only for semantically equal trees.
11388 2017-10-09 Richard Biener <rguenther@suse.de>
11390 PR tree-optimization/82449
11391 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
11392 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
11393 allow constant addresses.
11394 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
11397 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11399 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
11402 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11405 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
11408 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11411 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
11413 2017-10-09 Jakub Jelinek <jakub@redhat.com>
11416 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
11417 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
11419 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
11421 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
11422 (WI_BINARY_PREDICATE_RESULT): Likewise.
11423 (wi::binary_traits::operator_result): New type.
11424 (wi::binary_traits::predicate_result): Likewise.
11425 (generic_wide_int::operator~, unary generic_wide_int::operator-)
11426 (generic_wide_int::operator==, generic_wide_int::operator!=)
11427 (generic_wide_int::operator&, generic_wide_int::and_not)
11428 (generic_wide_int::operator|, generic_wide_int::or_not)
11429 (generic_wide_int::operator^, generic_wide_int::operator+
11430 (binary generic_wide_int::operator-, generic_wide_int::operator*):
11432 (operator~, unary operator-, operator==, operator!=, operator&)
11433 (operator|, operator^, operator+, binary operator-, operator*): New
11435 * expr.c (get_inner_reference): Use wi::bit_and_not.
11436 * fold-const.c (fold_binary_loc): Likewise.
11437 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
11438 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
11439 (bit_value_binop): Likewise.
11440 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
11441 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
11442 (extract_range_from_binary_expr_1): Likewise.
11443 (masked_increment): Likewise.
11444 (simplify_bit_ops_using_ranges): Likewise.
11446 2017-10-09 Martin Jambor <mjambor@suse.cz>
11449 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
11450 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
11451 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
11453 (hsa_fixup_mov_insn_type): New function.
11454 (hsa_op_with_type::get_in_type): Use it.
11455 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
11456 immediates in an assert.
11457 (hsa_op_with_type::extend_int_to_32bit): New method.
11458 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
11459 types. Convert to dest type if necessary.
11460 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
11461 (reg_for_gimple_ssa): Pass false as min32int to
11462 hsa_type_for_scalar_tree_type.
11463 (gen_hsa_addr): Fixup type when creating addresable temporary.
11464 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
11465 (gen_hsa_unary_operation): Extend operands and convert to dest type if
11466 necessary. Call hsa_fixup_mov_insn_type.
11467 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
11468 extend operands and convert to dest type if necessary.
11469 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
11470 to dest type if necessary.
11471 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
11472 if conversion nt necessary and size matches.
11473 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
11474 to dest type if necessary.
11475 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
11476 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
11478 (gen_hsa_clrsb): Likewise.
11479 (gen_hsa_ffs): Likewise.
11480 (gen_hsa_divmod): Extend operands and convert to dest type if
11482 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
11484 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
11486 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
11487 Remove empty default arguments. Use a brace block as output
11489 (conditional return): Ditto.
11491 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
11492 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
11494 (group_ending_nop): Ditto.
11495 (doloop_end): Ditto.
11496 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
11497 (splitters for those): Ditto.
11499 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
11501 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
11502 a conditional jump (and the compare for it) so that pc_rtx is the
11504 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
11505 for the deleted and renamed ctr<mode>_internal[234] patterns.
11506 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
11507 Delete second conditional return pattern.
11508 (ctr<mode>_internal2): Delete this second bdnz pattern.
11509 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
11510 (ctr<mode>_internal4): Delete this second bdz pattern.
11512 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
11514 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
11515 (always_initialized_rtx_for_ssa_name_p): New predicate.
11516 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
11517 (finish_out_of_ssa): Free new field of SA.
11518 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
11519 * tree-ssa-coalesce.c: Include tree-ssa.h.
11520 (get_parm_default_def_partitions): Remove extern keyword.
11521 (get_undefined_value_partitions): New function.
11522 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
11523 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
11524 uninitialized bits.
11525 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
11527 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
11529 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
11531 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
11533 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
11534 for targets that preffer 128bit.
11536 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
11538 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
11540 2017-10-08 Olivier Hainque <hainque@adacore.com>
11542 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
11543 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
11545 (thumb_set_return_address): Likewise.
11547 2017-10-08 Olivier Hainque <hainque@adacore.com>
11549 * common/config/arm/arm-common.c (arm_except_unwind_info):
11550 Handle DWARF2_UNWIND_INFO.
11552 2017-10-07 Michael Collison <michael.collison@arm.com>
11554 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
11557 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
11559 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
11560 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
11561 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
11562 defined, force the creation of a new block for a dispatch label.
11564 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
11566 * invoke.texi (Wsuggest-attribute=cold): Document.
11567 * common.opt (Wsuggest-attribute=cold): New
11568 * ipa-pure-const.c (warn_function_cold): New function.
11569 * predict.c (compute_function_frequency): Use it.
11570 * predict.h (warn_function_cold): Declare.
11572 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
11574 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
11577 2017-10-06 Martin Liska <mliska@suse.cz>
11579 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
11580 keyword for member functions.
11581 (struct sanopt_tree_couple): New struct.
11582 (struct sanopt_tree_couple_hash): New function.
11583 (struct sanopt_ctx): Add new hash_map.
11584 (has_dominating_ubsan_ptr_check): New function.
11585 (record_ubsan_ptr_check_stmt): Likewise.
11586 (maybe_optimize_ubsan_ptr_ifn): Likewise.
11587 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
11588 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
11590 2017-10-06 Sudakshina Das <sudi.das@arm.com>
11593 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
11594 aarch64_simd_valid_immediate on CONST_VECTORs.
11595 (aarch64_reg_or_bic_imm): Likewise.
11597 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
11599 PR rtl-optimization/82396
11600 * haifa-sched.c (ready_sort_real): Disable qsort checking.
11602 2017-10-06 Sebastian Pop <sebpop@gmail.com>
11604 * graphite-dependences.c (scop_get_reads): Move code to...
11605 (scop_get_must_writes): Move code to...
11606 (scop_get_may_writes): Move code to...
11607 (scop_get_reads_and_writes): ... here.
11608 (scop_get_dependences): Call scop_get_reads_and_writes.
11610 2017-10-06 Jakub Jelinek <jakub@redhat.com>
11612 PR tree-optimization/82434
11613 * fold-const.h (can_native_encode_type_p,
11614 can_native_encode_string_p): Remove.
11615 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
11616 don't encode anything, just return what would be otherwise returned.
11617 (native_encode_fixed, native_encode_complex, native_encode_vector):
11619 (native_encode_string): Likewise. Inline by hand
11620 can_native_encode_string_p.
11621 (can_native_encode_type_p): Remove.
11622 (can_native_encode_string_p): Remove.
11623 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
11624 STRING_CSTs using can_native_encode_string_p, test all
11625 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
11626 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
11627 argument from native_encode_expr.
11628 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
11629 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
11632 2017-10-06 Richard Biener <rguenther@suse.de>
11634 PR tree-optimization/82397
11635 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
11636 operand_equal_p but rely on data_ref_compare_tree for detecting
11638 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
11639 to match up with dr_group_sort_cmp.
11641 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11644 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
11646 * config/s390/s390-builtin-types.def: Regenerate.
11648 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11651 * config/s390/s390-builtin-types.def: Regenerate.
11652 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
11653 Change flag from B_VXE to B_VX.
11654 (s390_vec_min_dbl): Remove B_VXE flag.
11656 2017-10-06 Richard Biener <rguenther@suse.de>
11658 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
11659 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
11660 translate_isl_ast_to_gimple::is_valid_rename,
11661 translate_isl_ast_to_gimple::get_rename,
11662 translate_isl_ast_to_gimple::get_def_bb_for_const,
11663 translate_isl_ast_to_gimple::get_new_name,
11664 translate_isl_ast_to_gimple::collect_all_ssa_names,
11665 translate_isl_ast_to_gimple::copy_loop_phi_args,
11666 translate_isl_ast_to_gimple::collect_all_ssa_names,
11667 translate_isl_ast_to_gimple::copy_loop_phi_args,
11668 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
11669 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
11670 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
11671 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
11672 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
11673 translate_isl_ast_to_gimple::copy_cond_phi_args,
11674 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
11675 translate_isl_ast_to_gimple::edge_for_new_close_phis,
11676 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
11677 translate_isl_ast_to_gimple::rename_uses,
11678 translate_isl_ast_to_gimple::rename_all_uses): Remove.
11679 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
11680 (set_rename_for_each_def): Likewise.
11681 (graphite_copy_stmts_from_block): Handle debug stmt resetting
11682 here. Handle rewriting SCEV analyzable uses here.
11683 (copy_bb_and_scalar_dependences): Generate code for PHI
11685 (graphite_regenerate_ast_isl): Adjust.
11686 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
11687 (add_write, add_read): New functions.
11688 (build_cross_bb_scalars_def): Use it and simplify.
11689 (build_cross_bb_scalars_use): Likewise.
11690 (graphite_find_cross_bb_scalar_vars): Inline into...
11691 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
11692 simulating out-of-SSA. Compute liveout and add dependencies.
11693 (build_scops): Force an empty entry block.
11694 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
11696 (sese_build_liveouts): Declare.
11697 (sese_trivially_empty_bb_p): Likewise.
11698 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
11699 compute liveout and debug_liveout.
11700 (sese_bad_liveouts_use): Remove.
11701 (sese_reset_debug_liveouts_bb): Likewise.
11702 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
11703 (sese_build_liveouts): Build liveout and debug_liveout and store
11705 (new_sese_info): Adjust.
11706 (free_sese_info): Likewise.
11707 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
11708 do not build liveout here.
11709 (move_sese_in_condition): Adjust region entry.
11710 (scev_analyzable_p): Match up with chrec_apply requirements.
11711 (sese_trivially_empty_bb_p): New.
11712 * tree-into-ssa.c (get_reaching_def): Properly support generating
11713 default-defs for incremental rewrite of anonymous names.
11715 2017-10-06 Richard Biener <rguenther@suse.de>
11717 * graphite-sese-to-poly.c (extract_affine): For casts increasing
11718 precision do not perform modulo reduction.
11720 2017-10-06 Richard Biener <rguenther@suse.de>
11722 PR tree-optimization/82436
11723 * tree-vect-slp.c (vect_supported_load_permutation_p): More
11724 conservatively choose the vectorization factor when checking
11725 whether we can perform the required load permutation.
11726 (vect_transform_slp_perm_load): Assert when we may not fail.
11728 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
11730 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
11731 message for incompatible -msdata=* and -mcall-* options.
11733 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
11735 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
11736 rate for post-reload scheduling.
11738 2017-10-05 Tamar Christina <tamar.christina@arm.com>
11740 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
11742 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
11744 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
11745 to improve monte carlo in scimark.
11747 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
11749 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
11750 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
11751 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
11752 pentium4_cost, nocona_cost): Set reassociation width to 1.
11753 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
11754 width to 2 for fp operations and 1 otherwise.
11755 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
11757 (atom_cost): Set reassociation width to 2.
11758 (slm_cost, generic_cost): Set fp reassociation width
11759 to 2 and 1 otherwise.
11760 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
11761 (core_cost): Set fp reassociation width to 4 and vector to 2.
11762 (ix86_reassociation_width): Rewrite using cost table; special case
11763 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
11764 and TARGET_AVX128_OPTIMAL.
11765 * config/i386/i386.h (processor_costs): Add
11766 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
11767 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
11768 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
11769 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
11770 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
11771 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
11773 2017-10-05 Nathan Sidwell <nathan@acm.org>
11775 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
11777 2017-10-05 Tamar Christina <tamar.christina@arm.com>
11779 * config/arm/arm.c (arm_test_fpu_data): New.
11780 (arm_run_selftests): Call arm_test_fpu_data.
11782 2017-10-04 Nathan Sidwell <nathan@acm.org>
11784 * toplev.c (toplev::main): Remove excess parens on pretty_printer
11786 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
11788 2017-10-04 Sudakshina Das <sudi.das@arm.com>
11790 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
11791 check type for aarch64_simd_valid_immediate.
11792 (aarch64_output_simd_mov_immediate): Update prototype.
11793 (aarch64_simd_valid_immediate): Update prototype.
11794 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
11795 support for ORR-immediate.
11796 (and<mode>3): modified pattern to add support for BIC-immediate.
11797 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
11798 now checks for valid immediate for BIC and ORR based on new enum
11800 (aarch64_output_simd_mov_immediate): Function now used to output
11801 BIC/ORR imm as well based on new enum argument.
11802 * config/aarch64/constraints.md (Do): New vector immediate constraint.
11804 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
11805 (aarch64_reg_or_bic_imm): Likewise.
11807 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11809 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
11810 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
11812 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
11816 PR rtl-optimization/82396
11817 * haifa-sched.c (autopref_multipass_init): Simplify
11819 (autopref_rank_data): Simplify sort order.
11820 * sched-int.h (autopref_multipass_data_): Remove
11821 multi_mem_insn_p, min_offset and max_offset.
11823 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11825 * doc/sourcebuild.texi: Document vect_peeling_profitable.
11827 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11829 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
11830 vect_doubleint_cvt.
11832 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11834 * doc/sourcebuild.texi: Document vect_long_mult.
11836 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
11838 PR tree-optimization/82413
11839 * fold-const.c (build_range_check): Use widest_int when comparing
11840 the maximum ETYPE value with HIGH.
11842 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
11844 PR rtl-optimization/82396
11845 * haifa-sched.c (autopref_multipass_init): Simplify
11847 (autopref_rank_data): Simplify sort order.
11848 * sched-int.h (autopref_multipass_data_): Remove
11849 multi_mem_insn_p, min_offset and max_offset.
11851 2017-10-04 Jakub Jelinek <jakub@redhat.com>
11853 PR tree-optimization/82381
11854 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
11855 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
11858 PR tree-optimization/82374
11859 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
11860 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
11861 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
11862 current_function_decl to the new decl.
11864 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
11866 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
11867 helper macro for IEEE float128 hardware built-in functions.
11868 (SQRTF128_ODD): Add built-in functions with the round-to-odd
11870 (TRUNCF128_ODD): Likewise.
11871 (ADDF128_ODD): Likewise.
11872 (SUBF128_ODD): Likewise.
11873 (MULF128_ODD): Likewise.
11874 (DIVF128_ODD): Likewise.
11875 (FMAF128_ODD): Likewise.
11876 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
11877 UNSPEC_TRUNC_ROUND_TO_ODD.
11878 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
11879 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
11880 floating point round to odd instructions.
11881 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
11882 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
11883 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
11884 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
11885 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
11886 (trunc<mode>sf2_hw): Change the truncate with round to odd
11887 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
11888 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
11889 to odd hardware instructions.
11890 (sub<mode>3_odd): Likewise.
11891 (mul<mode>3_odd): Likewise.
11892 (div<mode>3_odd): Likewise.
11893 (sqrt<mode>2_odd): Likewise.
11894 (fma<mode>4_odd): Likewise.
11895 (fms<mode>4_odd): Likewise.
11896 (nfma<mode>4_odd): Likewise.
11897 (nfms<mode>4_odd): Likewise.
11898 (trunc<mode>df2_odd): Change the truncate with round to odd
11899 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
11901 * doc/extend.texi (PowerPC built-in functions): Update documentation
11902 for existing IEEE float128-bit built-in functions. Add built-in
11903 functions that generate the IEEE 128-bit floating point round to
11906 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
11908 PR rtl-optimization/77729
11909 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
11910 to (X&(C1&~C2))|C2 transformations.
11912 2017-10-03 Martin Jambor <mjambor@suse.cz>
11914 PR tree-optimization/82363
11915 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
11916 mismatch, mark lacc written regardless of racc.
11918 2017-10-03 Jakub Jelinek <jakub@redhat.com>
11920 PR tree-optimization/82381
11921 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
11922 stmt_to_insert nor wheather SSA_NAMEs are default defs.
11923 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
11924 fallthrough into reassoc_stmt_dominates_stmt_p.
11927 * combine.c (combine_instructions): Don't combine in unreachable
11930 2017-10-02 Peter Bergner <bergner@vnet.ibm.com>
11933 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
11934 function to not use the have_cpu variable. Do not set cpu_index,
11935 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
11936 or the default cpu.
11937 (rs6000_valid_attribute_p): Remove duplicate initializations of
11938 old_optimize and func_optimize.
11939 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
11940 (rs6000_activate_target_options): Make global.
11941 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
11944 2017-10-02 Jakub Jelinek <jakub@redhat.com>
11946 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
11947 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
11948 Set *poffset to 0 and *psize and *pmax_size to -1 if
11949 *poffset + *psize overflows in HOST_WIDE_INT.
11951 PR tree-optimization/82387
11952 PR tree-optimization/82388
11953 PR tree-optimization/82389
11954 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
11955 instead of live_bytes non-NULL.
11957 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
11960 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
11963 2017-10-02 Richard Biener <rguenther@suse.de>
11965 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
11966 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
11967 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
11969 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
11971 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
11972 requested precision matches the type's.
11973 * calls.c (alloc_max_size): Calculate the new candidate size as
11974 a widest_int and use wi::to_widest when comparing it with the
11975 current candidate size.
11976 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
11977 zero rather than integer_zero_node.
11978 * match.pd: Check for a no-op conversion before using wi::add
11979 rather than after. Use tree_to_uhwi when summing small shift
11980 counts into an unsigned int.
11982 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
11983 Alan Hayward <alan.hayward@arm.com>
11984 David Sherwood <david.sherwood@arm.com>
11987 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
11988 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
11989 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
11990 POINTER_AND_FP_REGS.
11992 2017-10-02 Richard Biener <rguenther@suse.de>
11994 PR tree-optimization/82355
11995 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
11996 a mapping for the enclosing loop but avoid generating one for
11997 the loop tree root.
11998 (copy_bb_and_scalar_dependences): Remove premature codegen
11999 error on PHIs in blocks duplicated into multiple places.
12000 * graphite-scop-detection.c
12001 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
12002 in the region use it as loop and nest to analyze the DR in.
12003 (try_generate_gimple_bb): Likewise.
12004 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
12005 (add_loop_constraints): For blocks in a loop not in the region
12006 create a dimension with a single iteration.
12007 * sese.h (gbb_loop_at_index): Remove assert.
12009 2017-10-01 Kevin Buettner <kevinb@redhat.com>
12011 * omp-expand.c (adjust_context_scope): New function.
12012 (expand_parallel_call): Call adjust_context_scope.
12014 2017-10-01 Jeff Law <law@redhat.com>
12016 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
12017 dom_opt_dom_walker class with direct access to private members.
12018 Add comments. Call test_for_singularity.
12019 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
12020 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
12021 m_dummy_cond anymore.
12022 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
12024 (pass_dominator:execute): Build the dummy_cond here and pass it
12025 to the dom_opt_dom_walker ctor.
12026 (test_for_singularity): New function.
12028 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
12029 Maya Rashish <coypu@sdf.org>
12031 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
12032 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
12033 (alpha*-*-netbsd*) Use nbsd_tm_file.
12034 (arm*-*-netbsdelf*) Likewise.
12035 (i[34567]86-*-netbsdelf*) Likewise.
12036 (x86_64-*-netbsd*) Likewise.
12037 (mips*-*-netbsd*) Likewise.
12038 (powerpc-*-netbsd*) Likewise.
12039 (sh*-*-netbsd*) Likewise.
12040 (sparc-*-netbsdelf*) Likewise.
12041 (sparc64-*-netbsd*) Likewise.
12042 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
12044 (vax-*-netbsdelf*) Likewise.
12045 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
12046 (UINT_FAST8_TYPE) Likewise.
12047 (INT_FAST16_TYPE) Check CHAR_FAST16.
12048 (UINT_FAST16_TYPE) Likewise.
12050 2017-09-30 Jakub Jelinek <jakub@redhat.com>
12053 * config/i386/i386.md
12054 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
12055 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
12056 *divmodsi4_zext_2): New define_insn_and_split.
12057 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
12058 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
12059 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
12060 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
12061 New define_insn_and_split.
12062 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
12063 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
12064 operands[1] having DImode when mode is SImode.
12066 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
12067 always SImode for DIV and MOD in REG_EQUAL notes.
12069 2017-09-29 Yury Gribov <tetra2005@gmail.com>
12071 PR middle-end/82319
12072 * match.pd: Fix handling of NaNs in pattern.
12074 2017-09-29 Jeff Law <law@redhat.com>
12076 * sbitmap.c (bitmap_bit_in_range_p): New function.
12077 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
12078 * tree-ssa-dse.c (live_bytes_read): New function.
12079 (dse_classify_store): Ignore reads of dead bytes.
12081 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
12082 typos and whitespace errors.
12083 * config/i386/predicates.md (address_no_seg_operand): Likewise.
12084 * config/s390/s390.c (s390_emit_prologue): Likewise.
12086 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
12089 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
12090 with a symbol for LRA.
12092 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
12094 PR rtl-optimization/82338
12095 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
12097 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
12099 * genmodes.c (calc_wider_mode): Suppress qsort macro.
12100 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
12101 (qsort_chk): Declare.
12102 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
12103 (qsort_chk): New function.
12105 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12107 PR tree-optimization/82337
12108 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
12109 phi definition if the PHI result appears in an abnormal PHI.
12110 (find_basis_for_base_expr): Don't record a basis if the LHS of the
12111 basis appears in an abnormal PHI.
12113 2017-09-29 Richard Biener <rguenther@suse.de>
12115 * graphite-isl-ast-to-gimple.c
12116 (translate_isl_ast_to_gimple::set_codegen_error): New function.
12117 (binary_op_to_tree): Use it.
12118 (get_rename_from_scev): Likewise.
12119 (copy_loop_phi_nodes): Likewise.
12120 (copy_bb_and_scalar_dependences): Likewise.
12121 (translate_pending_phi_nodes): Likewise.
12123 2017-09-29 Jakub Jelinek <jakub@redhat.com>
12126 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
12127 for movabsq $(i32 << shift), r64.
12129 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
12131 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
12132 index when encoding %esp as %rsp to avoid 0x67 prefix.
12134 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
12136 * config/i386/i386.md (*movsf_internal, *movdf_internal):
12137 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
12139 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
12141 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
12142 Extensions with more than 16 double VFP registers.
12143 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
12144 to_clear_mask and all code related to it. Replace the remaining
12145 entry by a sbitmap and adapt code accordingly.
12147 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
12149 * brig-builtins.def: Change pure attributes to const.
12151 2017-09-28 Joseph Myers <joseph@codesourcery.com>
12153 * config.gcc (default_gnu_indirect_function): Default to yes for
12154 sparc*-*-linux* with glibc.
12156 2017-09-28 Joseph Myers <joseph@codesourcery.com>
12158 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
12159 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
12160 when creating .init_array and .fini_array sections with priority
12163 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
12166 * config/aarch64/aarch64.c
12167 (aarch64_builtin_support_vector_misalignment): Always return false
12168 when misalignment is unknown.
12170 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
12172 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
12173 this function to return false if the definition used by the swap
12174 instruction is artificial, or if the memory address from which the
12175 constant value is loaded is not represented by a base address held
12176 in a register or if the base address register is a frame or stack
12177 pointer. Additionally, return false if the base address of the
12178 loaded constant is a SYMBOL_REF but is not considered to be a
12180 (replace_swapped_load_constant): New function.
12181 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
12182 loaded constant vector with a load of a swapped constant vector.
12184 2017-09-27 Carl Love <cel@us.ibm.com>
12186 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
12187 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
12188 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
12191 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
12193 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
12194 first, always call autopref_rank_data otherwise.
12196 2017-09-27 Richard Biener <rguenther@suse.de>
12198 * graphite-scop-detection.c (find_scop_parameters): Move
12199 loop bound handling ...
12200 (gather_bbs::before_dom_children): ... here, avoiding the need
12201 to build scop_info->loop_nest.
12202 (record_loop_in_sese): Remove.
12203 * sese.h (sese_info_t::loop_nest): Remove.
12204 * sese.c (new_sese_info): Do not allocate loop_nest.
12205 (free_sese_info): Do not free loop_nest.
12207 2017-09-27 Jakub Jelinek <jakub@redhat.com>
12210 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
12211 lhs from calls if the lhs has addressable type.
12213 2017-09-27 Richard Biener <rguenther@suse.de>
12215 * graphite.h (scop::max_alias_set): New member.
12216 * graphite-scop-detection.c: Remove references to non-existing
12217 --param in comments.
12218 (build_alias_sets): Record the maximum alias set used for drs.
12219 (build_scops): Support zero as unlimited for
12220 --param graphite-max-arrays-per-scop.
12221 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
12222 and inline into ...
12223 (build_poly_sr_1): ... here. Compute alias set based on the
12224 maximum alias set used for drs rather than
12225 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
12227 2017-09-27 Richard Biener <rguenther@suse.de>
12229 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
12230 --param loop-block-tile-size=0 to disable tiling.
12232 2017-09-27 Richard Biener <rguenther@suse.de>
12234 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
12235 (graphite-max-nb-scop-params): Document special value zero.
12236 * domwalk.h (dom_walker::STOP): New symbolical constant.
12237 (dom_walker::dom_walker): Add optional parameter for bb to
12239 (dom_walker::~dom_walker): Declare.
12240 (dom_walker::before_dom_children): Document STOP return value.
12241 (dom_walker::m_user_bb_to_rpo): New member.
12242 (dom_walker::m_bb_to_rpo): Likewise.
12243 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
12244 mapping here if not provided by the user.
12245 (dom_walker::~dom_walker): Free bb to RPO mapping if not
12246 provided by the user.
12247 (dom_walker::STOP): Define.
12248 (dom_walker::walk): Do not compute bb to RPO mapping here.
12249 Support STOP return value from before_dom_children to stop
12251 * graphite-optimize-isl.c (optimize_isl): If the schedule
12252 is the same still generate code if -fgraphite-identity
12253 or -floop-parallelize-all are given.
12254 * graphite-scop-detection.c: Include cfganal.h.
12255 (gather_bbs::gather_bbs): Get and pass through bb to RPO
12257 (gather_bbs::before_dom_children): Return STOP for BBs
12259 (build_scops): Compute bb to RPO mapping and pass it to
12260 the domwalk. Treat --param graphite-max-nb-scop-params=0
12261 as not limiting the number of params.
12262 * graphite.c (graphite_initialize): Remove limit on the
12263 number of basic-blocks in a function.
12264 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
12265 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
12266 default value of 10.
12268 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
12270 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
12271 Adjust code to eliminate needing to do the shift right 32-bits
12272 operation after XSCVDPSPN.
12274 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
12276 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
12277 ((X / Y) != 0 -> X >= Y): Likewise.
12279 2017-09-26 Carl Love <cel@us.ibm.com>
12281 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
12282 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
12283 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
12284 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
12285 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
12286 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
12287 definitions and overloading.
12288 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
12289 statement for P9V_BUILTIN_XST_LEN_R.
12290 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
12291 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
12292 define_expand and define_insn for the instructions and builtins.
12293 * doc/extend.texi: Update the built-in documentation file for the new
12294 built-in functions.
12295 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
12296 define_insn for the instructions
12298 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
12301 * gcc/config/netbsd-protos.h: New file.
12302 * gcc/config/netbsd.c: New file.
12303 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
12304 * gcc/config/t-netbsd: New file.
12305 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
12306 (tmake_file) Add t-netbsd.
12307 (extra_objs) Add netbsd.o.
12309 2017-09-26 Janus Weil <janus@gcc.gnu.org>
12313 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
12315 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
12317 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
12318 sign extension from a vector register to a GPR by doing a 32-bit
12319 direct move and then an EXTSW.
12320 (extendsi<mode>2 splitter): Likewise.
12321 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
12322 right or vector extract after doing XSCVDPSPN. Use
12323 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
12325 (movdi_from_sf_zero_ext): Likewise.
12326 (reload_gpr_from_vsxsf): Likewise.
12327 (p8_mfvsrd_4_disf): Delete, no longer used.
12328 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
12329 then needing to move the SFmode to a GPR to use the XSCVDPSP
12330 instruction instead of FRSP and XSCVDPSPN.
12331 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
12332 it is adjacent to the other XSCVSPDP insns.
12333 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
12334 SFmode to be in traditional Altivec registers.
12335 (vsx_xscvdpspn): Eliminate useless alternative constraint.
12336 (vsx_xscvspdpn): Likewise.
12337 (vsx_xscvspdpn_scalar): Likewise.
12339 2017-09-26 Martin Jambor <mjambor@suse.cz>
12341 * tree-sra.c (compare_access_positions): Put integral types first,
12342 stabilize sorting of integral types, remove conditions putting
12343 non-full-precision integers last.
12344 (sort_and_splice_var_accesses): Disable scalarization if a
12345 non-integert would be represented by a non-full-precision integer.
12347 2017-09-26 Joseph Myers <joseph@codesourcery.com>
12349 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
12350 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
12351 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
12352 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
12353 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
12354 conditionals inside the function instead of around it. Call
12355 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
12356 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
12358 2017-09-26 Richard Biener <rguenther@suse.de>
12360 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
12362 (scop_detection::build_scop_breadth): ... this. Removed.
12363 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
12364 (scop_detection::harmful_stmt_in_bb): Likewise.
12365 (scop_detection::graphite_can_represent_stmt): Likewise.
12366 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
12367 (scop_detection::can_represent_loop): Remove recursion, fold in ...
12368 (scop_detection::can_represent_loop_1): ... this. Removed.
12369 (scop_detection::harmful_loop_in_region): Simplify after inlining
12370 the above and remove more quadraticness.
12371 (build_scops): Adjust.
12372 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
12375 2017-09-26 Jakub Jelinek <jakub@redhat.com>
12378 * config/i386/i386.c (ix86_print_operand_address_as): Only test
12379 REGNO (base) == SP_REG if base is a REG.
12381 PR middle-end/35691
12382 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
12383 if it is different SSA_NAME.
12384 (optimize_range_tests_cmp_bitwise): New function.
12385 (optimize_range_tests): Call it.
12387 2017-09-26 Richard Biener <rguenther@suse.de>
12389 PR tree-optimization/82321
12390 * graphite.c (canonicalize_loop_closed_ssa): Properly check
12391 for the def being inside the loop.
12393 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12395 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
12397 * config/s390/s390-builtins.def: Fix constraint on op4.
12399 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12401 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
12402 independent expanders.
12403 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
12404 ("vec_ordered", "vec_unordered"): New expanders.
12406 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12408 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
12411 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12413 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
12414 vec_unpacks_lo_v16qi.
12415 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
12417 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12419 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
12420 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
12421 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
12423 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12425 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
12427 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
12429 ("*vec_slb<mode>"): New insn pattern.
12430 ("vec_shr_<mode>"): New expander.
12431 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
12432 and force the shift count operand to V16QImode.
12433 ("vec_srb<mode>"): Set shift count mode to V16QI.
12435 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12437 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
12438 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
12439 ("vec_widen_smult_hi_<mode>"): New expander definitions.
12441 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
12444 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
12446 2017-09-26 Richard Biener <rguenther@suse.de>
12448 PR tree-optimization/82320
12449 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
12452 2017-09-25 Jeff Law <law@redhat.com>
12454 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
12455 prototype for new argument.
12456 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
12457 mostly extracted from rs6000_emit_allocate_stack.
12458 (rs6000_emit_probe_stack_range_stack_clash): New function.
12459 (rs6000_emit_allocate_stack): Call
12460 rs6000_emit_probe_stack_range_stack_clash as needed.
12461 (rs6000_emit_probe_stack_range): Add additional argument
12462 to call to gen_probe_stack_range{si,di}.
12463 (output_probe_stack_range): New.
12464 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
12465 (output_probe_stack_range_stack_clash): New.
12466 (rs6000_emit_prologue): Emit notes into dump file as requested.
12467 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
12468 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
12469 Add additional operand and pass it to output_probe_stack_range.
12471 2017-09-25 Bin Cheng <bin.cheng@arm.com>
12473 PR tree-optimization/82163
12474 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
12475 (checking_verify_loop_closed_ssa): New parameter.
12476 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
12477 (check_loop_closed_ssa_stmt): Delete.
12478 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
12479 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
12480 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
12483 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
12485 * brig-builtins.def: Treat HSAIL barrier builtins as
12486 setjmp/longjump style functions.
12488 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
12490 * target.def (constant_alignment): New hook.
12491 * defaults.h (CONSTANT_ALIGNMENT): Delete.
12492 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
12493 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
12494 * doc/tm.texi: Regenerate.
12495 * targhooks.h (default_constant_alignment): Declare.
12496 (constant_alignment_word_strings): Likewise.
12497 * targhooks.c (default_constant_alignment): New function.
12498 (constant_alignment_word_strings): Likewise.
12499 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
12500 instead of CONSTANT_ALIGNMENT.
12501 * varasm.c (align_variable, get_variable_align, build_constant_desc)
12502 (force_const_mem): Likewise.
12503 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
12504 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
12505 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
12506 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12507 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
12509 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
12510 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12511 constant_alignment_word_strings.
12512 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
12513 (CONSTANT_ALIGNMENT): Likewise.
12514 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12515 (arm_constant_alignment): New function.
12516 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
12517 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12518 constant_alignment_word_strings.
12519 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
12520 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12521 constant_alignment_word_strings.
12522 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
12523 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12524 (cris_constant_alignment): New function.
12525 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
12526 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12527 (epiphany_constant_alignment): New function.
12528 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
12529 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12530 constant_alignment_word_strings.
12531 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
12532 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12533 constant_alignment_word_strings.
12534 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
12535 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12536 constant_alignment_word_strings.
12537 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
12538 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
12539 * config/i386/i386.c (ix86_constant_alignment): Make static.
12540 Use the same interface as the target hook.
12541 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12542 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
12543 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12544 constant_alignment_word_strings.
12545 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
12546 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
12547 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12548 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
12549 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12550 constant_alignment_word_strings.
12551 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
12552 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12553 constant_alignment_word_strings.
12554 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
12555 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12556 constant_alignment_word_strings.
12557 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
12558 * config/microblaze/microblaze.c (microblaze_constant_alignment):
12560 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12561 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
12562 * config/mips/mips.c (mips_constant_alignment): New function.
12563 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12564 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
12565 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
12566 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12567 (mmix_constant_alignment): Make static. Use the same interface
12568 as the target hook.
12569 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
12570 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12571 constant_alignment_word_strings.
12572 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
12573 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12574 constant_alignment_word_strings.
12575 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
12576 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12577 constant_alignment_word_strings.
12578 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
12579 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12580 (rs6000_constant_alignment): New function.
12581 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
12582 * config/riscv/riscv.c (riscv_constant_alignment): New function.
12583 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12584 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
12585 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12586 (rs6000_constant_alignment): New function.
12587 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
12588 * config/s390/s390.c (s390_constant_alignment): New function.
12589 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12590 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
12591 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12592 constant_alignment_word_strings.
12593 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
12594 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12595 (sparc_constant_alignment): New function.
12596 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
12597 * config/spu/spu.c (spu_constant_alignment): New function.
12598 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12599 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
12600 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12601 constant_alignment_word_strings.
12602 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
12603 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12604 constant_alignment_word_strings.
12605 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
12606 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12607 constant_alignment_word_strings.
12608 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
12609 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12610 (visium_constant_alignment): New function.
12611 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
12612 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12613 (xtensa_constant_alignment): New function.
12614 * system.h (CONSTANT_ALIGNMENT): Poison.
12616 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
12618 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
12619 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
12620 (rs6000_builtin_valid_without_lhs): New helper function.
12621 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12622 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
12624 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
12626 * target.h (vec_perm_indices): Use unsigned short rather than
12628 (auto_vec_perm_indices): Likewise.
12629 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
12630 Use unsigned int rather than unsigned char.
12631 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
12633 2017-09-25 Richard Biener <rguenther@suse.de>
12635 * cfgloop.h (sort_sibling_loops): Declare.
12636 * cfgloop.c (sort_sibling_loops_cmp): New helper.
12637 (sort_sibling_loops): New function sorting the sibling loop list
12639 * graphite.c (graphite_transform_loops): Sort sibling loops.
12641 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
12643 * target.def (vec_perm_const_ok): Change sel parameter to
12645 * optabs-query.c (can_vec_perm_p): Update accordingly.
12646 * doc/tm.texi: Regenerate.
12647 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
12648 auto_vec_perm_indices and remove separate nelt field.
12649 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
12650 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
12651 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
12652 (aarch64_expand_vec_perm_const): Update accordingly.
12653 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
12654 to vec_perm_indices.
12655 * config/arm/arm.c (expand_vec_perm_d): Change perm to
12656 auto_vec_perm_indices and remove separate nelt field.
12657 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
12658 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
12659 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
12661 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
12662 to vec_perm_indices.
12663 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
12664 sel to vec_perm_indices.
12665 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
12666 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
12667 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
12669 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
12672 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
12675 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
12676 on the FUNCTION_DECL function context if it has a DIE that is a
12679 2017-09-25 Richard Biener <rguenther@suse.de>
12681 PR tree-optimization/82285
12682 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
12685 2017-09-25 Tom de Vries <tom@codesourcery.com>
12689 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
12692 2017-09-25 Richard Biener <rguenther@suse.de>
12694 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
12695 ISL errors other than isl_error_quota happen. Dump if the
12696 schedule is the same.
12697 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
12698 errors instead of aborting inside ISL.
12700 2017-09-25 Iain Sandoe <iain@codesourcery.com>
12703 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
12704 of libgcc_eh for m64.
12705 * config/i386/darwin64.h: Likewise.
12707 2017-09-25 Richard Biener <rguenther@suse.de>
12709 PR middle-end/82144
12710 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
12711 attribute for incomplete types nor twice for complete ones.
12713 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
12716 * config/i386/i386.c (ix86_print_operand_address_as): Encode
12717 %esp as %rsp to avoid 0x67 prefix if there is no index or base
12720 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
12723 * config/i386/i386.opt (mprefer-avx256): Use
12724 ix86_target_flags variable.
12725 * config/i386/i386.c (ix86_target_string): Move
12726 -mprefer-avx256 to flag2_opts.
12728 2017-09-22 Jakub Jelinek <jakub@redhat.com>
12730 PR middle-end/35691
12731 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
12732 and x != -1 | y != -1 into (x & y) != -1.
12734 2017-09-22 Steve Ellcey <sellcey@cavium.com>
12736 * config.gcc: Add new case statement to set
12737 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
12738 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
12739 s390x-*-linux* case statements. Added aarch64 to the list of
12740 supported architectures.
12742 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
12744 PR tree-optimization/82289
12745 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
12746 STMT_VINFO_RELEVANT_P.
12748 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
12749 Alan Hayward <alan.hayward@arm.com>
12750 David Sherwood <david.sherwood@arm.com>
12752 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
12753 for VR_RANGE only; don't allow VR_ANTI_RANGE.
12754 (extract_range_from_binary_expr_1): Don't call
12755 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
12757 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
12758 Alan Hayward <alan.hayward@arm.com>
12759 David Sherwood <david.sherwood@arm.com>
12761 * target.def (preferred_vector_alignment): New hook.
12762 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
12764 * doc/tm.texi: Regenerate.
12765 * targhooks.h (default_preferred_vector_alignment): Declare.
12766 * targhooks.c (default_preferred_vector_alignment): New function.
12767 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
12769 (DR_TARGET_ALIGNMENT): New macro.
12770 (aligned_access_p): Update commentary.
12771 (vect_known_alignment_in_bytes): New function.
12772 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
12774 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
12775 Calculate the misalignment based on the target alignment rather than
12777 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
12778 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
12779 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
12780 the target alignment, rather than masking the element misalignment
12781 with the number of elements in a vector. Also use the target
12782 alignment when calculating the maximum number of peels.
12783 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
12784 instead of TYPE_ALIGN_UNIT.
12785 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
12786 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
12787 (vect_create_addr_base_for_vector_ref): Update call accordingly.
12788 (vect_create_data_ref_ptr): Likewise.
12789 (vect_setup_realignment): Realign by ANDing with
12790 -DR_TARGET_MISALIGNMENT.
12791 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
12792 the number of peels based on DR_TARGET_ALIGNMENT.
12793 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
12794 with the guaranteed alignment boundary when deciding whether
12796 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
12797 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
12798 (ensure_base_align): Remove stmt_info parameter. Get the
12799 target base alignment from DR_TARGET_ALIGNMENT.
12800 (vectorizable_store): Update call accordingly. Interpret
12801 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
12803 (vectorizable_load): Likewise.
12805 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
12806 Alan Hayward <alan.hayward@arm.com>
12807 David Sherwood <david.sherwood@arm.com>
12809 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
12810 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
12811 (vect_enhance_data_refs_alignment): Likewise.
12813 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
12815 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
12816 error. Only quit immediately if parsing is complete.
12817 (BEGIN): Initialize fatal_err and parse_done.
12818 (begin fpu, end fpu): Check number of arguments.
12819 (begin arch, end arch): Likewise.
12820 (begin cpu, end cpu): Likewise.
12821 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
12822 (optalias): Likewise.
12824 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
12826 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
12827 * config/arm/arm-isa.h: Delete. Move definitions to ...
12828 * arm-cpus.in: ... here. Use new feature and fgroup values.
12829 * config/arm/arm.c (arm_option_override): Use lower case for feature
12831 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
12832 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
12833 * config/arm/parsecpu.awk (END): Add new command 'isa'.
12835 (print_isa_bits_for): New function.
12836 (gen_isa): New function.
12837 (gen_comm_data): Use print_isa_bits_for.
12838 (define feature): New keyword.
12839 (define fgroup): New keyword.
12840 * config/arm/t-arm (TM_H): Remove.
12841 (GTM_H): Add arm-isa.h.
12842 (arm-isa.h): Add rule to generate file.
12843 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
12844 case for feature bit names.
12846 2017-09-22 Richard Biener <rguenther@suse.de>
12848 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
12850 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
12851 print of no dependency loops ...
12852 * graphite.c (graphite_transform_loops): ... here.
12853 (canonicalize_loop_closed_ssa_form): Work from inner to outer
12855 (same_close_phi_node, remove_duplicate_close_phi,
12856 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
12857 (canonicalize_loop_closed_ssa): ... here and simplify.
12858 * graphite-optimize-isl.c: Include tree-vectorizer.h.
12859 (optimize_isl): Use dump_printf_loc to tell when we stopped
12860 optimizing because of an ISL timeout.
12862 2017-09-22 Richard Biener <rguenther@suse.de>
12864 PR tree-optimization/82291
12865 * tree-if-conv.c (predicate_mem_writes): Make sure to
12866 remove writes in blocks predicated with false.
12868 2017-09-22 Richard Biener <rguenther@suse.de>
12870 * sese.c: Include cfganal.h.
12871 (if_region_set_false_region): Remove.
12872 (create_if_region_on_edge): Likewise.
12873 (move_sese_in_condition): Re-implement without destroying
12876 2017-09-22 Richard Biener <rguenther@suse.de>
12878 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
12879 Verify both BBs contain loop PHI nodes before dispatching to
12880 copy_loop_phi_args.
12881 (graphite_regenerate_ast_isl): Do not recompute dominators,
12882 do not verify three times. Restructure for clarity.
12883 * graphite-scop-detection.c (same_close_phi_node,
12884 remove_duplicate_close_phi, make_close_phi_nodes_unique,
12885 defined_in_loop_p, canonicalize_loop_closed_ssa,
12886 canonicalize_loop_closed_ssa_form): Simplify, remove excess
12887 checking and SSA rewrite, move to ...
12888 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
12889 (graphite_initialize): Do not pass in ctx, do not reset the
12890 SCEV cache, compute only dominators.
12891 (graphite_transform_loops): Allocate ISL ctx after
12892 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
12893 Maintain post-dominators only around build_scops.
12894 * sese.c (if_region_set_false_region): Make static. Free
12895 and recompute dominators.
12896 (move_sese_in_condition): Assert we don't get called with
12897 post-dominators computed.
12898 * sese.h (if_region_set_false_region): Remove.
12900 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
12902 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
12903 mode attribute for TARGET_AVX512VL.
12905 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
12907 * config/i386/i386.opt (mprefer-avx256): New option.
12908 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
12910 (ix86_preferred_simd_mode): Return 256-bit AVX modes
12911 for TARGET_PREFER_AVX256.
12912 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
12914 2017-09-21 Jeff Law <law@redhat.com>
12916 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
12917 Fix dump output if the only stack space is for pushed registers.
12919 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
12921 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
12924 2017-09-21 Martin Sebor <msebor@redhat.com>
12927 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
12928 code (in C++) or code that triggers warnings.
12930 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
12932 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
12934 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
12936 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
12937 * sched-rgn.c: Ditto.
12938 * sel-sched-ir.c: Ditto.
12940 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
12942 * toplev.h (set_random_seed): Adjust return type.
12943 * toplev.c (init_local_tick): Move eager initialization of random_seed
12944 to get_random_seed. Adjust comment.
12945 (init_random_seed): Inline to get_random_seed, delete.
12946 (get_random_seed): Initialize random_seed lazily.
12947 (set_random_seed): Do not return previous value.
12948 (print_switch_value): Do not call get_random_seed.
12950 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
12952 * cgraph.c (delete_function_version): New, broken out from...
12953 (cgraph_node::delete_function_version): ...here. Rename to
12954 cgraph_node::delete_function_version_by_decl. Update all uses.
12955 (cgraph_node::remove): Call delete_function_version.
12957 2017-09-21 Jakub Jelinek <jakub@redhat.com>
12960 * tree-inline.c (expand_call_inline): Emit clobber stmts for
12961 VAR_DECLs to which addressable non-volatile parameters are mapped
12962 and for id->retvar after the return value assignment. Clear
12963 id->retval and id->retbnd after inlining.
12965 2017-09-21 Richard Biener <rguenther@suse.de>
12967 PR tree-optimization/82276
12968 PR tree-optimization/82244
12969 * tree-vrp.c (build_assert_expr_for): Set
12970 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
12972 (remove_range_assertions): Revert earlier change.
12974 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
12977 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
12979 2017-09-21 Richard Biener <rguenther@suse.de>
12981 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
12982 Restore valid IL after code generation errors.
12983 * graphite.c (graphite_transform_loops): Diagnose code
12984 generation issues as MSG_MISSED_OPTIMIZATION and continue
12985 with processing SCOPs.
12987 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
12988 Alan Hayward <alan.hayward@arm.com>
12989 David Sherwood <david.sherwood@arm.com>
12991 * calls.c (compute_argument_addresses): Use simplify_gen_binary
12992 rather than choosing between plus_constant and gen_rtx_<CODE>.
12993 * expr.c (emit_push_insn): Likewise.
12994 (expand_expr_real_2): Likewise.
12996 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
12997 Alan Hayward <alan.hayward@arm.com>
12998 David Sherwood <david.sherwood@arm.com>
13000 * loop-unroll.c (split_iv): Call copy_rtx on the step.
13002 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13003 Alan Hayward <alan.hayward@arm.com>
13004 David Sherwood <david.sherwood@arm.com>
13006 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
13007 calling tree_to_uhwi.
13009 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13010 Alan Hayward <alan.hayward@arm.com>
13011 David Sherwood <david.sherwood@arm.com>
13013 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
13014 INTEGER_CST rather than a negative test for ADDR_EXPR.
13016 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13017 Alan Hayward <alan.hayward@arm.com>
13018 David Sherwood <david.sherwood@arm.com>
13020 * tree-vrp.c (extract_range_from_binary_expr_1): Check
13021 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
13023 2017-09-21 Richard Biener <rguenther@suse.de>
13025 PR tree-optimization/71351
13026 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
13027 graphite_create_new_loop_guard): Remove, fold remaining parts
13029 (translate_isl_ast_node_for): ... here and simplify.
13031 2017-09-21 Jakub Jelinek <jakub@redhat.com>
13034 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
13035 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
13036 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
13037 alternative always use QI mode, for -Os imov (=R,R) alternative
13038 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
13041 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13042 Jeff Law <law@redhat.com>
13044 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
13045 (allocate_stack_space): New function, partially extracted from
13046 s390_emit_prologue.
13047 (s390_emit_prologue): Track offset to most recent stack probe.
13048 Code to allocate space moved into allocate_stack_space.
13049 Dump actions when no stack is allocated.
13050 (s390_prologue_plus_offset): New function.
13051 (s390_emit_stack_probe): Likewise.
13053 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
13055 * common.opt (Wa, Wl, Wp, g, gz=): Add
13057 (gno-column-info): Remove.
13058 (gcolumn-info): Drop RejectNegative.
13059 (gno-): New prefix.
13060 (gno-record-gcc-switches): Remove.
13061 (grecord-gcc-switches): Drop RejectNegative.
13062 (gno-split-dwarf): Remove.
13063 (gsplit-dwarf): Drop RejectNegative.
13064 (gno-strict-dwarf): Remove.
13065 (gstrict-dwarf): Drop RejectNegative.
13066 * config/darwin.opt (gfull, gused): Add RejectNegative.
13067 * dwarf2out.c (gen_producer_string): Drop
13068 gno-record-gcc-switches handler.
13069 * optc-gen.awk: Add g to prefixes with negative forms.
13070 * opts-common.c (remapping_prefix_p): New.
13071 (find_opt): Check it.
13072 (generate_canonical_option): Test g prefix.
13073 (option_map): Add -gno- mapping.
13074 (add_misspelling_candidates): Check remapping_prefix_p.
13076 2017-09-20 Jeff Law <law@redhat.com>
13078 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
13079 thinko in stack clash protection support.
13081 * explow.c (compute_stack_clash_protection_loop_data): Use
13082 CONST_INT_P instead of explicit test. Verify object is a
13083 CONST_INT_P before looking at INTVAL.
13084 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
13085 instead of explicit test.
13087 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
13090 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
13091 address instead of to r1 and r11.
13093 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
13095 * config.gcc: Support "knm".
13096 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
13097 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
13099 * config/i386/i386.c (m_KNM): Define.
13100 (processor_target_table): Add "knm".
13102 (ix86_option_override_internal): Add "knm".
13103 (ix86_issue_rate): Add PROCESSOR_KNM.
13104 (ix86_adjust_cost): Ditto.
13105 (ia32_multipass_dfa_lookahead): Ditto.
13106 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
13107 (fold_builtin_cpu): Add M_INTEL_KNM.
13108 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
13109 (processor_type): Add PROCESSOR_KNM.
13110 * config/i386/x86-tune.def: Add m_KNM.
13111 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
13113 2017-09-20 Richard Biener <rguenther@suse.de>
13115 PR tree-optimization/80213
13116 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
13117 are allowed in empty BBs as well.
13118 (canonicalize_loop_closed_ssa): Also look for other complex
13120 (scop_detection::get_sese): Include the loop-closed PHI block
13122 (scop_detection::merge_sese): Remove code adding extra blocks.
13123 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
13124 (build_scops): Assert the final returned scop is invalid.
13126 2017-09-20 Richard Biener <rguenther@suse.de>
13128 PR tree-optimization/82264
13129 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
13131 (vn_phi_lookup): Likewise.
13132 (vn_phi_insert): Likewise.
13134 2017-09-20 Jakub Jelinek <jakub@redhat.com>
13136 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
13137 that fits into uhwi or shwi, add DW_AT_const_value regardless
13138 of early_dwarf without going through RTL, using add_AT_unsigned
13141 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
13142 (DEBUG_LTO_ABBREV_SECTION): Likewise.
13143 (DEBUG_LTO_MACINFO_SECTION): Likewise.
13144 (DEBUG_MACRO_SECTION): Likewise.
13145 (DEBUG_LTO_MACRO_SECTION): Likewise.
13146 (DEBUG_STR_DWO_SECTION): Likewise.
13147 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
13148 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
13149 (DEBUG_LTO_DWO_LINE_SECTION): Define.
13150 (DEBUG_LTO_LINE_STR_SECTION): Define.
13151 (init_sections_and_labels): Initialize debug_line_str_section
13152 variable. Initialize debug_loc_section for -gdwarf-5 to
13153 DEBUG_LOCLISTS_SECTION. Formatting fixes.
13155 2017-09-20 Richard Biener <rguenther@suse.de>
13157 * graphite-sese-to-poly.c (extract_affine): Properly handle
13158 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
13160 2017-09-20 Richard Biener <rguenther@suse.de>
13162 PR tree-optimization/81373
13163 * graphite-scop-detection.c (build_cross_bb_scalars_def):
13164 Force SESE live-out defs to be handled even if they are
13167 2017-09-19 Jeff Law <law@redhat.com>
13169 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
13170 nothing for stack adjustments with REG_STACK_CHECK.
13171 * sched-deps.c (parse_add_or_inc): Reject insns with
13172 REG_STACK_CHECK from dependency breaking.
13173 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
13174 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
13175 * reg-notes.def (STACK_CHECK): New note.
13177 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
13178 (ix86_expand_prologue): Dump stack clash info as needed.
13179 Call ix86_adjust_stack_and_probe_stack_clash as needed.
13181 * function.c (dump_stack_clash_frame_info): New function.
13182 * function.h (dump_stack_clash_frame_info): Prototype.
13183 (enum stack_clash_probes): New enum.
13185 * config/alpha/alpha.c (alpha_expand_prologue): Also check
13186 flag_stack_clash_protection.
13187 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
13188 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
13189 (arm_frame_pointer_required): Likewise.
13190 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
13191 (ia64_expand_prologue): Likewise.
13192 * config/mips/mips.c (mips_expand_prologue): Likewise.
13193 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
13194 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
13195 (sparc_flat_expand_prologue): Likewise.
13196 * config/spu/spu.c (spu_expand_prologue): Likewise.
13198 * explow.c: Include "params.h".
13199 (anti_adjust_stack_and_probe_stack_clash): New function.
13200 (get_stack_check_protect): Likewise.
13201 (compute_stack_clash_protection_loop_data): Likewise.
13202 (emit_stack_clash_protection_loop_start): Likewise.
13203 (emit_stack_clash_protection_loop_end): Likewise.
13204 (allocate_dynamic_stack_space): Use get_stack_check_protect.
13205 Use anti_adjust_stack_and_probe_stack_clash.
13206 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
13207 (emit_stack_clash_protection_loop_start): Likewise.
13208 (emit_stack_clash_protection_loop_end): Likewise.
13209 * rtl.h (get_stack_check_protect): Prototype.
13210 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
13211 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
13212 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
13214 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
13216 * doc/tm.texi: Rebuilt.
13217 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
13218 get_stack_check_protect.
13219 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
13220 * config/arm/arm.c (arm_expand_prologue): Likewise.
13221 (arm_frame_pointer_required): Likewise.
13222 * config/i386/i386.c (ix86_expand_prologue): Likewise.
13223 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
13224 * config/mips/mips.c (mips_expand_prologue): Likewise.
13225 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
13226 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
13227 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
13228 (sparc_flat_expand_prologue): Likewise.
13230 * common.opt (-fstack-clash-protection): New option.
13231 * flag-types.h (enum stack_check_type): Note difference between
13232 -fstack-check= and -fstack-clash-protection.
13233 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
13234 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
13235 * toplev.c (process_options): Issue warnings/errors for cases
13236 not handled with -fstack-clash-protection.
13237 * doc/invoke.texi (-fstack-clash-protection): Document new option.
13238 (-fstack-check): Note additional problem with -fstack-check=generic.
13239 Note that -fstack-check is primarily for Ada and refer users
13240 to -fstack-clash-protection for stack-clash-protection.
13241 Document new params for stack clash protection.
13243 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
13245 * config/i386/i386.c (ix86_split_long_move): Do not handle
13246 address used for LEA in a special way.
13248 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
13250 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
13252 2017-09-19 Martin Sebor <msebor@redhat.com>
13255 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
13256 of incompatible types.
13258 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
13260 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13261 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
13262 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13263 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
13265 2017-09-19 Richard Biener <rguenther@suse.de>
13267 PR tree-optimization/82244
13268 * tree-vrp.c (remove_range_assertions): Do not propagate
13269 a constant to abnormals but replace the assert with a copy.
13271 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
13273 PR rtl-optimization/57878
13274 PR rtl-optimization/68988
13275 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
13276 avoidance test involving non_reload_pseudos. Move frequency test
13277 below the general fragmentation avoidance test.
13279 2017-09-19 Richard Biener <rguenther@suse.de>
13281 PR tree-optimization/69728
13282 * graphite-sese-to-poly.c (schedule_error): New global.
13283 (add_loop_schedule): Handle empty domain by failing the
13285 (build_original_schedule): Handle schedule_error.
13287 2017-09-19 Richard Biener <rguenther@suse.de>
13289 * graphite-scop-detection.c (scop_detection::can_represent_loop):
13290 Do not iterate to sibling loops but only to siblings of inner
13293 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
13296 * config/m68k/m68k.md (moveq feeding equality comparison): Check
13297 that the registers are different.
13299 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
13301 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
13302 to processor_model and "amdfam17h" to arch_names_table.
13303 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
13305 2017-09-18 Jakub Jelinek <jakub@redhat.com>
13308 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
13310 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13311 Alan Hayward <alan.hayward@arm.com>
13312 David Sherwood <david.sherwood@arm.com>
13314 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
13316 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
13318 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
13319 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
13320 vect_schedule_slp_instance.
13321 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
13322 Update call to vect_slp_analyze_node_operations. Simplify return
13324 (vect_slp_analyze_bb_1): Update call accordingly.
13325 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
13326 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
13327 (vect_schedule_slp): Update call accordingly.
13329 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13330 Alan Hayward <alan.hayward@arm.com>
13331 David Sherwood <david.sherwood@arm.com>
13333 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
13334 with types that aren't in fact scalar.
13336 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13338 * tree-vect-slp.c (vect_record_max_nunits): New function,
13340 (vect_build_slp_tree_1): ...here.
13341 (vect_build_slp_tree_2): Call it for phis too.
13343 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13345 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
13346 to vect_get_vec_def_for_operand when getting the mask operand.
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-vect-loop.c (vectorizable_live_operation): Fix type of
13355 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13356 Alan Hayward <alan.hayward@arm.com>
13357 David Sherwood <david.sherwood@arm.com>
13359 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
13360 calculation for vector booleans.
13362 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13363 Alan Hayward <alan.hayward@arm.com>
13364 David Sherwood <david.sherwood@arm.com>
13366 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
13368 (vect_transform_stmt): ...here.
13369 (vect_analyze_stmt): Use it instead of calling
13370 vectorizable_live_operation directly.
13372 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
13374 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
13375 non-SIMT targets in acc vector loops.
13377 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
13379 * configure.ac: Add arc and check if assembler supports gdwarf2.
13380 * configure: Regenerate.
13382 2017-09-18 Richard Biener <rguenther@suse.de>
13384 PR tree-optimization/82220
13385 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
13386 epilogue niters from the min_profitable_iters compute.
13388 2017-09-18 Jakub Jelinek <jakub@redhat.com>
13391 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
13392 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
13393 (ix86_init_pic_reg): Revert 2017-09-01 changes.
13395 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
13398 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
13399 switching to a new text section.
13401 2017-09-18 Richard Biener <rguenther@suse.de>
13403 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
13405 (build_alias_set): Reject aliases with no access function.
13407 2017-09-18 Richard Biener <rguenther@suse.de>
13409 PR tree-optimization/79622
13410 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
13412 (build_cross_bb_scalars_use): Likewise.
13414 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
13416 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
13418 2017-09-18 Alan Modra <amodra@gmail.com>
13421 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
13422 stack_pointer_rtx for count 0. Update comments. Break up
13423 large rtl expression.
13425 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
13427 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
13428 Increase to 20 bytes.
13429 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
13430 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
13431 or avx version of the stub.
13433 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
13436 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
13437 compute the minimum stack alignment. Also update preferred stack
13438 boundary for leaf functions.
13440 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
13442 PR tree-optimization/82228
13443 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
13446 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
13448 * common/config/nds32/nds32-common.c
13449 (nds32_option_optimization_table): Refine formatting.
13450 (nds32_option_optimization_table): Use -fsched-pressure and
13451 -fomit-frame-pointer for specific optimization level.
13453 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
13455 * config/nds32/nds32.c: Refine formatting and comments.
13456 * config/nds32/nds32.h: Likewise.
13457 * config/nds32/nds32.md: Likewise.
13458 * config/nds32/nds32-cost.c: Likewise.
13459 * config/nds32/nds32-isr.c: Likewise.
13460 * config/nds32/nds32-md-auxiliary.c: Likewise.
13461 * config/nds32/nds32-multiple.md: Likewise.
13462 * config/nds32/nds32-predicates.c: Likewise.
13464 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
13465 Jakub Jelinek <jakub@redhat.com>
13467 Add support for -std=c++2a.
13468 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
13470 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
13472 2017-09-15 Steve Ellcey <sellcey@cavium.com>
13475 * doc/extend.texi (Common Function Attributes): Add
13476 references to ARM, AArch64, and S/390 specific attributes.
13477 (Function Specific Option Pragmas): Add AArch64 and S/390
13478 to list of back ends that support the target pragma.
13480 2017-09-15 Nathan Sidwell <nathan@acm.org>
13482 * doc/standards.texi: Fix C++17 description. Update URLs for
13485 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
13487 * common.opt (Wcast-align=strict): New warning option.
13488 * doc/invoke.texi: Document -Wcast-align=strict.
13490 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
13492 * cgraph.h (cgraph_thunk_info): Add comments.
13493 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
13494 assert for VIRTUAL_* arguments stricter.
13496 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
13498 PR tree-optimization/71026
13499 * match.pd: Move RDIV patterns from fold-const.c
13500 * fold-const.c (distribute_real_division): Removed.
13501 (fold_binary_loc): Remove calls to distribute_real_divison.
13503 2017-09-15 Jakub Jelinek <jakub@redhat.com>
13505 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
13506 c++1z and gnu++1z as deprecated. Change other references to
13507 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
13508 Change -Wc++1z-compat to -Wc++17-compat.
13509 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
13510 * dwarf2out.c (highest_c_language): Handle C++17.
13511 (gen_compile_unit_die): Likewise.
13513 2017-09-15 Jakub Jelinek <jakub@redhat.com>
13515 PR rtl-optimization/82192
13516 * combine.c (make_extraction): Don't look through non-paradoxical
13517 SUBREGs or TRUNCATE if pos + len is or might be bigger than
13520 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
13521 Alan Hayward <alan.hayward@arm.com>
13522 David Sherwood <david.sherwood@arm.com>
13524 * target.def (function_arg_offset): New hook.
13525 * targhooks.h (default_function_arg_offset): Declare.
13526 * targhooks.c (default_function_arg_offset): New function.
13527 * function.c (locate_and_pad_parm): Use
13528 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
13529 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
13530 (TARGET_FUNCTION_ARG_OFFSET): ...this.
13531 * doc/tm.texi: Regenerate.
13532 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
13533 * config/spu/spu.c (spu_function_arg_offset): New function.
13534 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
13535 * system.h (FUNCTION_ARG_OFFSET): Poison.
13537 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
13538 Alan Hayard <alan.hayward@arm.com>
13539 David Sherwood <david.sherwood@arm.com>
13541 * target.def (truly_noop_truncation): New hook.
13542 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
13543 than TRULY_NOOP_TRUNCATION.
13544 * hooks.h (hook_bool_uint_uint_true): Declare.
13545 * hooks.c (hook_bool_uint_uint_true): New function.
13546 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
13547 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
13548 * doc/tm.texi: Regenerate.
13549 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
13550 rather than TRULY_NOOP_TRUNCATION in comments.
13551 (simplify_comparison): Likewise.
13552 (record_truncated_value): Likewise.
13553 * expmed.c (extract_bit_field_1): Likewise.
13554 (extract_split_bit_field): Likewise.
13555 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
13556 instead of TRULY_NOOP_TRUNCATION.
13557 * function.c (assign_parm_setup_block): Likewise.
13558 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
13559 * rtlhooks.c: Include target.h.
13560 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
13561 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
13562 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
13563 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
13564 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
13565 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
13566 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
13567 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
13568 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
13569 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
13570 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
13571 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
13572 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
13573 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
13574 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
13575 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
13576 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
13577 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
13578 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
13579 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
13580 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
13581 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
13582 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
13583 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
13584 * config/mips/mips.c (mips_truly_noop_truncation): New function.
13585 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
13586 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
13587 rather than TRULY_NOOP_TRUNCATION in comments.
13588 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
13589 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
13590 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
13591 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
13592 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
13593 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
13594 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
13595 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
13596 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
13597 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
13598 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
13599 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
13600 rather than TRULY_NOOP_TRUNCATION in comments.
13601 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
13602 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
13603 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
13604 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
13605 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
13606 TRULY_NOOP_TRUNCATION condition.
13607 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
13608 (TRULY_NOOP_TRUNCATION): Delete.
13609 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
13610 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
13611 * config/spu/spu.c (spu_truly_noop_truncation): New function.
13612 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
13613 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
13614 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
13615 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
13616 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
13617 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
13618 rather than TRULY_NOOP_TRUNCATION in comments.
13619 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
13620 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
13621 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
13622 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
13623 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
13624 * system.h (TRULY_NOOP_TRUNCATION): Poison.
13626 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
13629 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
13630 (*cmp_ior): Likewise.
13631 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
13632 (*ior_scc_scc_cmp): Likewise.
13633 (*and_scc_scc): Likewise.
13634 (*and_scc_scc_cmp): Likewise.
13636 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
13637 Alan Hayard <alan.hayward@arm.com>
13638 David Sherwood <david.sherwood@arm.com>
13640 * target.def (can_change_mode_class): New hook.
13641 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
13642 (hard_regno_nregs): Likewise.
13643 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
13644 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
13645 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
13646 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
13647 (LOAD_EXTEND_OP): Update accordingly.
13648 * doc/tm.texi: Regenerate.
13649 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
13650 CANNOT_CHANGE_MODE_CLASS.
13651 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
13652 (REG_CAN_CHANGE_MODE_P): ...this new macro.
13653 * combine.c (simplify_set): Update accordingly.
13654 * emit-rtl.c (validate_subreg): Likewise.
13655 * recog.c (general_operand): Likewise.
13656 * regcprop.c (mode_change_ok): Likewise.
13657 * reload1.c (choose_reload_regs): Likewise.
13658 (inherit_piecemeal_p): Likewise.
13659 * rtlanal.c (simplify_subreg_regno): Likewise.
13660 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
13661 instead of CANNOT_CHANGE_MODE_CLASS.
13662 (reload_cse_simplify_operands): Likewise.
13663 * reload.c (push_reload): Use targetm.can_change_mode_class
13664 instead of CANNOT_CHANGE_MODE_CLASS.
13665 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
13666 REG_CANNOT_CHANGE_MODE_P.
13667 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13668 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
13669 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13670 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13671 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13672 (arm_can_change_mode_class): New function.
13673 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
13674 than CANNOT_CHANGE_MODE_CLASS in comments.
13675 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13676 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
13677 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
13678 (ix86_can_change_mode_class): ...this new function, inverting the
13679 sense of the return value.
13680 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13681 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13682 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13683 (ia64_can_change_mode_class): New function.
13684 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13685 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
13686 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
13687 (m32c_can_change_mode_class): ...this new function, inverting the
13688 sense of the return value.
13689 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13690 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13691 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
13692 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
13693 (mips_can_change_mode_class): ...this new function, inverting the
13694 sense of the return value.
13695 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13696 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13697 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13698 (msp430_can_change_mode_class): New function.
13699 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13700 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
13701 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13702 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13703 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13704 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
13705 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13706 (pa_cannot_change_mode_class): Replace with...
13707 (pa_can_change_mode_class): ...this new function, inverting the
13708 sense of the return value.
13709 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
13710 than CANNOT_CHANGE_MODE_CLASS in comments.
13711 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13712 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
13713 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13714 (pdp11_cannot_change_mode_class): Replace with...
13715 (pdp11_can_change_mode_class): ...this new function, inverting the
13716 sense of the return value.
13717 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13718 * config/powerpcspe/powerpcspe-protos.h
13719 (rs6000_cannot_change_mode_class_ptr): Delete.
13720 * config/powerpcspe/powerpcspe.c
13721 (rs6000_cannot_change_mode_class_ptr): Delete.
13722 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13723 (rs6000_option_override_internal): Assign to
13724 targetm.can_change_mode_class instead of
13725 rs6000_cannot_change_mode_class_ptr.
13726 (rs6000_cannot_change_mode_class): Replace with...
13727 (rs6000_can_change_mode_class): ...this new function, inverting the
13728 sense of the return value.
13729 (rs6000_debug_cannot_change_mode_class): Replace with...
13730 (rs6000_debug_can_change_mode_class): ...this new function.
13731 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13732 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
13733 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13734 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13735 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
13737 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
13738 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13739 (rs6000_option_override_internal): Assign to
13740 targetm.can_change_mode_class instead of
13741 rs6000_cannot_change_mode_class_ptr.
13742 (rs6000_cannot_change_mode_class): Replace with...
13743 (rs6000_can_change_mode_class): ...this new function, inverting the
13744 sense of the return value.
13745 (rs6000_debug_cannot_change_mode_class): Replace with...
13746 (rs6000_debug_can_change_mode_class): ...this new function.
13747 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13748 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
13749 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
13750 (s390_can_change_mode_class): ...this new function, inverting the
13751 sense of the return value.
13752 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13753 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13754 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
13755 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13756 (sh_cannot_change_mode_class): Replace with...
13757 (sh_can_change_mode_class): ...this new function, inverting the
13758 sense of the return value.
13759 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13760 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13761 (sparc_can_change_mode_class): New function.
13762 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13763 * config/spu/spu.c (spu_can_change_mode_class): New function.
13764 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13765 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13766 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13767 (visium_can_change_mode_class): New function.
13768 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
13770 2017-09-15 Richard Biener <rguenther@suse.de>
13772 PR tree-optimization/82217
13773 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
13774 but not undefined case.
13776 2017-09-15 Jakub Jelinek <jakub@redhat.com>
13779 * postreload.c (reload_cse_simplify_operands): Skip
13780 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
13782 2017-09-15 Richard Biener <rguenther@suse.de>
13784 PR tree-optimization/68823
13785 * graphite-scop-detection.c (build_alias_set): If we have a
13786 possible dependence check whether we can handle them by just
13787 looking at the DRs DR_ACCESS_FNs.
13788 (build_scops): If build_alias_set fails, fail the SCOP.
13790 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
13792 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
13793 to support float128 built-in functions that require the ISA 3.0
13795 (BU_FLOAT128_3_HW): Likewise.
13796 (SQRTF128): Add support for the IEEE 128-bit square root and fma
13797 built-in functions.
13798 (FMAF128): Likewise.
13800 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
13801 support for built-in functions that need the ISA 3.0 IEEE 128-bit
13802 floating point instructions.
13803 (rs6000_invalid_builtin): Likewise.
13804 (rs6000_builtin_mask_names): Likewise.
13805 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
13806 (RS6000_BTM_FLOAT128_HW): Likewise.
13807 (RS6000_BTM_COMMON): Likewise.
13808 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
13810 * doc/extend.texi (RS/6000 built-in functions): Document the
13811 IEEE 128-bit floating point square root and fused multiply-add
13812 built-in functions.
13814 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
13816 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
13817 reg (r2) isn't in the set of registers defined in the prologue.
13819 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13820 Alan Hayward <alan.hayward@arm.com>
13821 David Sherwood <david.sherwood@arm.com>
13823 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
13824 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
13825 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
13826 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
13827 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
13829 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
13830 max_vectorization_factor.
13831 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
13833 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13834 Alan Hayward <alan.hayward@arm.com>
13835 David Sherwood <david.sherwood@arm.com>
13837 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
13838 (vect_worthwhile_without_simd_p): Declare.
13839 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
13840 (vectorizable_reduction): Use it.
13841 * tree-vect-stmts.c (vectorizable_shift): Likewise.
13842 (vectorizable_operation): Likewise.
13844 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13845 Alan Hayward <alan.hayward@arm.com>
13846 David Sherwood <david.sherwood@arm.com>
13848 * tree-vectorizer.h (vect_get_num_copies): New function.
13849 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
13850 * tree-vect-loop.c (vectorizable_reduction): Likewise.
13851 (vectorizable_induction): Likewise.
13852 (vectorizable_live_operation): Likewise.
13853 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
13854 (vectorizable_bswap): Likewise.
13855 (vectorizable_call): Likewise.
13856 (vectorizable_conversion): Likewise.
13857 (vectorizable_assignment): Likewise.
13858 (vectorizable_shift): Likewise.
13859 (vectorizable_operation): Likewise.
13860 (vectorizable_store): Likewise.
13861 (vectorizable_load): Likewise.
13862 (vectorizable_condition): Likewise.
13863 (vectorizable_comparison): Likewise.
13864 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
13866 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13867 Alan Hayward <alan.hayward@arm.com>
13868 David Sherwood <david.sherwood@arm.com>
13870 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
13871 of vect_init_vector.
13873 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13874 Alan Hayward <alan.hayward@arm.com>
13875 David Sherwood <david.sherwood@arm.com>
13877 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
13878 an inline wrapper that provides a location.
13879 (gimple_build_vector): Likewise.
13880 * gimple-fold.c (gimple_build_vector_from_val): New function.
13881 (gimple_build_vector): Likewise.
13882 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
13883 functions to build the initial value. Always return a gimple value.
13884 (get_initial_defs_for_reduction): Likewise. Only compute
13886 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
13887 vect_init_vector on the results from get_initial_def(s)_for_reduction.
13888 (vectorizable_induction): Use gimple_build_vector rather than
13891 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13892 Alan Hayward <alan.hayward@arm.com>
13893 David Sherwood <david.sherwood@arm.com>
13895 * target.h (vec_perm_indices): New typedef.
13896 (auto_vec_perm_indices): Likewise.
13897 * optabs-query.h: Include target.h
13898 (can_vec_perm_p): Take a vec_perm_indices *.
13899 * optabs-query.c (can_vec_perm_p): Likewise.
13900 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
13901 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
13902 * tree-vect-generic.c (lower_vec_perm): Likewise.
13903 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
13904 (vect_grouped_load_supported): Likewise.
13905 (vect_shift_permute_load_chain): Likewise.
13906 (vect_permute_store_chain): Use auto_vec_perm_indices.
13907 (vect_permute_load_chain): Likewise.
13908 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
13909 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
13910 Update uses of can_vec_perm_p.
13911 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
13912 mode with a number of elements. Take a vec_perm_indices *.
13913 (vect_create_epilog_for_reduction): Update accordingly.
13914 Use auto_vec_perm_indices.
13915 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
13916 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
13917 (vect_transform_slp_perm_load): Likewise.
13918 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
13919 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
13920 (vect_gen_perm_mask_checked): Likewise.
13921 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
13922 (vect_gen_perm_mask_checked): Likewise.
13923 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
13924 (vectorizable_store): Likewise.
13925 (vectorizable_load): Likewise.
13926 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
13927 (vectorizable_bswap): Likewise.
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 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
13934 * tree.c (build_vector): Likewise.
13935 (build_vector_from_ctor): Update accordingly.
13936 (build_vector_from_val): Likewise.
13937 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
13938 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
13939 * tree-vect-generic.c (add_rshift): Likewise.
13940 (expand_vector_divmod): Likewise.
13941 (optimize_vector_constructor): Likewise.
13942 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
13943 (vect_transform_slp_perm_load): Likewise.
13944 (vect_schedule_slp_instance): Likewise.
13945 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
13946 (vectorizable_call): Likewise.
13947 (vect_gen_perm_mask_any): Likewise. Add elements in order.
13948 * expmed.c (make_tree): Likewise.
13949 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
13950 a vector passed to build_vector.
13951 (fold_convert_const): Likewise.
13952 (exact_inverse): Likewise.
13953 (fold_ternary_loc): Likewise.
13954 (fold_relational_const): Likewise.
13955 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
13956 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
13957 (const_unop): Likewise. Store the reduction accumulator in a
13958 variable rather than an array.
13959 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
13960 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
13961 the new vector, rather than constructing it after the input arrays.
13962 (native_interpret_vector): Use auto_vec<tree> when building
13963 a vector passed to build_vector. Add elements in order.
13964 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
13965 auto_vec<tree> when building a vector passed to build_vector.
13966 (vect_create_epilog_for_reduction): Likewise.
13967 (vectorizable_induction): Likewise.
13968 (get_initial_def_for_reduction): Likewise. Fix indentation of
13970 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
13972 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
13973 passed to build_vector.
13975 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13976 Alan Hayward <alan.hayward@arm.com>
13977 David Sherwood <david.sherwood@arm.com>
13979 * tree-core.h (tree_base::u): Add an "nelts" field.
13980 (tree_vector): Use VECTOR_CST_NELTS as the length.
13981 * tree.c (tree_size): Likewise.
13982 (make_vector): Initialize VECTOR_CST_NELTS.
13983 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
13984 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
13985 TYPE_VECTOR_SUBPARTS.
13986 * expr.c (const_vector_mask_from_tree): Consistently use "units"
13987 as the number of units, setting it from VECTOR_CST_NELTS.
13988 (const_vector_from_tree): Likewise.
13989 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
13990 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
13991 (fold_negate_expr_1): Likewise.
13992 (fold_convert_const): Likewise.
13993 (const_binop): Likewise. Differentiate the number of output and
13995 (const_unop): Likewise.
13996 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
13997 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
13998 in cases that did the opposite.
14000 2017-09-14 Richard Biener <rguenther@suse.de>
14002 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
14005 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
14007 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
14009 2017-09-14 Jakub Jelinek <jakub@redhat.com>
14012 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
14013 if and where to split a bb, except for splitting before debug insn
14014 sequences followed by non-label real insn. Delete debug insns
14015 in between basic blocks.
14017 * combine.c (make_compound_operation_int): Formatting fixes.
14019 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
14020 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14021 * config/netbsd.h (LINK_EH_SPEC): Likewise.
14022 * config/sol2.h (LINK_EH_SPEC): Likewise.
14023 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14024 * config/s390/linux.h (LINK_SPEC): Likewise.
14025 * config/freebsd.h (LINK_EH_SPEC): Likewise.
14026 * config/openbsd.h (LINK_EH_SPEC): Likewise.
14027 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14028 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
14029 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
14030 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14031 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
14032 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
14034 2017-09-13 Jakub Jelinek <jakub@redhat.com>
14036 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
14038 (ENDFILE_LINUX_SPEC): Likewise.
14039 (LINK_EH_SPEC): Likewise.
14040 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
14041 (LINK_OS_LINUX_SPEC32): Likewise.
14042 (LINK_OS_LINUX_SPEC64): Likewise.
14043 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
14044 (LINK_OS_LINUX_SPEC): Likewise.
14046 2017-09-13 Martin Liska <mliska@suse.cz>
14048 PR middle-end/82154
14049 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
14050 CASE_HIGH is NULL_TREE.
14052 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
14053 Alan Hayward <alan.hayward@arm.com>
14054 David Sherwood <david.sherwood@arm.com>
14056 * target.def (secondary_memory_needed): New hook.
14057 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
14058 instead of SECONDARY_MEMORY_NEEDED.
14059 (secondary_memory_needed_mode): Likewise.
14060 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
14061 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
14062 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
14063 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
14064 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
14065 * doc/tm.texi: Regenerate.
14066 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
14067 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
14068 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14069 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
14070 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
14071 * config/i386/i386.c (inline_secondary_memory_needed): Put the
14072 mode argument first and change the reg_class arguments to reg_class_t.
14073 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
14074 Make static. Update the call to inline_secondary_memory_needed.
14075 (ix86_register_move_cost): Update the call to
14076 inline_secondary_memory_needed.
14077 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14078 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
14080 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
14081 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
14083 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
14084 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
14085 * config/mips/mips.c (mips_secondary_memory_needed): Make static
14086 and match hook interface. Add comment from mips.h.
14087 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14088 * config/mmix/mmix.md (truncdfsf2): Refer to
14089 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
14091 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
14092 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
14093 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14094 (pa_secondary_memory_needed): New function.
14095 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
14096 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
14097 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14098 (pdp11_secondary_memory_needed): Make static and match hook interface.
14099 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
14100 * config/powerpcspe/powerpcspe-protos.h
14101 (rs6000_secondary_memory_needed_ptr): Delete.
14102 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
14104 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14105 (rs6000_option_override_internal): Assign to
14106 targetm.secondary_memory_needed rather than
14107 rs6000_secondary_memory_needed_ptr.
14108 (rs6000_secondary_memory_needed): Match hook interface.
14109 (rs6000_debug_secondary_memory_needed): Likewise.
14110 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
14111 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
14112 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
14113 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14114 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
14115 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
14117 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
14118 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14119 (rs6000_option_override_internal): Assign to
14120 targetm.secondary_memory_needed rather than
14121 rs6000_secondary_memory_needed_ptr.
14122 (rs6000_secondary_memory_needed): Match hook interface.
14123 (rs6000_debug_secondary_memory_needed): Likewise.
14124 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
14125 * config/s390/s390.c (s390_secondary_memory_needed): New function.
14126 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14127 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
14128 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14129 (sparc_secondary_memory_needed): New function.
14130 * lra-constraints.c (check_and_process_move): Refer to
14131 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
14133 (curr_insn_transform): Likewise.
14134 (process_alt_operands): Use targetm.secondary_memory_needed
14135 instead of TARGET_SECONDARY_MEMORY_NEEDED.
14136 (check_secondary_memory_needed_p): Likewise.
14137 (choose_split_class): Likewise.
14138 * reload.c: Unconditionally include code that was previously
14139 conditional on SECONDARY_MEMORY_NEEDED.
14140 (push_secondary_reload): Use targetm.secondary_memory_needed
14141 instead of TARGET_SECONDARY_MEMORY_NEEDED.
14142 (push_reload): Likewise.
14143 * reload1.c: Unconditionally include code that was previously
14144 conditional on SECONDARY_MEMORY_NEEDED.
14145 (choose_reload_regs): Use targetm.secondary_memory_needed
14146 instead of TARGET_SECONDARY_MEMORY_NEEDED.
14147 (gen_reload): Likewise.
14148 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
14150 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
14151 Alan Hayward <alan.hayward@arm.com>
14152 David Sherwood <david.sherwood@arm.com>
14154 * target.def (secondary_memory_needed_mode): New hook:
14155 * targhooks.c (default_secondary_memory_needed_mode): Declare.
14156 * targhooks.h (default_secondary_memory_needed_mode): New function.
14157 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
14158 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
14159 * doc/tm.texi: Regenerate.
14160 * lra-constraints.c (check_and_process_move): Use
14161 targetm.secondary_memory_needed_mode instead of
14162 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
14163 (curr_insn_transform): Likewise.
14164 * reload.c (get_secondary_mem): Likewise.
14165 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14166 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
14168 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14169 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14170 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
14171 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14172 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
14174 * config/powerpcspe/powerpcspe-protos.h
14175 (rs6000_secondary_memory_needed_mode): Delete.
14176 * config/powerpcspe/powerpcspe.c
14177 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14178 (rs6000_secondary_memory_needed_mode): Make static.
14179 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14180 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
14182 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
14184 (rs6000_secondary_memory_needed_mode): Make static.
14185 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14186 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
14187 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14188 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14189 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
14191 (sparc_secondary_memory_needed_mode): New function.
14192 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
14194 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
14196 * config/aarch64/constraints.md (Umq): New constraint.
14197 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
14199 (mov<mode>): Update condition.
14201 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14203 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
14204 when bitposition is the same.
14206 2017-09-13 Richard Biener <rguenther@suse.de>
14208 * dwarf2out.c (output_die_symbol): Remove.
14209 (output_die): Do not output a DIEs symbol.
14211 2017-09-13 Richard Biener <rguenther@suse.de>
14213 PR middle-end/82128
14214 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
14215 default-def to avoid breaking iterator update with the weird
14216 interaction with cgraph_update_edges_for_call_stmt_node.
14218 2017-09-13 Richard Biener <rguenther@suse.de>
14220 * tree-cfg.c (verify_gimple_assign_binary): Add verification
14221 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
14222 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
14223 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
14225 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
14227 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
14228 Disable pc relative literal load irrespective of
14229 TARGET_FIX_ERR_A53_84341 for default.
14231 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
14233 * config/sparc/sparc.c (output_return): Output the source location of
14234 the insn in the delay slot, if any.
14235 (output_sibcall): Likewise.
14237 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
14240 * common.opt (-static-pie): New alias.
14241 (shared): Negate static-pie.
14242 (-no-pie): Update help text.
14244 (static-pie): New option.
14245 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
14246 -static-pie support.
14247 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
14248 (LINK_EH_SPEC): Likewise.
14249 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14250 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
14251 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
14252 * gcc.c (LINK_COMMAND_SPEC): Likewise.
14253 (init_gcc_specs): Likewise.
14254 (init_spec): Likewise.
14255 (display_help): Update help message for -pie.
14256 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
14259 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
14261 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
14262 (movdi_aarch64): Likewise.
14263 (movti_aarch64): Likewise.
14265 2017-09-12 Simon Wright <simon@pushface.org>
14268 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
14269 calculation of the minor version, always output as 0.
14271 2017-09-12 Jakub Jelinek <jakub@redhat.com>
14274 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
14275 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
14276 on it early, rather than manual conversion late. For
14277 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
14278 instead of performing manual conversion.
14280 2017-09-12 Carl Love <cel@us.ibm.com>
14282 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
14283 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
14285 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
14286 VMULOSW): Add definitions.
14287 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14288 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
14289 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
14290 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
14291 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
14293 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
14295 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
14297 (movti_aarch64): Likewise.
14298 (movdf_aarch64): Likewise.
14299 (movtf_aarch64): Likewise.
14300 (load_pairdi): Likewise.
14301 (store_pairdi): Likewise.
14302 (load_pairdf): Likewise.
14303 (store_pairdf): Likewise.
14304 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
14305 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
14306 (ldr_got_small_<mode>): Likewise.
14307 (ldr_got_small_28k_<mode>): Likewise.
14308 (ldr_got_tiny): Likewise.
14309 * config/aarch64/iterators.md (ldst_sz): New.
14310 (ldpstp_sz): Likewise.
14311 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
14313 (thunderx_load): Split load_8 to load_16.
14314 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
14316 (thunderx2t99_storepair_basic): Split store_8 to store_16.
14317 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
14318 (xgene1_store_pair): Split store_8 to store_16.
14319 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
14320 (falkor_st_0_st_sd): Split store_8 to store_16.
14322 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
14324 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
14325 and store1/2/3/4 to store_4/8/12/16.
14326 * config/aarch64/aarch64.md: Update for rename.
14327 * config/arm/arm.md: Likewise.: Likewise.
14328 * config/arm/arm.c: Likewise.
14329 * config/arm/thumb1.md: Likewise.
14330 * config/arm/thumb2.md: Likewise.
14331 * config/arm/vfp.md: Likewise.
14332 * config/arm/arm-generic.md: Likewise.
14333 * config/arm/arm1020e.md: Likewise.
14334 * config/arm/arm1026ejs.md: Likewise.
14335 * config/arm/arm1136jfs.md: Likewise.
14336 * config/arm/arm926ejs.md: Likewise.
14337 * config/arm/cortex-a15.md: Likewise.
14338 * config/arm/cortex-a17.md: Likewise.
14339 * config/arm/cortex-a5.md: Likewise.
14340 * config/arm/cortex-a53.md: Likewise.
14341 * config/arm/cortex-a57.md: Likewise.
14342 * config/arm/cortex-a7.md: Likewise.
14343 * config/arm/cortex-a8.md: Likewise.
14344 * config/arm/cortex-a9.md: Likewise.
14345 * config/arm/cortex-m4.md: Likewise.
14346 * config/arm/cortex-m7.md: Likewise.
14347 * config/arm/cortex-r4.md: Likewise.
14348 * config/arm/exynos-m1.md: Likewise.
14349 * config/arm/fa526.md: Likewise.
14350 * config/arm/fa606te.md: Likewise.
14351 * config/arm/fa626te.md: Likewise.
14352 * config/arm/fa726te.md: Likewise.
14353 * config/arm/fmp626.md: Likewise.
14354 * config/arm/iwmmxt.md: Likewise.
14355 * config/arm/ldmstm.md: Likewise.
14356 * config/arm/marvell-pj4.md: Likewise.
14357 * config/arm/xgene1.md: Likewise.
14358 * config/aarch64/thunderx.md: Likewise.
14359 * config/aarch64/thunderx2t99.md: Likewise.
14360 * config/aarch64/falkor.md: Likewise.
14362 2017-09-12 Martin Liska <mliska@suse.cz>
14364 * attribs.c (private_lookup_attribute): New function.
14365 * attribs.h (private_lookup_attribute): Declared here.
14366 (lookup_attribute): Called from this place.
14368 2017-09-12 Richard Biener <rguenther@suse.de>
14370 PR tree-optimization/82157
14371 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
14372 stmts with side-effects.
14374 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14375 Alan Hayward <alan.hayward@arm.com>
14376 David Sherwood <david.sherwood@arm.com>
14378 * target.def (hard_regno_nregs): New hook.
14379 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
14380 * targhooks.h (default_hard_regno_nregs): Declare.
14381 * targhooks.c (default_hard_regno_nregs): New function.
14382 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
14383 (TARGET_HARD_REGNO_NREGS): ...this hook.
14384 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
14385 (CLASS_MAX_NREGS): Likewise.
14386 * doc/tm.texi: Regenerate.
14387 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
14388 instead of HARD_REGNO_NREGS.
14389 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
14390 HARD_REGNO_NREGS in the comment.
14391 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
14392 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
14393 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
14394 Return an unsigned int.
14395 (TARGET_HARD_REGNO_NREGS): Redefine.
14396 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
14397 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
14398 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
14399 (arc_hard_regno_nregs): New function.
14400 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
14401 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
14402 (arm_hard_regno_nregs): New function.
14403 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
14404 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
14405 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
14406 (TARGET_HARD_REGNO_NREGS): Redefine.
14407 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
14408 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
14409 (HARD_REGNO_NREGS): Delete.
14410 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
14411 (cr16_hard_regno_nregs): New function.
14412 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
14413 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
14414 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
14415 (cris_hard_regno_nregs): New function.
14416 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
14417 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
14418 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
14419 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
14420 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
14421 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
14422 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
14423 (frv_hard_regno_nregs): Make static. Take and return an
14425 (frv_class_max_nregs): Remove outdated copy of documentation.
14426 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
14427 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
14428 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
14429 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
14430 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
14431 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
14432 (TARGET_HARD_REGNO_NREGS): Redefine.
14433 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
14434 (CLASS_MAX_NREGS): Update comment.
14435 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
14436 (ia64_hard_regno_nregs): New function.
14437 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
14438 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
14439 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
14440 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
14441 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
14443 (m32c_hard_regno_nregs): Likewise. Make static.
14444 (TARGET_HARD_REGNO_NREGS): Redefine.
14445 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
14446 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
14447 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
14448 (m68k_hard_regno_nregs): New function.
14449 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
14450 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
14451 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
14452 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
14453 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
14454 Take and return an unsigned int.
14455 (TARGET_HARD_REGNO_NREGS): Redefine.
14456 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
14457 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
14458 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
14459 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
14460 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
14461 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
14462 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
14463 (msp430_hard_regno_nregs): Make static. Take and return an
14465 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
14466 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
14467 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
14468 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
14469 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
14470 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
14471 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
14472 (TARGET_HARD_REGNO_NREGS): Redefine.
14473 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
14474 (PA_HARD_REGNO_NREGS): ...this.
14475 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
14476 (PA_HARD_REGNO_NREGS): ...this.
14477 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
14478 (pa_hard_regno_nregs): New function.
14479 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
14480 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
14481 (pdp11_hard_regno_nregs): New function.
14482 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
14483 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
14484 (rs6000_hard_regno_nregs_hook): New function.
14485 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
14486 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
14487 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
14488 Take and return an unsigned int. Move earlier in file.
14489 (TARGET_HARD_REGNO_NREGS): Redefine.
14490 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
14491 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
14492 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
14493 (rl78_hard_regno_nregs): Make static. Take and return an
14495 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
14496 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
14497 (rs6000_hard_regno_nregs_hook): New function.
14498 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
14499 * config/rx/rx.c (rx_hard_regno_nregs): New function.
14500 (TARGET_HARD_REGNO_NREGS): Redefine.
14501 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
14502 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
14503 instead of HARD_REGNO_NREGS.
14504 (s390_hard_regno_nregs): New function.
14505 (s390_hard_regno_mode_ok): Add comment from s390.h.
14506 (TARGET_HARD_REGNO_NREGS): Redefine.
14507 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
14508 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
14509 (sh_hard_regno_nregs): New function.
14510 (sh_pass_in_reg_p): Use it.
14511 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
14512 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
14513 (sparc_hard_regno_nregs): New function.
14514 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
14515 * config/spu/spu.c (spu_hard_regno_nregs): New function.
14516 (spu_function_arg_advance): Use it, supplying a valid register number.
14517 (TARGET_HARD_REGNO_NREGS): Redefine.
14518 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
14519 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
14520 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
14521 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
14522 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
14523 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
14524 (CLASS_MAX_NREGS): Remove copy of old documentation.
14525 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
14526 (visium_hard_regno_nregs): New function.
14527 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
14528 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
14529 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
14530 xtensa_hard_regno_nregs): New function.
14531 * system.h (HARD_REGNO_NREGS): Poison.
14533 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14535 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
14536 hard_regno_nregs instead of HARD_REGNO_NREGS.
14537 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
14538 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
14539 (c6x_expand_epilogue): Likewise.
14540 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
14541 (frv_read_iacc_argument): Likewise.
14542 * config/sh/sh.c: Include regs.h.
14543 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
14544 (regs_used): Likewise.
14545 (output_stack_adjust): Likewise.
14546 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
14547 * expmed.c: Include regs.h.
14548 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
14549 * ree.c: Include regs.h.
14550 (combine_reaching_defs): Use hard_regno_nregs instead of
14552 (add_removable_extension): Likewise.
14554 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14556 * regs.h (hard_regno_nregs): Turn into a function.
14557 (end_hard_regno): Update accordingly.
14558 * caller-save.c (setup_save_areas): Likewise.
14559 (save_call_clobbered_regs): Likewise.
14560 (replace_reg_with_saved_mem): Likewise.
14561 (insert_restore): Likewise.
14562 (insert_save): Likewise.
14563 * combine.c (can_change_dest_mode): Likewise.
14564 (move_deaths): Likewise.
14565 (distribute_notes): Likewise.
14566 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
14567 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
14568 (rs6000_split_multireg_move): Likewise.
14569 (rs6000_register_move_cost): Likewise.
14570 (rs6000_memory_move_cost): Likewise.
14571 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
14572 (rs6000_split_multireg_move): Likewise.
14573 (rs6000_register_move_cost): Likewise.
14574 (rs6000_memory_move_cost): Likewise.
14575 * cselib.c (cselib_reset_table): Likewise.
14576 (cselib_lookup_1): Likewise.
14577 * emit-rtl.c (set_mode_and_regno): Likewise.
14578 * function.c (aggregate_value_p): Likewise.
14579 * ira-color.c (setup_profitable_hard_regs): Likewise.
14580 (check_hard_reg_p): Likewise.
14581 (calculate_saved_nregs): Likewise.
14582 (assign_hard_reg): Likewise.
14583 (improve_allocation): Likewise.
14584 (calculate_spill_cost): Likewise.
14585 * ira-emit.c (modify_move_list): Likewise.
14586 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
14587 (ira_hard_reg_in_set_p): Likewise.
14588 * ira.c (setup_reg_mode_hard_regset): Likewise.
14589 (clarify_prohibited_class_mode_regs): Likewise.
14590 (check_allocation): Likewise.
14591 * lra-assigns.c (find_hard_regno_for_1): Likewise.
14592 (lra_setup_reg_renumber): Likewise.
14593 (setup_try_hard_regno_pseudos): Likewise.
14594 (spill_for): Likewise.
14595 (assign_hard_regno): Likewise.
14596 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
14597 * lra-constraints.c (in_class_p): Likewise.
14598 (lra_constraint_offset): Likewise.
14599 (simplify_operand_subreg): Likewise.
14600 (lra_constraints): Likewise.
14601 (split_reg): Likewise.
14602 (split_if_necessary): Likewise.
14603 (invariant_p): Likewise.
14604 (inherit_in_ebb): Likewise.
14605 * lra-lives.c (process_bb_lives): Likewise.
14606 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
14607 (get_hard_regs): Likewise.
14608 (do_remat): Likewise.
14609 * lra-spills.c (assign_spill_hard_regs): Likewise.
14610 * mode-switching.c (create_pre_exit): Likewise.
14611 * postreload.c (reload_combine_recognize_pattern): Likewise.
14612 * recog.c (peep2_find_free_register): Likewise.
14613 * regcprop.c (kill_value_regno): Likewise.
14614 (set_value_regno): Likewise.
14615 (copy_value): Likewise.
14616 (maybe_mode_change): Likewise.
14617 (find_oldest_value_reg): Likewise.
14618 (copyprop_hardreg_forward_1): Likewise.
14619 * regrename.c (check_new_reg_p): Likewise.
14620 (regrename_do_replace): Likewise.
14621 * reload.c (push_reload): Likewise.
14622 (combine_reloads): Likewise.
14623 (find_dummy_reload): Likewise.
14624 (operands_match_p): Likewise.
14625 (find_reloads): Likewise.
14626 (find_equiv_reg): Likewise.
14627 (reload_adjust_reg_for_mode): Likewise.
14628 * reload1.c (count_pseudo): Likewise.
14629 (count_spilled_pseudo): Likewise.
14630 (find_reg): Likewise.
14631 (clear_reload_reg_in_use): Likewise.
14632 (free_for_value_p): Likewise.
14633 (allocate_reload_reg): Likewise.
14634 (choose_reload_regs): Likewise.
14635 (reload_adjust_reg_for_temp): Likewise.
14636 (emit_reload_insns): Likewise.
14637 (delete_output_reload): Likewise.
14638 * rtlanal.c (subreg_get_info): Likewise.
14639 * sched-deps.c (sched_analyze_reg): Likewise.
14640 * sel-sched.c (init_regs_for_mode): Likewise.
14641 (mark_unavailable_hard_regs): Likewise.
14642 (choose_best_reg_1): Likewise.
14643 (verify_target_availability): Likewise.
14644 * valtrack.c (dead_debug_insert_temp): Likewise.
14645 * var-tracking.c (track_loc_p): Likewise.
14646 (emit_note_insn_var_location): Likewise.
14647 * varasm.c (make_decl_rtl): Likewise.
14648 * reginfo.c (choose_hard_reg_mode): Likewise.
14649 (init_reg_modes_target): Refer directly to
14650 this_target_regs->x_hard_regno_nregs.
14652 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14654 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
14655 instead of hard_regno_nregs.
14657 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14659 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
14660 end_hard_regno instead of hard_regno_nregs.
14661 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
14662 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
14663 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
14664 * ira-color.c (improve_allocation): Likewise.
14665 * lra-assigns.c (find_hard_regno_for_1): Likewise.
14666 * lra-lives.c (mark_regno_live): Likewise.
14667 (mark_regno_dead): Likewise.
14668 * lra-remat.c (operand_to_remat): Likewise.
14669 * lra.c (collect_non_operand_hard_regs): Likewise.
14670 * postreload.c (reload_combine_note_store): Likewise.
14671 (move2add_valid_value_p): Likewise.
14672 * reload.c (regno_clobbered_p): Likewise.
14674 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14676 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
14678 * config/v850/v850.c (v850_reorg): Likewise.
14679 * reload.c (refers_to_regno_for_reload_p): Likewise.
14680 (find_equiv_reg): Likewise.
14681 * reload1.c (reload_reg_reaches_end_p): Likewise.
14683 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14685 * caller-save.c (add_used_regs): Use REG_NREGS instead of
14687 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
14688 * config/arm/arm.c (output_move_neon): Likewise.
14689 (arm_attr_length_move_neon): Likewise.
14690 (neon_split_vcombine): Likewise.
14691 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
14692 (c6x_mark_reg_written): Likewise.
14693 (c6x_dwarf_register_span): Likewise.
14694 * config/i386/i386.c (ix86_save_reg): Likewise.
14695 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
14696 (rws_access_reg): Likewise.
14697 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
14698 * mode-switching.c (create_pre_exit): Likewise.
14699 * ree.c (combine_reaching_defs): Likewise.
14700 (add_removable_extension): Likewise.
14701 * regcprop.c (find_oldest_value_reg): Likewise.
14702 (copyprop_hardreg_forward_1): Likewise.
14703 * reload.c (reload_inner_reg_of_subreg): Likewise.
14704 (push_reload): Likewise.
14705 (combine_reloads): Likewise.
14706 (find_dummy_reload): Likewise.
14707 (reload_adjust_reg_for_mode): Likewise.
14708 * reload1.c (find_reload_regs): Likewise.
14709 (forget_old_reloads_1): Likewise.
14710 (reload_reg_free_for_value_p): Likewise.
14711 (reload_adjust_reg_for_temp): Likewise.
14712 (emit_reload_insns): Likewise.
14713 (delete_output_reload): Likewise.
14714 * sel-sched.c (choose_best_reg_1): Likewise.
14715 (choose_best_pseudo_reg): Likewise.
14717 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14718 Alan Hayward <alan.hayward@arm.com>
14719 David Sherwood <david.sherwood@arm.com>
14721 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
14722 * target.def (slow_unaligned_access): New hook.
14723 * targhooks.h (default_slow_unaligned_access): Declare.
14724 * targhooks.c (default_slow_unaligned_access): New function.
14725 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
14726 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
14727 * doc/tm.texi: Regenerate.
14728 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
14729 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
14730 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
14732 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
14733 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
14735 (rs6000_slow_unaligned_access): New function.
14736 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
14737 (expand_block_compare): Likewise.
14738 (expand_strn_compare): Likewise.
14739 (rs6000_rtx_costs): Likewise.
14740 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
14741 (riscv_slow_unaligned_access): Likewise.
14742 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
14743 (riscv_slow_unaligned_access_p): ...this and make static.
14744 (riscv_option_override): Update accordingly.
14745 (riscv_slow_unaligned_access): New function.
14746 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
14747 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
14748 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
14749 (rs6000_slow_unaligned_access): New function.
14750 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
14751 (rs6000_rtx_costs): Likewise.
14752 * config/rs6000/rs6000-string.c (expand_block_compare)
14753 (expand_strn_compare): Use targetm.slow_unaligned_access instead
14754 of SLOW_UNALIGNED_ACCESS.
14755 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
14756 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
14757 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
14758 of SLOW_UNALIGNED_ACCESS.
14759 * expmed.c (simple_mem_bitfield_p): Likewise.
14760 * expr.c (alignment_for_piecewise_move): Likewise.
14761 (emit_group_load_1): Likewise.
14762 (emit_group_store): Likewise.
14763 (copy_blkmode_from_reg): Likewise.
14764 (emit_push_insn): Likewise.
14765 (expand_assignment): Likewise.
14766 (store_field): Likewise.
14767 (expand_expr_real_1): Likewise.
14768 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
14769 * lra-constraints.c (simplify_operand_subreg): Likewise.
14770 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
14771 * gimple-ssa-store-merging.c: Likewise in block comment at start
14773 * tree-ssa-strlen.c: Include target.h.
14774 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
14775 of SLOW_UNALIGNED_ACCESS.
14776 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
14778 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14780 PR rtl-optimization/82185
14781 * expmed.c (emit_store_flag_int): Only test tem if it has been
14784 2017-09-12 Richard Biener <rguenther@suse.de>
14786 PR middle-end/82149
14787 * match.pd ((FTYPE) N CMP CST): Fix typo.
14789 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
14791 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
14793 (mips_near_type_p): Add 'short_call' attribute as a synonym
14795 * doc/extend.texi (short_call): Document new function attribute.
14797 2017-09-12 Jakub Jelinek <jakub@redhat.com>
14800 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
14801 assertion check that in the condition.
14802 (get_atomic_generic_size): Likewise. Before testing if parameter
14803 has pointer type, if it has array type, call for C++
14804 default_conversion to perform array-to-pointer conversion.
14806 2017-09-12 Richard Biener <rguenther@suse.de>
14808 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
14809 for operations we cannot scalarize.
14811 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
14813 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
14814 vectors heap vectors. Clean up comments.
14815 Make visited_bbs a reference.
14816 (profitable_jump_thread_path): Make GC
14817 vectors heap vectors. Clean up comments.
14819 (convert_and_register_jump_thread_path): Make GC vectors heap
14821 (check_subpath_and_update_thread_path): Same. Clean up comments.
14822 Make visited_bbs a reference.
14823 (handle_phi): Abstract common code to to
14824 register_jump_thread_path_if_profitable.
14825 Rename VAR_BB to DEF_BB.
14827 Make GC vectors heap vectors.
14828 Make visited_bbs a reference.
14829 (handle_assignment): Same.
14830 (register_jump_thread_path_if_profitable): New.
14831 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
14833 Make GC vectors heap vectors. Clean up comments.
14834 Make visited_bbs a reference.
14835 (find_jump_threads_backwards): Make visited_bbs live in the stack.
14836 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
14839 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
14842 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
14843 words of E_DImode object are reachable by xtensa_uimm8x4 access.
14845 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
14847 Revert r251800 and r251799.
14849 2017-09-11 Martin Jambor <mjambor@suse.cz>
14852 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
14853 arguments in advance.
14854 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
14855 use it to find predecessor edges.
14856 (naive_outof_ssa): Collect vector of predecessors.
14858 2017-09-08 Jason Merrill <jason@redhat.com>
14860 PR c++/70029 - ICE with ref-qualifier and -flto
14861 * langhooks.h (struct lang_hooks_for_types): Add
14862 copy_lang_qualifiers.
14863 * attribs.c (build_type_attribute_qual_variant): Use it.
14864 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
14866 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
14867 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
14869 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
14872 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
14873 (*mulsi3_sp64): New instruction.
14874 (mulsi3): New expander.
14876 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
14878 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
14880 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
14882 * sancov.c: Include memmodel.h.
14884 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
14887 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
14890 2017-09-07 Carl Love <cel@us.ibm.com>
14892 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
14893 the sldi instruction.
14895 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
14897 * sancov.c: Include tm_p.h.
14899 2017-09-07 Jakub Jelinek <jakub@redhat.com>
14902 * output.h (switch_to_other_text_partition): New declaration.
14903 * varasm.c (switch_to_other_text_partition): New function.
14904 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
14905 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
14906 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
14907 to the other text partition before emitting LCL label and switch back
14908 after emitting the word after it.
14910 2017-09-07 Richard Biener <rguenther@suse.de>
14912 * passes.def (pass_split_crit_edges): Remove instance before PRE.
14913 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
14914 critical edges here, after loop init.
14915 (pass_data_pre): Remove PROP_no_crit_edges flags.
14916 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
14917 for valueization of call args to avoid leaking VN_TOP.
14918 (visit_use): Assert we do not visit default defs.
14919 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
14920 Use error_mark_node to more easily detect leaking VN_TOP.
14921 All default-defs are varying, not VN_TOP. Mark them visited.
14922 (run_scc_vn): Make code match comment.
14924 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
14926 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
14927 OPTION_MASK_FLOAT128_KEYWORD.
14928 (POWERPC_MASKS): Likewise.
14929 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
14930 support for the -mfloat128-type option, and make -mfloat128
14931 default on PowerPC Linux systems. Define or undefine
14932 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
14933 Define __float128 to be __ieee128 if IEEE 128-bit support is
14934 enabled, or undefine it.
14935 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
14936 Delete defining __FLOAT128_TYPE__.
14937 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
14938 -mfloat128-type option and make -mfloat128 default on PowerPC
14940 (TARGET_FLOAT128_TYPE): Likewise.
14941 (-mfloat128-type): Likewise.
14942 * config/rs6000/rs6000.c (rs6000_option_override_internal):
14943 Delete the -mfloat128-type option and make -mfloat128 default on
14944 PowerPC Linux systems. Always use __ieee128 to be the keyword for
14945 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
14946 128-bit floating point is enabled. Change tests from using
14947 -mfloat128-type to -mfloat128.
14948 (rs6000_mangle_type): Use the correct mangling for the __float128
14949 type even if normal long double is restricted to 64-bits.
14950 (floatn_mode): Enable the _Float128 type by default on VSX Linux
14952 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
14953 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
14954 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
14956 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
14957 documentation for -mfloat128.
14959 2017-09-06 Olivier Hainque <hainque@adacore.com>
14961 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
14963 2017-09-06 Wish Wu <wishwu007@gmail.com>
14964 Jakub Jelinek <jakub@redhat.com>
14966 * asan.c (initialize_sanitizer_builtins): Add
14967 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
14968 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
14969 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
14970 BT_FN_VOID_UINT64_PTR variables.
14971 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
14972 (BT_FN_VOID_UINT16_UINT16): Likewise.
14973 (BT_FN_VOID_UINT32_UINT32): Likewise.
14974 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
14975 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
14976 (BT_FN_VOID_UINT64_PTR): Likewise.
14977 * common.opt (flag_sanitize_coverage): New variable.
14978 (fsanitize-coverage=trace-pc): Remove.
14979 (fsanitize-coverage=): Add.
14980 * flag-types.h (enum sanitize_coverage_code): New enum.
14981 * fold-const.c (fold_range_test): Disable non-short-circuit
14982 optimization if flag_sanitize_coverage.
14983 (fold_truth_andor): Likewise.
14984 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
14985 * opts.c (COVERAGE_SANITIZER_OPT): Define.
14986 (coverage_sanitizer_opts): New array.
14987 (get_closest_sanitizer_option): Add OPTS argument, handle also
14988 OPT_fsanitize_coverage_.
14989 (parse_sanitizer_options): Adjusted to also handle
14990 OPT_fsanitize_coverage_.
14991 (common_handle_option): Add OPT_fsanitize_coverage_.
14992 * sancov.c (instrument_comparison, instrument_switch): New function.
14993 (sancov_pass): Add trace-cmp support.
14994 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
14995 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
14996 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
14997 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
14998 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
14999 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
15000 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
15001 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
15002 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
15003 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
15005 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
15007 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
15008 error. Only quit immediately if parsing is complete.
15009 (BEGIN): Initialize fatal_err and parse_done.
15010 (begin fpu, end fpu): Check number of arguments.
15011 (begin arch, end arch): Likewise.
15012 (begin cpu, end cpu): Likewise.
15013 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
15014 (optalias): Likewise.
15016 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
15018 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
15019 * config/arm/arm-isa.h: Delete. Move definitions to ...
15020 * arm-cpus.in: ... here. Use new feature and fgroup values.
15021 * config/arm/arm.c (arm_option_override): Use lower case for feature
15023 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
15024 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
15025 * config/arm/parsecpu.awk (END): Add new command 'isa'.
15027 (print_isa_bits_for): New function.
15028 (gen_isa): New function.
15029 (gen_comm_data): Use print_isa_bits_for.
15030 (define feature): New keyword.
15031 (define fgroup): New keyword.
15032 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
15033 (arm-isa.h): Add rule to generate file.
15034 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
15035 case for feature bit names.
15037 2017-09-06 Richard Biener <rguenther@suse.de>
15039 * tree-ssa-pre.c (NECESSARY): Remove.
15040 (create_expression_by_pieces): Do not touch pass-local flags.
15041 (insert_into_preds_of_block): Likewise.
15042 (do_pre_regular_insertion): Likewise.
15043 (eliminate_insert): Likewise.
15044 (eliminate_dom_walker::before_dom_children): Likewise.
15045 (fini_eliminate): Do not look at inserted_exprs.
15046 (mark_operand_necessary): Remove.
15047 (remove_dead_inserted_code): Replace with simple work-list
15048 algorithm based on inserted_exprs and SSA uses.
15049 (pass_pre::execute): Re-order fini_eliminate and
15050 remove_dead_inserted_code.
15052 2017-09-06 Olivier Hainque <hainque@adacore.com>
15054 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
15055 for VxWorks 7. Adjust surrounding comments.
15057 2017-09-06 Richard Biener <rguenther@suse.de>
15059 * gimple-ssa-strength-reduction.c
15060 (find_candidates_dom_walker::before_dom_children): Also allow
15063 2017-09-06 Richard Biener <rguenther@suse.de>
15065 PR tree-optimization/82108
15066 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
15067 for gap in the non-permutation SLP case.
15069 2017-09-06 Martin Jambor <mjambor@suse.cz>
15071 PR tree-optimization/82078
15072 * tree-sra.c (sort_and_splice_var_accesses): Move call to
15073 add_access_to_work_queue...
15074 (build_accesses_from_assign): ...here.
15075 (propagate_all_subaccesses): Make sure racc is the group
15076 representative, if there is one.
15078 2017-09-06 Jakub Jelinek <jakub@redhat.com>
15080 PR middle-end/82095
15081 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
15084 2017-09-06 Richard Biener <rguenther@suse.de>
15086 * gimple-ssa-strength-reduction.c
15087 (find_candidates_dom_walker::before_doom_children): Use a
15088 type and not a mode check.
15090 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
15093 * config/arm/predicates.md (arm_general_adddi_operand): Create new
15095 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
15097 2017-09-05 Jeff Law <law@redhat.com>
15099 PR tree-optimization/64910
15100 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
15101 cases where we have 3 or more operands.
15103 2017-09-05 Jakub Jelinek <jakub@redhat.com>
15105 PR middle-end/81768
15106 * omp-low.c (lower_omp_for): Recompute tree invariant if
15107 gimple_omp_for_initial/final is ADDR_EXPR.
15109 PR middle-end/81768
15110 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
15111 into gimple val before gimplification fo the COND_EXPR.
15113 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
15115 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
15116 REGION_COPY argument.
15117 (thread_through_all_blocks): Remove unused argument to
15118 duplicate_thread_path.
15120 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15121 Alan Hayward <alan.hayward@arm.com>
15122 David Sherwood <david.sherwood@arm.com>
15124 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
15125 Take a scalar_mode rather than a machine_mode.
15126 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
15127 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
15128 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
15129 (aarch64_gen_adjusted_ldpstp): Likewise.
15130 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
15132 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15133 Alan Hayward <alan.hayward@arm.com>
15134 David Sherwood <david.sherwood@arm.com>
15136 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
15137 Take a scalar_int_mode instead of a machine_mode.
15138 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
15139 (aarch64_output_scalar_simd_mov_immediate): Likewise.
15140 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
15141 (aarch64_simd_attr_length_rglist): Delete.
15142 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
15143 a scalar_int_mode instead of a machine_mode.
15144 (aarch64_add_offset): Likewise.
15145 (aarch64_internal_mov_immediate): Likewise
15146 (aarch64_add_constant_internal): Likewise.
15147 (aarch64_add_constant): Likewise.
15148 (aarch64_movw_imm): Likewise.
15149 (aarch64_rtx_arith_op_extract_p): Likewise.
15150 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
15151 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
15152 Remove assert that the mode isn't a vector.
15153 (aarch64_output_scalar_simd_mov_immediate): Likewise.
15154 (aarch64_expand_mov_immediate): Update calls after above changes.
15155 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
15156 (aarch64_and_bitmask_imm): Check for scalar integer modes.
15157 (aarch64_move_imm): Likewise.
15158 (aarch64_can_const_movi_rtx_p): Likewise.
15159 (aarch64_strip_extend): Likewise.
15160 (aarch64_extr_rtx_p): Likewise.
15161 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
15162 a CONST_INT when the mode parameter is VOIDmode.
15163 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
15165 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15167 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
15168 * stor-layout.c (bitwise_mode_for_mode): Likewise.
15169 (bitwise_type_for_mode): Update accordingly.
15171 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15173 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
15174 * stor-layout.c (mode_for_size_tree): Likewise.
15175 (mode_for_array): Update accordingly.
15176 (layout_decl): Likewise.
15177 (compute_record_mode): Likewise. Only set the mode once.
15179 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15181 * target.def (get_mask_mode): Change return type to opt_mode.
15183 * doc/tm.texi: Regenerate.
15184 * targhooks.h (default_get_mask_mode): Return an opt_mode.
15185 * targhooks.c (default_get_mask_mode): Likewise.
15186 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
15187 * optabs-query.c (can_vec_mask_load_store_p): Update use of
15188 targetm.get_mask_mode.
15189 * tree.c (build_truth_vector_type): Likewise.
15191 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15193 * machmode.h (mode_for_vector): Return an opt_mode.
15194 * stor-layout.c (mode_for_vector): Likewise.
15195 (mode_for_int_vector): Update accordingly.
15196 (layout_type): Likewise.
15197 * config/i386/i386.c (emit_memmov): Likewise.
15198 (ix86_expand_set_or_movmem): Likewise.
15199 (ix86_expand_vector_init): Likewise.
15200 (ix86_get_mask_mode): Likewise.
15201 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
15203 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
15204 * expmed.c (extract_bit_field_1): Likewise.
15205 * expr.c (expand_expr_real_2): Likewise.
15206 * optabs-query.c (can_vec_perm_p): Likewise.
15207 (can_vec_mask_load_store_p): Likewise.
15208 * optabs.c (expand_vec_perm): Likewise.
15209 * targhooks.c (default_get_mask_mode): Likewise.
15210 * tree-vect-stmts.c (vectorizable_store): Likewise.
15211 (vectorizable_load): Likewise.
15212 (get_vectype_for_scalar_type_and_size): Likewise.
15214 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15216 * machmode.h (mode_for_int_vector): New function.
15217 * stor-layout.c (mode_for_int_vector): Likewise.
15218 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
15219 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
15220 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
15221 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
15222 (s390_expand_vcond): Likewise.
15224 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15226 * machmode.h (opt_machine_mode): New type.
15227 (opt_mode<T>): Allow construction from anything that can be
15229 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
15230 (mode_for_size): Return an opt_machine_mode.
15231 * stor-layout.c (mode_for_size): Likewise.
15232 (mode_for_size_tree): Update call accordingly.
15233 (bitwise_mode_for_mode): Likewise.
15234 (make_fract_type): Likewise.
15235 (make_accum_type): Likewise.
15236 * caller-save.c (replace_reg_with_saved_mem): Update call
15238 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15239 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15240 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15241 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15242 * expmed.c (extract_bit_field_1): Likewise.
15243 * reload.c (get_secondary_mem): Likewise.
15244 * varasm.c (assemble_integer): Likewise.
15245 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
15248 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15250 * machmode.h (decimal_float_mode_for_size): New function.
15251 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
15252 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
15253 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
15254 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
15255 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
15256 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
15258 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15260 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
15261 (get_builtin_sync_mode): Likewise.
15262 (expand_ifn_atomic_compare_exchange): Likewise.
15263 (expand_builtin_atomic_clear): Likewise.
15264 (expand_builtin_atomic_test_and_set): Likewise.
15265 (fold_builtin_atomic_always_lock_free): Likewise.
15266 * calls.c (compute_argument_addresses): Likewise.
15267 (emit_library_call_value_1): Likewise.
15268 (store_one_arg): Likewise.
15269 * combine.c (combine_instructions): Likewise.
15270 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
15271 * config/arm/arm.c (arm_function_value): Likewise.
15272 (aapcs_allocate_return_reg): Likewise.
15273 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
15274 * config/i386/i386.c (construct_container): Likewise.
15275 (ix86_gimplify_va_arg): Likewise.
15276 (ix86_expand_sse_cmp): Likewise.
15277 (emit_memmov): Likewise.
15278 (emit_memset): Likewise.
15279 (expand_small_movmem_or_setmem): Likewise.
15280 (ix86_expand_pextr): Likewise.
15281 (ix86_expand_pinsr): Likewise.
15282 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
15283 * config/microblaze/microblaze.c (microblaze_block_move_straight):
15285 * config/mips/mips.c (mips_function_value_1) Likewise.
15286 (mips_block_move_straight): Likewise.
15287 (mips_expand_ins_as_unaligned_store): Likewise.
15288 * config/powerpcspe/powerpcspe.c
15289 (rs6000_darwin64_record_arg_advance_flush): Likewise.
15290 (rs6000_darwin64_record_arg_flush): Likewise.
15291 * config/rs6000/rs6000.c
15292 (rs6000_darwin64_record_arg_advance_flush): Likewise.
15293 (rs6000_darwin64_record_arg_flush): Likewise.
15294 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
15295 (sparc_function_value_1): Likewise.
15296 * config/spu/spu.c (adjust_operand): Likewise.
15297 (spu_emit_branch_or_set): Likewise.
15298 (arith_immediate_p): Likewise.
15299 * emit-rtl.c (gen_lowpart_common): Likewise.
15300 * expr.c (expand_expr_real_1): Likewise.
15301 * function.c (assign_parm_setup_block): Likewise.
15302 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
15303 * reload1.c (alter_reg): Likewise.
15304 * stor-layout.c (mode_for_vector): Likewise.
15305 (layout_type): Likewise.
15307 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15309 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
15310 (spu_convert_move): Likewise.
15311 * lower-subreg.c (resolve_simple_move): Likewise.
15313 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15316 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
15317 define_insn to a define_expand.
15318 (altivec_vsum2sws_direct): New define_insn.
15319 (altivec_vsumsws): Convert from a define_insn to a define_expand.
15321 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
15323 * config/arm/arm.c (arm_option_params_internal): Improve setting of
15326 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
15331 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
15332 realign stack if stack alignment needed is less than incoming
15335 2017-09-05 Marek Polacek <polacek@redhat.com>
15338 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
15341 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
15343 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
15345 2017-09-05 Richard Biener <rguenther@suse.de>
15347 PR tree-optimization/82084
15348 * fold-const.c (can_native_encode_string_p): Handle wide characters.
15350 2017-09-05 Richard Biener <rguenther@suse.de>
15352 PR tree-optimization/82102
15353 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
15355 2017-09-05 Martin Liska <mliska@suse.cz>
15357 PR tree-optimization/82032
15358 * tree-cfg.c (generate_range_test): New function.
15359 * tree-cfg.h (generate_range_test): Declared here.
15360 * tree-cfgcleanup.c (convert_single_case_switch): New function.
15361 (cleanup_control_expr_graph): Use it.
15362 * tree-switch-conversion.c (try_switch_expansion): Remove
15364 (emit_case_nodes): Use generate_range_test.
15366 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
15369 * config/i386/i386.md (*<btsc><mode>_mask): Add
15370 TARGET_USE_BT to insn constraint.
15371 (*btr<mode>_mask): Ditto.
15373 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
15375 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
15376 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
15378 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
15381 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
15382 TARGET_NEON and TARGET_IWMMXT.
15383 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
15384 TARGET_NEON and TARGET_IWMMXT.
15385 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
15387 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
15389 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
15390 (ix86_rewrite_tls_address): Ditto.
15391 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
15392 (ix86_rewrite_tls_address_1): Ditto.
15393 (ix86_rewrite_tls_address): Ditto.
15394 * config/i386/predicates.md (tls_address_pattern): New predicate.
15395 * config/i386/i386.md (TLS address splitter): New splitter.
15397 2017-09-04 Richard Biener <rguenther@suse.de>
15399 PR tree-optimization/82084
15400 * fold-const.h (can_native_encode_string_p): Declare.
15401 * fold-const.c (can_native_encode_string_p): Factor out from ...
15402 (native_encode_string): ... here.
15403 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
15404 vectorizing stores from constants we later cannot handle.
15406 2017-09-04 Marek Polacek <polacek@redhat.com>
15409 * doc/invoke.texi: Update -Wtautological-compare documentation.
15411 2017-09-04 Jeff Law <law@redhat.com>
15413 PR tree-optimization/64910
15414 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
15415 swap the first and last operand if the last is a constant.
15417 2017-09-04 Marek Polacek <polacek@redhat.com>
15420 * convert.c (do_narrow): When sanitizing signed integer overflows,
15421 bail out for signed types.
15422 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
15424 2017-09-04 Richard Biener <rguenther@suse.de>
15426 PR tree-optimization/82060
15427 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
15428 Move devirtualization after stmt folding and before EH/AB/noreturn
15429 cleanup to get the stmt refs canonicalized. Use a bool instead
15430 of gimple_modified_p since that doesn't work for NOPs. Schedule
15431 NOPs generated by folding for removal.
15433 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
15434 Alan Hayward <alan.hayward@arm.com>
15435 David Sherwood <david.sherwood@arm.com>
15437 * coretypes.h (pad_direction): New enum.
15438 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
15439 (FUNCTION_ARG_PADDING): Likewise.
15440 * target.def (function_arg_padding): New hook.
15441 * targhooks.h (default_function_arg_padding): Declare.
15442 * targhooks.c (default_function_arg_padding): New function.
15443 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
15444 (TARGET_FUNCTION_ARG_PADDING): ...this.
15445 * doc/tm.texi: Regenerate.
15446 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
15447 instead of direction.
15448 (compute_argument_addresses): Likewise.
15449 (load_register_parameters): Likewise.
15450 (emit_library_call_value_1): Likewise.
15451 (store_one_arg): Use targetm.calls.function_arg_padding instead
15452 of FUNCTION_ARG_PADDING.
15453 (must_pass_in_stack_var_size_or_pad): Likewise.
15454 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
15455 (emit_group_store): Likewise.
15456 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
15457 instead of FUNCTION_ARG_PADDING.
15458 (emit_push_insn): Likewise, and propagate enum change throughout
15460 * function.h (direction): Delete.
15461 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
15463 * function.c (assign_parm_find_stack_rtl): Likewise.
15464 (assign_parm_setup_block_p): Likewise.
15465 (assign_parm_setup_block): Likewise.
15466 (gimplify_parameters): Likewise.
15467 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
15468 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
15470 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
15471 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
15472 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
15473 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
15474 (aarch64_function_arg_padding): ...this new function.
15475 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
15476 (TARGET_FUNCTION_ARG_PADDING): Redefine.
15477 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
15478 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
15479 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
15480 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
15481 (arm_pad_arg_upward): Replace with...
15482 (arm_function_arg_padding): ...this new function.
15483 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
15485 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
15486 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
15487 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
15488 (ia64_hpux_function_arg_padding): Replace with...
15489 (ia64_function_arg_padding): ...this new function. Use pad_direction
15490 instead of direction. Check for TARGET_HPUX.
15491 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
15492 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
15493 (iq2000_function_arg_padding): New function.
15494 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
15495 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
15496 (mips_function_arg_padding): ...this new function.
15497 (mips_pad_reg_upward): Update accordingly.
15498 (TARGET_FUNCTION_ARG_PADDING): Redefine.
15499 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
15500 targetm.calls.function_arg_padding.
15501 (FUNCTION_ARG_PADDING): Delete.
15502 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
15503 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
15504 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
15505 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
15506 (nios2_block_reg_padding): Return pad_direction instead of direction.
15507 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
15508 instead of direction.
15509 (nios2_function_arg_padding): Likewise. Make static.
15510 (TARGET_FUNCTION_ARG_PADDING): Redefine.
15511 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
15512 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
15513 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
15514 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
15515 (pa_function_arg_padding): Make static. Return pad_direction instead
15517 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
15518 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
15519 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
15520 instead of direction. Use targetm.calls.function_arg_padding.
15521 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
15522 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
15523 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
15524 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
15525 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
15527 (function_arg_padding): Rename to...
15528 (rs6000_function_arg_padding): ...this. Make static. Return
15529 pad_direction instead of direction.
15530 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
15531 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
15532 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
15533 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
15534 instead of direction. Use targetm.calls.function_arg_padding.
15535 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
15536 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
15537 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
15538 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
15539 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
15540 (function_arg_padding): Rename to...
15541 (rs6000_function_arg_padding): ...this. Make static. Return
15542 pad_direction instead of direction.
15543 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
15544 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
15545 * config/s390/s390.c (s390_function_arg_padding): New function.
15546 (TARGET_FUNCTION_ARG_PADDING): Redefine.
15547 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
15548 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
15549 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
15550 (function_arg_padding): Rename to...
15551 (sparc_function_arg_padding): ...this. Make static. Return
15552 pad_direction instead of direction.
15553 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
15554 * config/spu/spu.c (spu_function_arg_padding): New function.
15555 (TARGET_FUNCTION_ARG_PADDING): Redefine.
15556 * system.h (FUNCTION_ARG_PADDING): Poison.
15558 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
15559 Alan Hayward <alan.hayward@arm.com>
15560 David Sherwood <david.sherwood@arm.com>
15562 * target.def (modes_tieable_p): New hook.
15563 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
15564 (TARGET_MODES_TIEABLE_P): ...this.
15565 * doc/tm.texi.in: Regenerate.
15566 * hooks.h (hook_bool_mode_mode_true): Declare.
15567 * hooks.c (hook_bool_mode_mode_true): New function.
15568 * combine.c (subst): Use targetm.modes_tieable_p instead of
15570 * dse.c (find_shift_sequence): Likewise.
15571 * expmed.c (extract_low_bits): Likewise.
15572 * lower-subreg.c: Include target.h.
15573 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
15575 * rtlanal.c (rtx_cost): Likewise.
15576 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
15577 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
15578 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
15579 (TARGET_MODES_TIEABLE_P): Redefine.
15580 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
15581 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
15582 (TARGET_MODES_TIEABLE_P): Redefine.
15583 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
15584 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
15585 (arc_modes_tieable_p): New function.
15586 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
15587 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
15588 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
15589 (arm_modes_tieable_p): Make static.
15590 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
15591 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
15592 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
15593 (TARGET_MODES_TIEABLE_P): Redefine.
15594 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
15595 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
15596 (TARGET_MODES_TIEABLE_P): Redefine.
15597 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
15598 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
15599 (cr16_modes_tieable_p): New function.
15600 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
15601 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
15602 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
15603 (TRULY_NOOP_TRUNCATION): Update comment.
15604 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
15605 (TRULY_NOOP_TRUNCATION): Update comment.
15606 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
15607 (frv_modes_tieable_p): New function.
15608 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
15609 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
15610 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
15611 (TARGET_MODES_TIEABLE_P): Redefine.
15612 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
15613 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
15614 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
15615 (TARGET_MODES_TIEABLE_P): Redefine.
15616 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
15617 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
15618 (ia64_modes_tieable_p): New function.
15619 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
15620 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
15621 (iq2000_modes_tieable_p): New function.
15622 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
15623 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
15624 (lm32_modes_tieable_p): New function.
15625 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
15626 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
15627 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
15628 (TARGET_MODES_TIEABLE_P): Redefine.
15629 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
15630 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
15631 (m32r_modes_tieable_p): New function.
15632 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
15633 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
15634 (m68k_modes_tieable_p): New function.
15635 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
15636 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
15637 (mcore_modes_tieable_p): New function.
15638 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
15639 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
15641 (TARGET_MODES_TIEABLE_P): Redefine.
15642 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
15643 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
15644 * config/mips/mips.c (mips_modes_tieable_p): Make static.
15645 (TARGET_MODES_TIEABLE_P): Redefine.
15646 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
15647 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
15648 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
15649 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
15650 (mn10300_modes_tieable_p): ...this and make static.
15651 (TARGET_MODES_TIEABLE_P): Redefine.
15652 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
15653 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
15654 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
15655 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
15656 (msp430_modes_tieable_p): Make static.
15657 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
15658 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
15659 (TARGET_MODES_TIEABLE_P): Redefine.
15660 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
15661 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
15662 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
15663 (TARGET_MODES_TIEABLE_P): Redefine.
15664 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
15665 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
15666 * config/pa/pa.c (pa_modes_tieable_p): Make static.
15667 (TARGET_MODES_TIEABLE_P): Redefine.
15668 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
15669 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
15670 (pdp11_modes_tieable_p): New function.
15671 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
15672 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
15673 (rs6000_modes_tieable_p): New function.
15674 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
15675 * config/powerpcspe/powerpcspe.md: Update comment.
15676 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
15677 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
15678 (TARGET_MODES_TIEABLE_P): Redefine.
15679 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
15680 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
15681 (rl78_modes_tieable_p): New function.
15682 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
15683 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
15684 (rs6000_modes_tieable_p): New function.
15685 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
15686 * config/rs6000/rs6000.md: Update comment.
15687 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
15688 * config/rx/rx.c (rx_modes_tieable_p): New function.
15689 (TARGET_MODES_TIEABLE_P): Redefine.
15690 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
15691 * config/s390/s390.c (s390_modes_tieable_p): New function.
15692 (TARGET_MODES_TIEABLE_P): Redefine.
15693 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
15694 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
15695 (sh_modes_tieable_p): New function.
15696 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
15697 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
15698 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
15699 (sparc_modes_tieable_p): Make static.
15700 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
15701 * config/spu/spu.c (spu_modes_tieable_p): New function.
15702 (TARGET_MODES_TIEABLE_P): Redefine.
15703 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
15704 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
15705 (TARGET_MODES_TIEABLE_P): Redefine.
15706 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
15707 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
15708 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
15709 * config/v850/v850.c (v850_modes_tieable_p): New function.
15710 (TARGET_MODES_TIEABLE_P): Redefine.
15711 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
15712 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
15713 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
15714 (visium_modes_tieable_p): New function.
15715 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
15716 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
15717 (xtensa_modes_tieable_p): New function.
15718 * system.h (MODES_TIEABLE_P): Poison.
15720 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
15721 Alan Hayward <alan.hayward@arm.com>
15722 David Sherwood <david.sherwood@arm.com>
15724 * target.def (hard_regno_mode_ok): New hook.
15725 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
15726 (TARGET_HARD_REGNO_MODE_OK): ...this.
15727 * doc/tm.texi.in: Regenerate.
15728 * hooks.h (hook_bool_uint_mode_true): Declare.
15729 * hooks.c (hook_bool_uint_mode_true): New function.
15730 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
15731 HARD_REGNO_MODE_OK.
15732 * genpreds.c (write_insn_preds_c): Add an include of target.h.
15733 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
15734 instead of HARD_REGNO_MODE_OK.
15735 * caller-save.c: Include target.h.
15736 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
15737 HARD_REGNO_MODE_OK.
15738 * combine.c (can_combine_p): Likewise.
15739 (combinable_i3pat): Likewise.
15740 (can_change_dest_mode): Likewise.
15741 * expr.c (init_expr_target): Likewise.
15742 (convert_move): Likewise.
15743 (convert_modes): Likewise.
15744 * ira.c (setup_prohibited_class_mode_regs): Likewise.
15745 (setup_prohibited_mode_move_regs): Likewise.
15746 * ira.h (target_ira): Likewise.
15747 * lra-assigns.c (find_hard_regno_for_1): Likewise.
15748 * lra-constraints.c (process_alt_operands): Likewise.
15749 (split_reg): Likewise.
15750 * recog.c (peep2_find_free_register): Likewise.
15751 * ree.c (combine_reaching_defs): Likewise.
15752 * regcprop.c (maybe_mode_change): Likewise.
15753 * reginfo.c (init_reg_sets_1): Likewise.
15754 (choose_hard_reg_mode): Likewise.
15755 (simplifiable_subregs): Likewise.
15756 * regrename.c (check_new_reg_p): Likewise.
15757 * reload.c (find_valid_class): Likewise.
15758 (find_valid_class_1): Likewise.
15759 (reload_inner_reg_of_subreg): Likewise.
15760 (push_reload): Likewise.
15761 (combine_reloads): Likewise.
15762 (find_dummy_reload): Likewise.
15763 (find_reloads): Likewise.
15764 * reload1.c (find_reg): Likewise.
15765 (set_reload_reg): Likewise.
15766 (allocate_reload_reg): Likewise.
15767 (choose_reload_regs): Likewise.
15768 (reload_adjust_reg_for_temp): Likewise.
15769 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
15770 (simplify_subreg_regno): Likewise.
15771 * sel-sched.c (init_regs_for_mode): Likewise.
15772 * varasm.c (make_decl_rtl): Likewise.
15773 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
15774 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
15775 HARD_REGNO_MODE_OK.
15776 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
15777 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
15778 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15779 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
15780 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
15781 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15782 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
15783 (arc_mode_class): Delete.
15784 (HARD_REGNO_MODE_OK): Delete.
15785 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15786 (arc_hard_regno_mode_ok): Rename old array to...
15787 (arc_hard_regno_mode_ok_modes): ...this.
15788 (arc_conditional_register_usage): Update accordingly.
15789 (arc_mode_class): Make static.
15790 (arc_hard_regno_mode_ok): New function.
15791 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
15792 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
15793 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15794 (arm_hard_regno_mode_ok): Make static.
15795 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
15796 HARD_REGNO_MODE_OK.
15797 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
15798 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
15799 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
15801 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15802 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
15803 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
15804 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
15805 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
15806 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15807 * config/bfin/predicates.md (valid_reg_operand): Use
15808 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
15809 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
15810 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
15811 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15812 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
15813 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
15814 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15815 (cr16_hard_regno_mode_ok): Make static and return a bool.
15816 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
15817 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15818 (cris_hard_regno_mode_ok): New function.
15819 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
15820 (epiphany_mode_class): Delete.
15821 (HARD_REGNO_MODE_OK): Delete.
15822 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
15823 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15824 (hard_regno_mode_ok): Rename to...
15825 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
15826 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
15827 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
15828 HARD_REGNO_MODE_OK.
15829 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
15830 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
15831 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15832 (frv_hard_regno_mode_ok): Make static and return a bool.
15833 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
15834 HARD_REGNO_MODE_OK.
15835 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
15836 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
15837 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
15838 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
15840 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15841 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
15842 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
15843 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
15845 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15846 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
15847 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15848 (ia64_hard_regno_mode_ok): New function.
15849 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
15850 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15851 (iq2000_hard_regno_mode_ok): New function.
15852 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
15853 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15854 (lm32_hard_regno_mode_ok): New function.
15855 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
15856 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
15857 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
15858 instead of HARD_REGNO_MODE_OK.
15859 (m32c_hard_regno_ok): Rename to...
15860 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
15861 (m32c_cannot_change_mode_class): Update accordingly.
15862 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15863 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
15864 (m32r_mode_class): Delete.
15865 (HARD_REGNO_MODE_OK): Delete.
15866 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15867 (m32r_hard_regno_mode_ok): Rename to...
15868 (m32r_hard_regno_modes): ...this.
15869 (m32r_mode_class): Make static.
15870 (m32r_hard_regno_mode_ok): New function.
15871 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
15872 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
15873 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15874 (m68k_hard_regno_mode_ok): Make static.
15875 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
15876 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15877 (mcore_hard_regno_mode_ok): New function.
15878 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
15879 (HARD_REGNO_MODE_OK): Delete.
15880 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
15882 (microblaze_hard_regno_mode_ok_p): ...this and make static.
15883 (microblaze_hard_regno_mode_ok): New function.
15884 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15885 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
15886 (mips_hard_regno_mode_ok): Delete.
15887 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
15888 (mips_hard_regno_mode_ok_p): ...this and make static.
15889 (mips_hard_regno_mode_ok_p): Rename to...
15890 (mips_hard_regno_mode_ok_uncached): ...this.
15891 (mips_hard_regno_mode_ok): New function.
15892 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
15893 of HARD_REGNO_MODE_OK.
15894 (mips_option_override): Update after above name changes.
15895 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15896 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
15897 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
15898 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
15899 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
15900 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15901 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
15902 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
15903 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
15904 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15905 (msp430_hard_regno_mode_ok): Make static and return a bool.
15906 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
15907 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
15908 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
15910 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15911 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
15912 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
15913 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
15914 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
15915 (PA_HARD_REGNO_MODE_OK): ...this
15916 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
15917 (PA_HARD_REGNO_MODE_OK): ...this.
15918 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15919 (pa_hard_regno_mode_ok): New function.
15920 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
15921 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15922 (pdp11_hard_regno_mode_ok): New function.
15923 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
15924 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
15926 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
15928 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15929 (rs6000_hard_regno_mode_ok): Rename to...
15930 (rs6000_hard_regno_mode_ok_uncached): ...this.
15931 (rs6000_init_hard_regno_mode_ok): Update accordingly.
15932 (rs6000_hard_regno_mode_ok): New function.
15933 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
15934 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
15935 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
15936 (riscv_hard_regno_mode_ok): ...this and make static.
15937 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15938 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
15939 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
15940 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15941 (rl78_hard_regno_mode_ok): Make static and return bool.
15942 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
15943 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
15945 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
15946 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15947 (rs6000_hard_regno_mode_ok): Rename to...
15948 (rs6000_hard_regno_mode_ok_uncached): ...this.
15949 (rs6000_init_hard_regno_mode_ok): Update accordingly.
15950 (rs6000_hard_regno_mode_ok): New function.
15951 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
15952 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
15953 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15954 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
15955 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
15956 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
15957 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15958 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
15959 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
15960 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15961 (sh_hard_regno_mode_ok): Make static.
15962 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
15963 instead of HARD_REGNO_MODE_OK.
15964 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
15965 (sparc_mode_class): Delete.
15966 (HARD_REGNO_MODE_OK): Delete.
15967 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15968 (hard_regno_mode_classes): Make static.
15969 (sparc_mode_class): Likewise.
15970 (sparc_hard_regno_mode_ok): New function.
15971 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
15972 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
15973 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
15975 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15976 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
15977 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
15978 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
15979 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
15980 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15981 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
15982 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
15983 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15984 (visium_hard_regno_mode_ok): New function.
15985 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
15986 instead of HARD_REGNO_MODE_OK.
15987 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
15988 (HARD_REGNO_MODE_OK): Delete.
15989 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
15990 (xtensa_hard_regno_mode_ok_p): ...this and make static.
15991 (xtensa_option_override): Update accordingly.
15992 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15993 (xtensa_hard_regno_mode_ok): New function.
15994 * system.h (HARD_REGNO_MODE_OK): Poison.
15996 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
15997 Alan Hayward <alan.hayward@arm.com>
15998 David Sherwood <david.sherwood@arm.com>
16000 * target.def (hard_regno_call_part_clobbered): New hook.
16001 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
16002 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
16003 * doc/tm.texi: Regenerate.
16004 * hooks.h (hook_bool_uint_mode_false): Declare.
16005 * hooks.c (hook_bool_uint_mode_false): New function.
16006 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16007 * cselib.c (cselib_process_insn): Use
16008 targetm.hard_regno_call_part_clobbered instead of
16009 HARD_REGNO_CALL_PART_CLOBBERED.
16010 * ira-conflicts.c (ira_build_conflicts): Likewise.
16011 * ira-costs.c (ira_tune_allocno_costs): Likewise.
16012 * lra-constraints.c (need_for_call_save_p): Likewise.
16013 * lra-lives.c: Include target.h.
16014 (check_pseudos_live_through_calls): Use
16015 targetm.hard_regno_call_part_clobbered instead of
16016 HARD_REGNO_CALL_PART_CLOBBERED.
16017 * regcprop.c: Include target.h.
16018 (copyprop_hardreg_forward_1): Use
16019 targetm.hard_regno_call_part_clobbered instead of
16020 HARD_REGNO_CALL_PART_CLOBBERED.
16021 * reginfo.c (choose_hard_reg_mode): Likewise.
16022 * regrename.c (check_new_reg_p): Likewise.
16023 * reload.c (find_equiv_reg): Likewise.
16024 * reload1.c (emit_reload_insns): Likewise.
16025 * sched-deps.c (deps_analyze_insn): Likewise.
16026 * sel-sched.c (init_regs_for_mode): Likewise.
16027 (mark_unavailable_hard_regs): Likewise.
16028 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
16029 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16030 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
16032 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16033 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16034 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
16036 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
16038 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16039 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16040 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
16042 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16043 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16044 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
16046 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16047 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
16049 * config/powerpcspe/powerpcspe.c
16050 (rs6000_hard_regno_call_part_clobbered): New function.
16051 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16052 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16053 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
16055 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16056 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16057 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
16059 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16060 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16061 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
16063 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16064 Alan Hayward <alan.hayward@arm.com>
16065 David Sherwood <david.sherwood@arm.com>
16067 * rtl.h (subreg_memory_offset): Declare.
16068 * emit-rtl.c (subreg_memory_offset): New function.
16069 * expmed.c (store_bit_field_1): Use it.
16070 * expr.c (undefined_operand_subword_p): Likewise.
16071 * simplify-rtx.c (simplify_subreg): Likewise.
16073 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
16075 PR rtl-optimization/57448
16078 * optabs.c (expand_atomic_load): Place compiler memory barriers if
16079 using atomic_load pattern.
16080 (expand_atomic_store): Likewise.
16082 2017-09-04 Jakub Jelinek <jakub@redhat.com>
16085 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
16086 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
16087 handling. Use replace_call_with_value with NULL instead of
16088 gsi_replace, unlink_stmt_vdef and release_defs.
16090 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
16093 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
16095 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16098 * rtl.h (emit_library_call_value_1): Declare.
16099 (emit_library_call): Replace declaration with a series of overloads.
16100 Remove the parameter count argument.
16101 (emit_library_call_value): Likewise.
16102 * calls.c (emit_library_call_value_1): Make global. Replace varargs
16103 with an "rtx_mode_t *".
16104 (emit_library_call_value): Delete.
16105 (emit_library_call): Likewise.
16106 * asan.c (asan_emit_stack_protection): Update calls accordingly.
16107 (asan_emit_allocas_unpoison): Likewise.
16108 * builtins.c (expand_builtin_powi): Likewise.
16109 (expand_asan_emit_allocas_unpoison): Likewise.
16110 * cfgexpand.c (expand_main_function): Likewise.
16111 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
16112 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
16113 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
16114 * config/arm/arm.c (arm_trampoline_init): Likewise.
16115 (arm_call_tls_get_addr): Likewise.
16116 (arm_expand_divmod_libfunc): Likewise.
16117 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
16118 (smulsi3_highpart): Likewise.
16119 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
16120 (c6x_expand_compare): Likewise.
16121 (c6x_expand_movmem): Likewise.
16122 * config/frv/frv.c (frv_trampoline_init): Likewise.
16123 * config/i386/i386.c (ix86_trampoline_init): Likewise.
16124 (ix86_expand_divmod_libfunc): Likewise.
16125 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
16126 (ia64_expand_compare): Likewise.
16127 (ia64_profile_hook): Likewise.
16128 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
16129 (nonlocal_goto): Likewise.
16130 (restore_stack_nonlocal): Likewise.
16131 * config/m32r/m32r.c (block_move_call): Likewise.
16132 (m32r_trampoline_init): Likewise.
16133 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
16134 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
16135 (m68k_call_m68k_read_tp): Likewise.
16136 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
16137 (microblaze_expand_divide): Likewise.
16138 * config/mips/mips.h (mips_args): Likewise.
16139 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
16140 (MIPS_ICACHE_SYNC): Likewise.
16141 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
16142 (nios2_trampoline_init): Likewise.
16143 * config/pa/pa.c (hppa_tls_call): Likewise.
16144 (pa_trampoline_init): Likewise.
16145 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
16146 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
16147 (expand_strn_compare): Likewise.
16148 (rs6000_generate_compare): Likewise.
16149 (rs6000_expand_float128_convert): Likewise.
16150 (output_profile_hook): Likewise.
16151 (rs6000_trampoline_init): Likewise.
16152 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
16153 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
16154 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
16155 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
16156 (rs6000_generate_compare): Likewise.
16157 (rs6000_expand_float128_convert): Likewise.
16158 (output_profile_hook): Likewise.
16159 (rs6000_trampoline_init): Likewise.
16160 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
16161 * config/sh/sh.c (sh_trampoline_init): Likewise.
16162 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
16163 (sparc_emit_float_lib_cmp): Likewise.
16164 (sparc32_initialize_trampoline): Likewise.
16165 (sparc64_initialize_trampoline): Likewise.
16166 (sparc_profile_hook): Likewise.
16167 * config/spu/spu.c (ea_load_store): Likewise.
16168 * config/spu/spu.md (floatunssidf2): Likewise.
16169 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
16170 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
16171 * config/visium/visium.c (expand_block_move_4): Likewise.
16172 (expand_block_move_2): Likewise.
16173 (expand_block_move_1): Likewise.
16174 (expand_block_set_4): Likewise.
16175 (expand_block_set_2): Likewise.
16176 (expand_block_set_1): Likewise.
16177 (visium_trampoline_init): Likewise.
16178 (visium_profile_hook): Likewise.
16179 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
16180 (xtensa_setup_frame_addresses): Likewise.
16181 (xtensa_trampoline_init): Likewise.
16182 * except.c (sjlj_emit_function_enter): Likewise.
16183 (sjlj_emit_function_exit): Likewise.
16184 * explow.c (allocate_dynamic_stack_space): Likewise.
16185 (probe_stack_range): Likewise.
16186 * expr.c (convert_mode_scalar): Likewise.
16187 * optabs.c (expand_binop): Likewise.
16188 (expand_twoval_binop_libfunc): Likewise.
16189 (expand_unop): Likewise.
16190 (prepare_cmp_insn): Likewise.
16191 (prepare_float_lib_cmp): Likewise.
16192 (expand_float): Likewise.
16193 (expand_fix): Likewise.
16194 (expand_fixed_convert): Likewise.
16195 (maybe_emit_sync_lock_test_and_set): Likewise.
16196 (expand_atomic_compare_and_swap): Likewise.
16197 (expand_mem_thread_fence): Likewise.
16198 (expand_atomic_fetch_op): Likewise.
16200 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
16202 * doc/generic.texi (OpenACC): Adjust URL.
16203 * doc/invoke.texi (C Dialect Options): Ditto.
16205 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
16207 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
16208 predicate for operand 1. Add (m,<S>) constraint.
16209 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
16210 Prevent memory operand 1 with register operand 2.
16212 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
16214 PR rtl-optimization/82024
16215 * combine.c (try_combine): If the combination result is a PARALLEL,
16216 and we only need to retain the SET in there that would be placed
16217 at I2, check that we can place that at I3 instead, before doing so.
16219 2017-09-01 Jakub Jelinek <jakub@redhat.com>
16222 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
16224 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
16225 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
16228 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
16229 Jeff Law <law@redhat.com>
16231 * varasm.c (bss_initializer_p): Do not put constants into .bss
16232 (categorize_decl_for_section): Handle bss_initializer_p returning
16233 false when DECL_INITIAL is NULL.
16235 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16238 * config/s390/s390.c (s390_can_inline_p): New function.
16240 2017-09-01 Jeff Law <law@redhat.com>
16242 PR tree-optimization/82052
16243 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
16244 Always initialize the returned slot after a hash table miss
16245 when INSERT is true.
16247 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
16249 * config/s390/s390.md (mem_signal_fence): Remove.
16250 * doc/md.texi (mem_signal_fence): Remove.
16251 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
16253 * target-insns.def (mem_signal_fence): Remove.
16255 2017-09-01 Jakub Jelinek <jakub@redhat.com>
16258 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
16261 * asan.c (create_odr_indicator): Strip name encoding from assembler
16262 name before appending it after __odr_asan_.
16264 2017-09-01 Martin Liska <mliska@suse.cz>
16266 PR tree-optimization/82059
16267 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
16268 frequency only when an edge is redirected.
16270 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
16272 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
16273 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
16274 (arc_conditional_register_usage): Remove ARC600 lp_count
16276 (arc_file_start): Emit Tag_ARC_CPU_variation.
16277 (arc_can_use_doloop_p): New conditions to use ZOLs.
16278 (hwloop_fail): New function.
16279 (hwloop_optimize): Likewise.
16280 (hwloop_pattern_reg): Likewise.
16281 (arc_doloop_hooks): New struct, to be used with reorg_loops.
16282 (arc_reorg_loops): New function, calls reorg_loops.
16283 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
16284 (arc600_corereg_hazard): Remove ZOL checking, case handled by
16286 (arc_loop_hazard): Remove function, functionality moved into
16288 (arc_hazard): Remove arc_loop_hazard call.
16289 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
16290 into hwloop_optimize.
16291 (arc_label_align): Remove ZOL handling.
16292 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
16293 * config/arc/arc.md (doloop_begin): Remove pattern.
16294 (doloop_begin_i): Likewise.
16295 (doloop_end_i): Likewise.
16296 (doloop_fallback): Likewise.
16297 (doloop_fallback_m): Likewise.
16298 (doloop_end): Reimplement expand.
16299 (arc_lp): New pattern for LP instruction.
16300 (loop_end): New pattern.
16301 (loop_fail): Likewise.
16302 (decrement_and_branch_until_zero): Likewise.
16303 * config/arc/arc.opt (mlpc-width): New option.
16304 * doc/invoke.texi (mlpc-width): Document option.
16306 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
16308 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
16309 (arc_ccfsm_advance): Fix checking for delay slots.
16310 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
16312 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
16314 * config/arc/arc.md (movqi_insn): Add stores to save constant long
16316 (movhi_insn): Update store instruction constraint which are saving
16317 6-bit short immediates.
16318 (movsi_insn): Consider also short scaled load operations.
16319 (zero_extendhisi2_i): Use Usd constraint instead of T.
16320 (extendhisi2_i): Add q constraint.
16321 (arc_clzsi2): Add type and length attributes.
16322 (arc_ctzsi2): Likewise.
16323 * config/arc/constraints.md (Usc): Update constraint, the
16324 assembler can parse two relocations for a single instruction.
16326 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
16328 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
16329 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
16331 2017-08-31 Olivier Hainque <hainque@adacore.com>
16333 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
16334 match as powerpc-wrs-vxworks*.
16336 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
16338 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
16339 register constraint for by-element operand.
16340 (aarch64_mls_elt_merge<mode>): Likewise.
16342 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
16344 * config/arc/arc.c (arc_can_follow_jump): Check for short
16347 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
16349 * config.gcc: Use g.opt for arc.
16350 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
16351 functionality moved to ...
16352 (legitimate_scaled_address_p): New function, ...here.
16353 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
16354 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
16355 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
16357 (arc_override_options): Handle G option.
16358 (arc_output_pic_addr_const): Correct function definition.
16359 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
16360 (arc_decl_anon_ns_mem_p): Delete.
16361 (arc_in_small_data_p): Overhaul this function to take into
16362 consideration the value given via G option.
16363 (arc_rewrite_small_data_1): Renamed and corrected old
16364 arc_rewrite_small_data function.
16365 (arc_rewrite_small_data): New function.
16366 (small_data_pattern): Don't use pic_offset_table_rtx.
16367 * config/arc/arc.h (CC1_SPEC): Recognize G option.
16368 * config/arc/simdext.md (movmisalignv2hi): Use
16369 prepare_move_operands function.
16371 (movmisalign*): Likewise.
16372 * doc/invoke.texi (ARC options): Document -G option.
16374 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
16376 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
16378 * config/arc/arc.c (arc_print_operand): Output scalled address for
16379 sdata whenever is possible.
16380 (arc_in_small_data_p): Allow sdata for 64bit datum when double
16381 load/stores are available.
16382 (compact_sda_memory_operand): Check for the alignment required by
16383 code density instructions.
16384 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
16386 * config/arc/constraints.md (Usd): Update constraint.
16387 (Us0): New constraint.
16388 (Usc): Update constraint.
16390 2017-08-31 Richard Biener <rguenther@suse.de>
16392 PR middle-end/82054
16393 * dwarf2out.c (dwarf2out_early_global_decl): Process each
16394 function only once.
16396 2017-08-31 Tamar Christina <tamar.christina@arm.com>
16398 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
16399 Resize type_signature.
16401 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
16402 Alan Hayward <alan.hayward@arm.com>
16403 David Sherwood <david.sherwood@arm.com>
16405 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
16406 subregs whose inner modes can be stored in GPRs.
16407 (aarch64_classify_index): Likewise.
16409 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
16410 Alan Hayward <alan.hayward@arm.com>
16411 David Sherwood <david.sherwood@arm.com>
16413 * config/aarch64/iterators.md (V_cmp_result): Rename to...
16414 (V_INT_EQUIV): ...this.
16415 (v_cmp_result): Rename to...
16416 (v_int_equiv): ...this.
16417 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
16418 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
16419 (copysign<mode>3): Likewise.
16420 (aarch64_simd_bsl<mode>_internal): Likewise.
16421 (aarch64_simd_bsl<mode>): Likewise.
16422 (vec_cmp<mode><mode>): Likewise.
16423 (vcond<mode><mode>): Likewise.
16424 (vcond<v_cmp_mixed><mode>): Likewise.
16425 (vcondu<mode><v_cmp_mixed>): Likewise.
16426 (aarch64_cm<optab><mode>): Likewise.
16427 (aarch64_cmtst<mode>): Likewise.
16428 (aarch64_fac<optab><mode>): Likewise.
16429 (vec_perm_const<mode>): Likewise.
16430 (vcond_mask_<mode><v_cmp_result>): Rename to...
16431 (vcond_mask_<mode><v_int_equiv>): ...this.
16432 (vec_cmp<mode><v_cmp_result>): Rename to...
16433 (vec_cmp<mode><v_int_equiv>): ...this.
16435 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
16436 Alan Hayward <alan.hayward@arm.com>
16437 David Sherwood <david.sherwood@arm.com>
16439 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
16441 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
16442 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
16443 (UNSPEC_LD4_DREG): New unspecs.
16444 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
16445 (aarch64_ld2<mode>_dreg_be): Replace with...
16446 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
16448 (aarch64_ld3<mode>_dreg_le)
16449 (aarch64_ld3<mode>_dreg_be): Replace with...
16450 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
16452 (aarch64_ld4<mode>_dreg_le)
16453 (aarch64_ld4<mode>_dreg_be): Replace with...
16454 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
16457 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16459 PR tree-optimization/81987
16460 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
16461 insert an initializer in a location not dominated by the stride
16464 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
16466 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
16467 on the entire header of the finally block in the fallthru case.
16469 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
16471 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
16473 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
16475 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
16476 rs6000_emit_move_from_cr and call renamed function.
16477 (rs6000_emit_prologue): Call renamed functions.
16478 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
16479 movesi_from_cr, remove volatile CRs.
16481 2017-08-30 Jon Beniston <jon@beniston.com>
16482 Richard Biener <rguenther@suse.de>
16484 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
16485 of VECTOR_MODE_P check.
16486 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
16487 element vector types.
16489 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16491 * df.h (df_read_modify_subreg_p): Remove in favor of...
16492 * rtl.h (read_modify_subreg_p): ...this new function. Take a
16493 const_rtx instead of an rtx.
16494 * cprop.c (local_cprop_find_used_regs): Update accordingly.
16495 * df-problems.c (df_word_lr_mark_ref): Likewise.
16496 * ira-lives.c (mark_pseudo_reg_live): Likewise.
16497 (mark_pseudo_reg_dead): Likewise.
16498 (mark_ref_dead): Likewise.
16499 * reginfo.c (init_subregs_of_mode): Likewise.
16500 * sched-deps.c (sched_analyze_1): Likewise.
16501 * df-scan.c (df_def_record_1): Likewise.
16502 (df_uses_record): Likewise.
16503 (df_read_modify_subreg_p): Remove in favor of...
16504 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
16505 const_rtx instead of an rtx.
16507 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16508 Alan Hayward <alan.hayward@arm.com>
16509 David Sherwood <david.sherwood@arm.com>
16511 * rtl.h (partial_subreg_p): New function.
16512 * caller-save.c (save_call_clobbered_regs): Use it.
16513 * calls.c (expand_call): Likewise.
16514 * combine.c (combinable_i3pat): Likewise.
16515 (simplify_set): Likewise.
16516 (make_extraction): Likewise.
16517 (make_compound_operation_int): Likewise.
16518 (gen_lowpart_or_truncate): Likewise.
16519 (force_to_mode): Likewise.
16520 (make_field_assignment): Likewise.
16521 (reg_truncated_to_mode): Likewise.
16522 (record_truncated_value): Likewise.
16523 (move_deaths): Likewise.
16524 * cse.c (record_jump_cond): Likewise.
16525 (cse_insn): Likewise.
16526 * cselib.c (cselib_lookup_1): Likewise.
16527 * expmed.c (extract_bit_field_using_extv): Likewise.
16528 * function.c (assign_parm_setup_reg): Likewise.
16529 * ifcvt.c (noce_convert_multiple_sets): Likewise.
16530 * ira-build.c (create_insn_allocnos): Likewise.
16531 * lra-coalesce.c (merge_pseudos): Likewise.
16532 * lra-constraints.c (match_reload): Likewise.
16533 (simplify_operand_subreg): Likewise.
16534 (curr_insn_transform): Likewise.
16535 * lra-lives.c (process_bb_lives): Likewise.
16536 * lra.c (new_insn_reg): Likewise.
16537 (lra_substitute_pseudo): Likewise.
16538 * regcprop.c (mode_change_ok): Likewise.
16539 (maybe_mode_change): Likewise.
16540 (copyprop_hardreg_forward_1): Likewise.
16541 * reload.c (push_reload): Likewise.
16542 (find_reloads): Likewise.
16543 (find_reloads_subreg_address): Likewise.
16544 * reload1.c (alter_reg): Likewise.
16545 (eliminate_regs_1): Likewise.
16546 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
16548 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
16550 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
16551 back to if statements, including unpack.
16553 2017-08-30 Martin Liska <mliska@suse.cz>
16555 PR inline-asm/82001
16556 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
16558 (func_checker::compare_asm_inputs_outputs): ... this function.
16559 (func_checker::compare_gimple_asm): Use the function to compare
16560 also ASM constrains.
16561 * ipa-icf-gimple.h: Rename the function.
16563 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16564 Alan Hayward <alan.hayward@arm.com>
16565 David Sherwood <david.sherwood@arm.com>
16567 * coretypes.h (complex_mode): New type.
16568 * gdbhooks.py (build_pretty_printer): Handle it.
16569 * machmode.h (complex_mode): New class.
16570 (complex_mode::includes_p): New function.
16571 (is_complex_int_mode): Likewise.
16572 (is_complex_float_mode): Likewise.
16573 * genmodes.c (get_mode_class): Handle complex mode classes.
16574 * function.c (expand_function_end): Use is_complex_int_mode.
16576 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16577 Alan Hayward <alan.hayward@arm.com>
16578 David Sherwood <david.sherwood@arm.com>
16580 * coretypes.h (scalar_mode_pod): New typedef.
16581 * gdbhooks.py (build_pretty_printer): Handle it.
16582 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
16583 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
16584 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
16585 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
16586 as_a <scalar_mode>.
16588 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16589 Alan Hayward <alan.hayward@arm.com>
16590 David Sherwood <david.sherwood@arm.com>
16592 * machmode.h (mode_for_vector): Take a scalar_mode instead
16594 * stor-layout.c (mode_for_vector): Likewise.
16595 * explow.c (promote_mode): Use as_a <scalar_mode>.
16596 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
16598 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16599 Alan Hayward <alan.hayward@arm.com>
16600 David Sherwood <david.sherwood@arm.com>
16602 * target.def (preferred_simd_mode): Take a scalar_mode
16603 instead of a machine_mode.
16604 * targhooks.h (default_preferred_simd_mode): Likewise.
16605 * targhooks.c (default_preferred_simd_mode): Likewise.
16606 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
16607 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
16608 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
16609 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
16610 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
16611 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
16612 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
16613 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
16615 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
16616 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
16617 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
16618 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
16619 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
16620 * doc/tm.texi: Regenerate.
16621 * optabs-query.c (can_vec_mask_load_store_p): Return false for
16624 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16625 Alan Hayward <alan.hayward@arm.com>
16626 David Sherwood <david.sherwood@arm.com>
16628 * target.def (scalar_mode_supported_p): Take a scalar_mode
16629 instead of a machine_mode.
16630 * targhooks.h (default_scalar_mode_supported_p): Likewise.
16631 * targhooks.c (default_scalar_mode_supported_p): Likewise.
16632 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
16633 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
16634 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
16635 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
16636 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
16637 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
16638 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
16639 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
16640 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
16641 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
16642 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
16643 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
16645 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
16646 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
16647 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
16648 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
16649 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
16651 * doc/tm.texi: Regenerate.
16653 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16654 Alan Hayward <alan.hayward@arm.com>
16655 David Sherwood <david.sherwood@arm.com>
16657 * coretypes.h (opt_scalar_mode): New typedef.
16658 * gdbhooks.py (build_pretty_printers): Handle it.
16659 * machmode.h (mode_iterator::get_2xwider): Add overload for
16661 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
16663 * expr.c (convert_mode_scalar): Likewise.
16664 * omp-low.c (omp_clause_aligned_alignment): Likewise.
16665 * optabs.c (expand_float): Likewise.
16666 (expand_fix): Likewise.
16667 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
16669 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16670 Alan Hayward <alan.hayward@arm.com>
16671 David Sherwood <david.sherwood@arm.com>
16673 * optabs.c (expand_float): Explicitly check for scalars before
16674 using a branching expansion.
16675 (expand_fix): Likewise.
16677 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16678 Alan Hayward <alan.hayward@arm.com>
16679 David Sherwood <david.sherwood@arm.com>
16681 * expr.c (convert_mode): Split scalar handling out into...
16682 (convert_mode_scalar): ...this new function. Treat the modes
16685 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16686 Alan Hayward <alan.hayward@arm.com>
16687 David Sherwood <david.sherwood@arm.com>
16689 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
16691 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
16693 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16694 Alan Hayward <alan.hayward@arm.com>
16695 David Sherwood <david.sherwood@arm.com>
16697 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
16698 rather than a machine_mode.
16699 (fixed_from_string): Likewise.
16700 (fixed_convert): Likewise.
16701 (fixed_convert_from_int): Likewise.
16702 (fixed_convert_from_real): Likewise.
16703 (real_convert_from_fixed): Likewise.
16704 * fixed-value.c (fixed_from_double_int): Likewise.
16705 (fixed_from_string): Likewise.
16706 (fixed_convert): Likewise.
16707 (fixed_convert_from_int): Likewise.
16708 (fixed_convert_from_real): Likewise.
16709 (real_convert_from_fixed): Likewise.
16710 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
16712 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16713 Alan Hayward <alan.hayward@arm.com>
16714 David Sherwood <david.sherwood@arm.com>
16716 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
16717 of separate mode class checks. Do not allow vector modes here.
16718 (immed_wide_int_const): Use as_a <scalar_mode>.
16719 * explow.c (trunc_int_for_mode): Likewise.
16720 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
16721 (wi::shwi): Likewise.
16722 (wi::min_value): Likewise.
16723 (wi::max_value): Likewise.
16724 * dwarf2out.c (loc_descriptor): Likewise.
16725 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
16726 for CONST_WIDE_INT.
16728 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16729 Alan Hayward <alan.hayward@arm.com>
16730 David Sherwood <david.sherwood@arm.com>
16732 * tree.h (SCALAR_TYPE_MODE): New macro.
16733 * expr.c (expand_expr_addr_expr_1): Use it.
16734 (expand_expr_real_2): Likewise.
16735 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
16736 (fold_convert_const_fixed_from_int): Likewise.
16737 (fold_convert_const_fixed_from_real): Likewise.
16738 (native_encode_fixed): Likewise
16739 (native_encode_complex): Likewise
16740 (native_encode_vector): Likewise.
16741 (native_interpret_fixed): Likewise.
16742 (native_interpret_real): Likewise.
16743 (native_interpret_complex): Likewise.
16744 (native_interpret_vector): Likewise.
16745 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
16746 (simd_clone_adjust_argument_types): Likewise.
16747 (simd_clone_init_simd_arrays): Likewise.
16748 (simd_clone_adjust): Likewise.
16749 * stor-layout.c (layout_type): Likewise.
16750 * tree.c (build_minus_one_cst): Likewise.
16751 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
16752 * tree-inline.c (estimate_move_cost): Likewise.
16753 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
16754 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
16755 (vectorizable_reduction): Likewise.
16756 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
16757 (vect_recog_mixed_size_cond_pattern): Likewise.
16758 (check_bool_pattern): Likewise.
16759 (adjust_bool_pattern): Likewise.
16760 (search_type_for_mask_1): Likewise.
16761 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
16762 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
16763 (vectorizable_load): Likewise.
16764 (vectorizable_store): Likewise.
16765 * ubsan.c (ubsan_encode_value): Likewise.
16766 * varasm.c (output_constant): Likewise.
16768 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16769 Alan Hayward <alan.hayward@arm.com>
16770 David Sherwood <david.sherwood@arm.com>
16772 * coretypes.h (scalar_mode): New class.
16773 * machmode.h (scalar_mode): Likewise.
16774 (scalar_mode::includes_p): New function.
16775 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
16776 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
16777 * genmodes.c (get_mode_class): Handle remaining scalar modes.
16778 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
16779 * expmed.c (store_bit_field_1): Likewise.
16780 (extract_bit_field_1): Likewise.
16781 * expr.c (write_complex_part): Likewise.
16782 (read_complex_part): Likewise.
16783 (emit_move_complex_push): Likewise.
16784 (expand_expr_real_2): Likewise.
16785 * function.c (assign_parm_setup_reg): Likewise.
16786 (assign_parms_unsplit_complex): Likewise.
16787 * optabs.c (expand_binop): Likewise.
16788 * rtlanal.c (subreg_get_info): Likewise.
16789 * simplify-rtx.c (simplify_immed_subreg): Likewise.
16790 * varasm.c (output_constant_pool_2): Likewise.
16792 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16793 Alan Hayward <alan.hayward@arm.com>
16794 David Sherwood <david.sherwood@arm.com>
16796 * expmed.c (extract_high_half): Use scalar_int_mode and remove
16798 (expmed_mult_highpart_optab): Likewise.
16799 (expmed_mult_highpart): Likewise.
16801 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16802 Alan Hayward <alan.hayward@arm.com>
16803 David Sherwood <david.sherwood@arm.com>
16805 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
16806 instead of a machine_mode.
16807 (builtin_memset_read_str): Likewise.
16808 * builtins.c (c_readstr): Likewise.
16809 (builtin_memcpy_read_str): Likewise.
16810 (builtin_strncpy_read_str): Likewise.
16811 (builtin_memset_read_str): Likewise.
16812 (builtin_memset_gen_str): Likewise.
16813 (expand_builtin_signbit): Use scalar_int_mode for local variables.
16814 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
16815 instead of a machine_mode.
16816 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
16818 (make_extraction): Likewise.
16819 (try_widen_shift_mode): Take and return scalar_int_modes instead
16821 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
16822 a scalar_int_mode instead of a machine_mode.
16823 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
16824 (avr_addr_space_pointer_mode): Likewise.
16825 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
16826 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
16827 (msp430_unwind_word_mode): Likewise.
16828 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
16829 (spu_addr_space_pointer_mode): Likewise.
16830 (spu_addr_space_address_mode): Likewise.
16831 (spu_libgcc_cmp_return_mode): Likewise.
16832 (spu_libgcc_shift_count_mode): Likewise.
16833 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
16834 (rl78_addr_space_pointer_mode): Likewise.
16835 (fl78_unwind_word_mode): Likewise.
16836 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
16838 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
16839 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
16840 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
16841 (mips_valid_pointer_mode): Likewise.
16842 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
16843 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
16844 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
16846 (ft32_addr_space_address_mode): Likewise.
16847 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
16848 scalar_int_mode instead of a machine_mode.
16849 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
16851 (m32c_addr_space_address_mode): Likewise.
16852 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
16853 (rs6000_eh_return_filter_mode): Likewise.
16854 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
16855 (rs6000_eh_return_filter_mode): Likewise.
16856 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
16857 (s390_libgcc_shift_count_mode): Likewise.
16858 (s390_unwind_word_mode): Likewise.
16859 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
16861 * target.def (mode_rep_extended): Likewise.
16862 (valid_pointer_mode): Likewise.
16863 (addr_space.valid_pointer_mode): Likewise.
16864 (eh_return_filter_mode): Return a scalar_int_mode rather than
16866 (libgcc_cmp_return_mode): Likewise.
16867 (libgcc_shift_count_mode): Likewise.
16868 (unwind_word_mode): Likewise.
16869 (addr_space.pointer_mode): Likewise.
16870 (addr_space.address_mode): Likewise.
16871 * doc/tm.texi: Regenerate.
16872 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
16874 (do_jump): Use scalar_int_mode for local variables.
16875 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
16876 rather than a machine_mode.
16877 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
16878 (scompare_loc_descriptor_wide): Likewise.
16879 (scompare_loc_descriptor_narrow): Likewise.
16880 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
16882 * except.c (sjlj_emit_dispatch_table): Likewise.
16883 (expand_builtin_eh_copy_values): Likewise.
16884 * explow.c (convert_memory_address_addr_space_1): Likewise.
16885 Take a scalar_int_mode rather than a machine_mode.
16886 (convert_memory_address_addr_space): Take a scalar_int_mode rather
16887 than a machine_mode.
16888 (memory_address_addr_space): Use scalar_int_mode for local variables.
16889 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
16890 rather than a machine_mode.
16891 * expmed.c (mask_rtx): Likewise.
16892 (init_expmed_one_conv): Likewise.
16893 (expand_mult_highpart_adjust): Likewise.
16894 (extract_high_half): Likewise.
16895 (expmed_mult_highpart_optab): Likewise.
16896 (expmed_mult_highpart): Likewise.
16897 (expand_smod_pow2): Likewise.
16898 (expand_sdiv_pow2): Likewise.
16899 (emit_store_flag_int): Likewise.
16900 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
16902 (extract_low_bits): Likewise.
16903 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
16905 * expr.c (pieces_addr::adjust): Likewise.
16906 (can_store_by_pieces): Likewise.
16907 (store_by_pieces): Likewise.
16908 (clear_by_pieces_1): Likewise.
16909 (expand_expr_addr_expr_1): Likewise.
16910 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
16911 (expand_expr_real_1): Likewise.
16912 (try_casesi): Likewise.
16913 * final.c (shorten_branches): Likewise.
16914 * fold-const.c (fold_convert_const_int_from_fixed): Change the
16915 type of "mode" to machine_mode.
16916 * internal-fn.c (expand_arith_overflow_result_store): Take a
16917 scalar_int_mode rather than a machine_mode.
16918 (expand_mul_overflow): Use scalar_int_mode for local variables.
16919 * loop-doloop.c (doloop_modify): Likewise.
16920 (doloop_optimize): Likewise.
16921 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
16922 than a machine_mode.
16923 (expand_doubleword_shift_condmove): Likewise.
16924 (expand_doubleword_shift): Likewise.
16925 (expand_doubleword_clz): Likewise.
16926 (expand_doubleword_popcount): Likewise.
16927 (expand_doubleword_parity): Likewise.
16928 (expand_absneg_bit): Use scalar_int_mode for local variables.
16929 (prepare_float_lib_cmp): Likewise.
16930 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
16931 rather than a machine_mode.
16932 (convert_memory_address_addr_space): Likewise.
16933 (get_mode_bounds): Likewise.
16934 (get_address_mode): Return a scalar_int_mode rather than a
16936 * rtlanal.c (get_address_mode): Likewise.
16937 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
16938 than a machine_mode.
16939 * targhooks.c (default_mode_rep_extended): Likewise.
16940 (default_valid_pointer_mode): Likewise.
16941 (default_addr_space_valid_pointer_mode): Likewise.
16942 (default_eh_return_filter_mode): Return a scalar_int_mode rather
16943 than a machine_mode.
16944 (default_libgcc_cmp_return_mode): Likewise.
16945 (default_libgcc_shift_count_mode): Likewise.
16946 (default_unwind_word_mode): Likewise.
16947 (default_addr_space_pointer_mode): Likewise.
16948 (default_addr_space_address_mode): Likewise.
16949 * targhooks.h (default_eh_return_filter_mode): Likewise.
16950 (default_libgcc_cmp_return_mode): Likewise.
16951 (default_libgcc_shift_count_mode): Likewise.
16952 (default_unwind_word_mode): Likewise.
16953 (default_addr_space_pointer_mode): Likewise.
16954 (default_addr_space_address_mode): Likewise.
16955 (default_mode_rep_extended): Take a scalar_int_mode rather than
16957 (default_valid_pointer_mode): Likewise.
16958 (default_addr_space_valid_pointer_mode): Likewise.
16959 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
16961 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
16962 rather than a machine_mode.
16963 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
16964 for local variables.
16965 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
16966 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
16967 than a machine_mode.
16969 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16970 Alan Hayward <alan.hayward@arm.com>
16971 David Sherwood <david.sherwood@arm.com>
16973 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
16974 the mode argument to scalar_int_mode.
16975 (do_jump_by_parts_zero_rtx): Likewise.
16976 (do_jump_by_parts_equality_rtx): Likewise.
16977 (do_jump_by_parts_greater): Take a mode argument.
16978 (do_jump_by_parts_equality): Likewise.
16979 (do_jump_1): Update calls accordingly.
16981 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16982 Alan Hayward <alan.hayward@arm.com>
16983 David Sherwood <david.sherwood@arm.com>
16985 * is-a.h (safe_dyn_cast): New function.
16986 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
16987 (jump_table_for_label): Likewise.
16988 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
16989 instead of an rtx_insn *.
16990 (shorten_branches): Use dyn_cast instead of LABEL_P and
16991 JUMP_TABLE_DATA_P. Use jump_table_for_label and
16992 rtx_jump_table_data::get_data_mode.
16993 (final_scan_insn): Likewise.
16995 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16996 Alan Hayward <alan.hayward@arm.com>
16997 David Sherwood <david.sherwood@arm.com>
16999 * combine.c (try_combine): Use is_a <scalar_int_mode> when
17000 trying to combine a full-register integer set with a subreg
17003 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17004 Alan Hayward <alan.hayward@arm.com>
17005 David Sherwood <david.sherwood@arm.com>
17007 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
17008 that is always either address_mode or pointer_mode.
17010 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17011 Alan Hayward <alan.hayward@arm.com>
17012 David Sherwood <david.sherwood@arm.com>
17014 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
17015 when the two are known to be equal.
17017 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17018 Alan Hayward <alan.hayward@arm.com>
17019 David Sherwood <david.sherwood@arm.com>
17021 * simplify-rtx.c (simplify_const_unary_operation): Use
17022 is_a <scalar_int_mode> instead of checking for a nonzero
17023 precision. Forcibly convert op_mode to a scalar_int_mode
17024 in that case. More clearly differentiate the operand and
17025 result modes and use the former when deciding what the value
17026 of a count-bits operation should be. Use is_int_mode instead
17027 of checking for a MODE_INT. Remove redundant check for whether
17028 this mode has a zero precision.
17030 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17031 Alan Hayward <alan.hayward@arm.com>
17032 David Sherwood <david.sherwood@arm.com>
17034 * optabs.c (widen_leading): Change the type of the mode argument
17035 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
17036 (widen_bswap): Likewise.
17037 (expand_parity): Likewise.
17038 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
17039 (expand_ffs): Likewise.
17040 (epand_unop): Check for scalar integer modes before calling the
17043 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17044 Alan Hayward <alan.hayward@arm.com>
17045 David Sherwood <david.sherwood@arm.com>
17047 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
17049 (expand_expr_real_1): Update call accordingly.
17051 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17052 Alan Hayward <alan.hayward@arm.com>
17053 David Sherwood <david.sherwood@arm.com>
17055 * expmed.c (store_bit_field_using_insv): Add op0_mode and
17056 value_mode arguments. Use scalar_int_mode internally.
17057 (store_bit_field_1): Rename the new integer mode from imode
17058 to op0_mode and use it instead of GET_MODE (op0). Update calls
17059 to store_split_bit_field, store_bit_field_using_insv and
17060 store_fixed_bit_field.
17061 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
17062 Use scalar_int_mode internally. Use a bit count rather than a mode
17063 when calculating the largest bit size for get_best_mode.
17064 Update calls to store_split_bit_field and store_fixed_bit_field_1.
17065 (store_fixed_bit_field_1): Add mode and value_mode arguments.
17066 Remove assertion that OP0 has a scalar integer mode.
17067 (store_split_bit_field): Add op0_mode and value_mode arguments.
17068 Update calls to extract_fixed_bit_field.
17069 (extract_bit_field_using_extv): Add an op0_mode argument.
17070 Use scalar_int_mode internally.
17071 (extract_bit_field_1): Rename the new integer mode from imode to
17072 op0_mode and use it instead of GET_MODE (op0). Update calls to
17073 extract_split_bit_field, extract_bit_field_using_extv and
17074 extract_fixed_bit_field.
17075 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
17076 to extract_split_bit_field and extract_fixed_bit_field_1.
17077 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
17078 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
17079 on the target mode.
17080 (extract_split_bit_field): Add an op0_mode argument. Update call
17081 to extract_fixed_bit_field.
17083 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17084 Alan Hayward <alan.hayward@arm.com>
17085 David Sherwood <david.sherwood@arm.com>
17087 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
17088 * explow.c (hard_function_value): Likewise.
17089 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
17090 convert_to_mode call outside the loop.
17091 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
17092 for the mode iterator. Require the mode specified by max_pieces
17094 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
17096 (copy_blkmode_to_reg): Likewise.
17097 (set_storage_via_setmem): Likewise.
17098 * optabs.c (prepare_cmp_insn): Likewise.
17099 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
17100 * stor-layout.c (finish_bitfield_representative): Likewise.
17102 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17103 Alan Hayward <alan.hayward@arm.com>
17104 David Sherwood <david.sherwood@arm.com>
17106 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
17107 * expr.c (convert_move): Use them.
17108 (convert_modes): Likewise.
17109 (store_expr_with_bounds): Likewise.
17111 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17112 Alan Hayward <alan.hayward@arm.com>
17113 David Sherwood <david.sherwood@arm.com>
17115 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
17116 parameter for the mode of "x". Remove the "known_x", "known_mode"
17117 and "known_ret" arguments. Change the type of the mode argument
17118 to scalar_int_mode.
17119 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
17120 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
17121 (reg_num_sign_bit_copies_for_combine): Likewise.
17122 * rtlanal.c (nonzero_bits1): Likewise.
17123 (num_sign_bit_copies1): Likewise.
17124 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
17125 (reg_num_sign_bit_copies_general): Likewise.
17126 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
17127 (reg_nonzero_bits_general): Likewise.
17129 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17130 Alan Hayward <alan.hayward@arm.com>
17131 David Sherwood <david.sherwood@arm.com>
17133 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
17134 than in subroutines. Return 1 for non-integer modes.
17135 (cached_num_sign_bit_copies): Change the type of the mode parameter
17136 to scalar_int_mode.
17137 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
17138 classes. Handle CONST_INT_P first and then check whether X also
17139 has a scalar integer mode. Check the same thing for inner registers
17140 of a SUBREG and for values that are being extended or truncated.
17142 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17143 Alan Hayward <alan.hayward@arm.com>
17144 David Sherwood <david.sherwood@arm.com>
17146 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
17147 in subroutines. Return the mode mask for non-integer modes.
17148 (cached_nonzero_bits): Change the type of the mode parameter
17149 to scalar_int_mode.
17150 (nonzero_bits1): Likewise. Remove early exit for other mode
17151 classes. Handle CONST_INT_P first and then check whether X
17152 also has a scalar integer mode.
17154 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17155 Alan Hayward <alan.hayward@arm.com>
17156 David Sherwood <david.sherwood@arm.com>
17158 * expr.c (widest_int_mode_for_size): Make the comment match the code.
17159 Return a scalar_int_mode and assert that the size is greater than
17161 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
17162 (op_by_pieces_d::op_by_pieces_d): Likewise.
17163 (op_by_pieces_d::run): Likewise.
17164 (can_store_by_pieces): Likewise.
17166 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17167 Alan Hayward <alan.hayward@arm.com>
17168 David Sherwood <david.sherwood@arm.com>
17170 * combine.c (extract_left_shift): Add a mode argument and update
17172 (make_compound_operation_int): Change the type of the mode parameter
17173 to scalar_int_mode and update the call to extract_left_shift.
17175 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17176 Alan Hayward <alan.hayward@arm.com>
17177 David Sherwood <david.sherwood@arm.com>
17179 * combine.c (simplify_and_const_int): Change the type of the mode
17180 parameter to scalar_int_mode.
17181 (simplify_and_const_int_1): Likewise. Update recursive call.
17183 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17184 Alan Hayward <alan.hayward@arm.com>
17185 David Sherwood <david.sherwood@arm.com>
17187 * combine.c (simplify_compare_const): Check that the mode is a
17188 scalar_int_mode (rather than VOIDmode) before testing its
17190 (simplify_comparison): Move COMPARISON_P handling out of the
17191 loop and restrict the latter part of the loop to scalar_int_modes.
17192 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
17193 and when considering SUBREG_REGs. Use is_int_mode instead of
17194 checking GET_MODE_CLASS against MODE_INT.
17196 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17197 Alan Hayward <alan.hayward@arm.com>
17198 David Sherwood <david.sherwood@arm.com>
17200 * combine.c (try_widen_shift_mode): Move check for equal modes to...
17201 (simplify_shift_const_1): ...here. Use scalar_int_mode for
17202 shift_unit_mode and for modes involved in scalar shifts.
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 (force_int_to_mode): New function, split out from...
17209 (force_to_mode): ...here. Keep xmode up-to-date and use it
17210 instead of GET_MODE (x).
17212 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17213 Alan Hayward <alan.hayward@arm.com>
17214 David Sherwood <david.sherwood@arm.com>
17216 * optabs-query.h (extraction_insn::struct_mode): Change type to
17217 opt_scalar_int_mode and update comment.
17218 (extraction_insn::field_mode): Change type to scalar_int_mode.
17219 (extraction_insn::pos_mode): Likewise.
17220 * combine.c (make_extraction): Update accordingly.
17221 * optabs-query.c (get_traditional_extraction_insn): Likewise.
17222 (get_optab_extraction_insn): Likewise.
17223 * recog.c (simplify_while_replacing): Likewise.
17224 * expmed.c (narrow_bit_field_mem): Change the type of the mode
17225 parameter to opt_scalar_int_mode.
17227 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17228 Alan Hayward <alan.hayward@arm.com>
17229 David Sherwood <david.sherwood@arm.com>
17231 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
17232 to a scalar_int_mode instead of a machine_mode.
17233 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
17234 (get_best_mode): Return a boolean and use a pointer argument to store
17235 the selected mode. Replace the limit mode parameter with a bit limit.
17236 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
17237 for the values returned by bit_field_mode_iterator::next_mode.
17238 (store_bit_field): Update call to get_best_mode.
17239 (store_fixed_bit_field): Likewise.
17240 (extract_fixed_bit_field): Likewise.
17241 * expr.c (optimize_bitfield_assignment_op): Likewise.
17242 * fold-const.c (optimize_bit_field_compare): Likewise.
17243 (fold_truth_andor_1): Likewise.
17244 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
17245 Update for new type of m_mode.
17246 (get_best_mode): As above.
17248 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17249 Alan Hayward <alan.hayward@arm.com>
17250 David Sherwood <david.sherwood@arm.com>
17252 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
17253 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
17254 (store_bit_field): Check is_a <scalar_int_mode> before calling
17255 strict_volatile_bitfield_p.
17256 (extract_bit_field): Likewise.
17258 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17259 Alan Hayward <alan.hayward@arm.com>
17260 David Sherwood <david.sherwood@arm.com>
17262 * target.def (cstore_mode): Return a scalar_int_mode.
17263 * doc/tm.texi: Regenerate.
17264 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
17265 * targhooks.h (default_cstore_mode): Likewise.
17266 * targhooks.c (default_cstore_mode): Likewise, using a forced
17268 * expmed.c (emit_cstore): Expect the target of the cstore to be
17271 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17272 Alan Hayward <alan.hayward@arm.com>
17273 David Sherwood <david.sherwood@arm.com>
17275 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
17277 (niter_desc): Likewise mode.
17278 (iv_analyze): Add a mode parameter.
17280 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
17281 and change its type to scalar_int_mode.
17282 * loop-iv.c: Update commentary at head of file.
17283 (iv_constant): Pass the mode paraeter before the rtx it describes
17284 and change its type to scalar_int_mode. Remove VOIDmode handling.
17285 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
17286 (iv_extend): Likewise.
17287 (shorten_into_mode): Likewise.
17288 (iv_add): Use scalar_int_mode.
17289 (iv_mult): Likewise.
17290 (iv_shift): Likewise.
17291 (canonicalize_iv_subregs): Likewise.
17292 (get_biv_step_1): Pass the outer_mode parameter before the rtx
17293 it describes and change its mode to scalar_int_mode. Also change
17294 the type of the returned inner_mode to scalar_int_mode.
17295 (get_biv_step): Likewise, turning outer_mode from a pointer
17296 into a direct parameter. Update call to get_biv_step_1.
17297 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
17298 iv_constant and get_biv_step.
17299 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
17300 and change its type to scalar_int_mode. Don't initialise iv->mode
17301 to VOIDmode and remove later checks for its still being VOIDmode.
17302 Update calls to iv_analyze_op and iv_analyze_expr. Check
17303 is_a <scalar_int_mode> when changing the mode under consideration.
17304 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
17305 Update call to iv_analyze_expr.
17306 (iv_analyze_op): Add a mode parameter. Reject subregs whose
17307 inner register is not also a scalar_int_mode. Update call to
17309 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
17310 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
17311 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
17312 separate mode class checks. Update calls to iv_analyze. Remove
17313 fix-up of VOIDmodes after iv_analyze_biv.
17314 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
17315 don't have a scalar_int_mode. Update call to biv_p.
17317 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17318 Alan Hayward <alan.hayward@arm.com>
17319 David Sherwood <david.sherwood@arm.com>
17321 * cfgexpand.c (convert_debug_memory_address): Use
17322 as_a <scalar_int_mode>.
17323 * combine.c (expand_compound_operation): Likewise.
17324 (make_extraction): Likewise.
17325 (change_zero_ext): Likewise.
17326 (simplify_comparison): Likewise.
17327 * cse.c (cse_insn): Likewise.
17328 * dwarf2out.c (minmax_loc_descriptor): Likewise.
17329 (mem_loc_descriptor): Likewise.
17330 (loc_descriptor): Likewise.
17331 * expmed.c (init_expmed_one_mode): Likewise.
17332 (synth_mult): Likewise.
17333 (emit_store_flag_1): Likewise.
17334 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
17335 of a comparison with size.
17336 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
17337 (reduce_to_bit_field_precision): Likewise.
17338 * function.c (expand_function_end): Likewise.
17339 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
17340 * loop-doloop.c (doloop_modify): Likewise.
17341 * optabs.c (expand_binop): Likewise.
17342 (expand_unop): Likewise.
17343 (expand_copysign_absneg): Likewise.
17344 (prepare_cmp_insn): Likewise.
17345 (maybe_legitimize_operand): Likewise.
17346 * recog.c (const_scalar_int_operand): Likewise.
17347 * rtlanal.c (get_address_mode): Likewise.
17348 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17349 (simplify_cond_clz_ctz): Likewise.
17350 * tree-nested.c (get_nl_goto_field): Likewise.
17351 * tree.c (build_vector_type_for_mode): Likewise.
17352 * var-tracking.c (use_narrower_mode): Likewise.
17354 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17355 Alan Hayward <alan.hayward@arm.com>
17356 David Sherwood <david.sherwood@arm.com>
17358 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
17359 * builtins.c (expand_builtin_signbit): Use it.
17360 * cfgexpand.c (expand_debug_expr): Likewise.
17361 * dojump.c (do_jump): Likewise.
17362 (do_compare_and_jump): Likewise.
17363 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
17364 * expmed.c (make_tree): Likewise.
17365 * expr.c (expand_expr_real_2): Likewise.
17366 (expand_expr_real_1): Likewise.
17367 (try_casesi): Likewise.
17368 * fold-const-call.c (fold_const_call_ss): Likewise.
17369 * fold-const.c (unextend): Likewise.
17370 (extract_muldiv_1): Likewise.
17371 (fold_single_bit_test): Likewise.
17372 (native_encode_int): Likewise.
17373 (native_encode_string): Likewise.
17374 (native_interpret_int): Likewise.
17375 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
17376 * internal-fn.c (expand_addsub_overflow): Likewise.
17377 (expand_neg_overflow): Likewise.
17378 (expand_mul_overflow): Likewise.
17379 (expand_arith_overflow): Likewise.
17380 * match.pd: Likewise.
17381 * stor-layout.c (layout_type): Likewise.
17382 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
17383 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
17384 * tree-ssanames.c (get_range_info): Likewise.
17385 * tree-switch-conversion.c (array_value_type) Likewise.
17386 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
17387 (vect_recog_divmod_pattern): Likewise.
17388 (vect_recog_mixed_size_cond_pattern): Likewise.
17389 * tree-vrp.c (extract_range_basic): Likewise.
17390 (simplify_float_conversion_using_ranges): Likewise.
17391 * tree.c (int_fits_type_p): Likewise.
17392 * ubsan.c (instrument_bool_enum_load): Likewise.
17393 * varasm.c (mergeable_string_section): Likewise.
17394 (narrowing_initializer_constant_valid_p): Likewise.
17395 (output_constant): Likewise.
17397 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17398 Alan Hayward <alan.hayward@arm.com>
17399 David Sherwood <david.sherwood@arm.com>
17401 * machmode.h (NARROWEST_INT_MODE): New macro.
17402 * expr.c (alignment_for_piecewise_move): Use it instead of
17403 GET_CLASS_NARROWEST_MODE (MODE_INT).
17404 (push_block): Likewise.
17405 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
17407 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
17409 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17410 Alan Hayward <alan.hayward@arm.com>
17411 David Sherwood <david.sherwood@arm.com>
17413 * postreload.c (move2add_valid_value_p): Change the type of the
17414 mode parameter to scalar_int_mode.
17415 (move2add_use_add2_insn): Add a mode parameter and use it instead
17417 (move2add_use_add3_insn): Likewise.
17418 (reload_cse_move2add): Update accordingly.
17420 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17421 Alan Hayward <alan.hayward@arm.com>
17422 David Sherwood <david.sherwood@arm.com>
17424 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
17426 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
17427 * optabs.c (expand_unop): Likewise.
17429 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17430 Alan Hayward <alan.hayward@arm.com>
17431 David Sherwood <david.sherwood@arm.com>
17433 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
17434 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
17435 (popcount_loc_descriptor): Likewise.
17436 (bswap_loc_descriptor): Likewise.
17437 (rotate_loc_descriptor): Likewise.
17438 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
17439 calling the functions above.
17441 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17442 Alan Hayward <alan.hayward@arm.com>
17443 David Sherwood <david.sherwood@arm.com>
17445 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
17447 (try_combine): Likewise.
17448 (simplify_if_then_else): Likewise.
17449 * cse.c (cse_insn): Likewise.
17450 * dwarf2out.c (mem_loc_descriptor): Likewise.
17451 * emit-rtl.c (gen_lowpart_common): Likewise.
17452 * simplify-rtx.c (simplify_truncation): Likewise.
17453 (simplify_binary_operation_1): Likewise.
17454 (simplify_const_relational_operation): Likewise.
17455 (simplify_ternary_operation): Likewise.
17456 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
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 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
17463 * reload.c (push_reload): Likewise.
17464 (find_reloads): Likewise.
17466 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17467 Alan Hayward <alan.hayward@arm.com>
17468 David Sherwood <david.sherwood@arm.com>
17470 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
17471 (make_compound_operation_int): Likewise.
17472 (change_zero_ext): Likewise.
17473 * expr.c (convert_move): Likewise.
17474 (convert_modes): Likewise.
17475 * fwprop.c (forward_propagate_subreg): Likewise.
17476 * loop-iv.c (get_biv_step_1): Likewise.
17477 * optabs.c (widen_operand): Likewise.
17478 * postreload.c (move2add_valid_value_p): Likewise.
17479 * recog.c (simplify_while_replacing): Likewise.
17480 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17481 (simplify_binary_operation_1): Likewise. Remove redundant
17482 mode equality check.
17484 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17485 Alan Hayward <alan.hayward@arm.com>
17486 David Sherwood <david.sherwood@arm.com>
17488 * combine.c (combine_simplify_rtx): Add checks for
17489 is_a <scalar_int_mode>.
17490 (simplify_if_then_else): Likewise.
17491 (make_field_assignment): Likewise.
17492 (simplify_comparison): Likewise.
17493 * ifcvt.c (noce_try_bitop): Likewise.
17494 * loop-invariant.c (canonicalize_address_mult): Likewise.
17495 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17497 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17498 Alan Hayward <alan.hayward@arm.com>
17499 David Sherwood <david.sherwood@arm.com>
17501 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
17502 is_a <scalar_int_mode> instead of != BLKmode.
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 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
17509 instead of != VOIDmode.
17510 * combine.c (if_then_else_cond): Likewise.
17511 (change_zero_ext): Likewise.
17512 * dwarf2out.c (mem_loc_descriptor): Likewise.
17513 (loc_descriptor): Likewise.
17514 * rtlanal.c (canonicalize_condition): Likewise.
17515 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
17517 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17518 Alan Hayward <alan.hayward@arm.com>
17519 David Sherwood <david.sherwood@arm.com>
17521 * simplify-rtx.c (simplify_binary_operation_1): Use
17522 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
17524 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17525 Alan Hayward <alan.hayward@arm.com>
17526 David Sherwood <david.sherwood@arm.com>
17528 * wide-int.h (int_traits<unsigned char>) New class.
17529 (int_traits<unsigned short>) Likewise.
17530 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
17531 Use GET_MODE_UNIT_PRECISION and remove redundant test for
17533 * combine.c (set_nonzero_bits_and_sign_copies): Use
17534 is_a <scalar_int_mode>.
17535 (find_split_point): Likewise.
17536 (combine_simplify_rtx): Likewise.
17537 (simplify_logical): Likewise.
17538 (expand_compound_operation): Likewise.
17539 (expand_field_assignment): Likewise.
17540 (make_compound_operation): Likewise.
17541 (extended_count): Likewise.
17542 (change_zero_ext): Likewise.
17543 (simplify_comparison): Likewise.
17544 * dwarf2out.c (scompare_loc_descriptor): Likewise.
17545 (ucompare_loc_descriptor): Likewise.
17546 (minmax_loc_descriptor): Likewise.
17547 (mem_loc_descriptor): Likewise.
17548 (loc_descriptor): Likewise.
17549 * expmed.c (init_expmed_one_mode): Likewise.
17550 * lra-constraints.c (lra_constraint_offset): Likewise.
17551 * optabs.c (prepare_libcall_arg): Likewise.
17552 * postreload.c (move2add_note_store): Likewise.
17553 * reload.c (operands_match_p): Likewise.
17554 * rtl.h (load_extend_op): Likewise.
17555 * rtlhooks.c (gen_lowpart_general): Likewise.
17556 * simplify-rtx.c (simplify_truncation): Likewise.
17557 (simplify_unary_operation_1): Likewise.
17558 (simplify_binary_operation_1): Likewise.
17559 (simplify_const_binary_operation): Likewise.
17560 (simplify_const_relational_operation): Likewise.
17561 (simplify_subreg): Likewise.
17562 * stor-layout.c (bitwise_mode_for_mode): Likewise.
17563 * var-tracking.c (adjust_mems): Likewise.
17564 (prepare_call_arguments): Likewise.
17566 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17567 Alan Hayward <alan.hayward@arm.com>
17568 David Sherwood <david.sherwood@arm.com>
17570 * machmode.h (is_int_mode): New fuction.
17571 * combine.c (find_split_point): Use it.
17572 (combine_simplify_rtx): Likewise.
17573 (simplify_if_then_else): Likewise.
17574 (simplify_set): Likewise.
17575 (simplify_shift_const_1): Likewise.
17576 (simplify_comparison): Likewise.
17577 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
17578 * cse.c (notreg_cost): Likewise.
17579 (cse_insn): Likewise.
17580 * cselib.c (cselib_lookup_1): Likewise.
17581 * dojump.c (do_jump_1): Likewise.
17582 (do_compare_rtx_and_jump): Likewise.
17583 * dse.c (get_call_args): Likewise.
17584 * dwarf2out.c (rtl_for_decl_init): Likewise.
17585 (native_encode_initializer): Likewise.
17586 * expmed.c (emit_store_flag_1): Likewise.
17587 (emit_store_flag): Likewise.
17588 * expr.c (convert_modes): Likewise.
17589 (store_field): Likewise.
17590 (expand_expr_real_1): Likewise.
17591 * fold-const.c (fold_read_from_constant_string): Likewise.
17592 * gimple-ssa-sprintf.c (get_format_string): Likewise.
17593 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
17594 * optabs.c (expand_binop): Likewise.
17595 (expand_unop): Likewise.
17596 (expand_abs_nojump): Likewise.
17597 (expand_one_cmpl_abs_nojump): Likewise.
17598 * simplify-rtx.c (mode_signbit_p): Likewise.
17599 (val_signbit_p): Likewise.
17600 (val_signbit_known_set_p): Likewise.
17601 (val_signbit_known_clear_p): Likewise.
17602 (simplify_relational_operation_1): Likewise.
17603 * tree.c (vector_type_mode): Likewise.
17605 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17606 Alan Hayward <alan.hayward@arm.com>
17607 David Sherwood <david.sherwood@arm.com>
17609 * machmode.h (smallest_mode_for_size): Fix formatting.
17610 (smallest_int_mode_for_size): New function.
17611 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
17612 instead of smallest_mode_for_size.
17613 * combine.c (make_extraction): Likewise.
17614 * config/arc/arc.c (arc_expand_movmem): Likewise.
17615 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
17616 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
17617 * config/s390/s390.c (s390_expand_insv): Likewise.
17618 * config/sparc/sparc.c (assign_int_registers): Likewise.
17619 * config/spu/spu.c (spu_function_value): Likewise.
17620 (spu_function_arg): Likewise.
17621 * coverage.c (get_gcov_type): Likewise.
17622 (get_gcov_unsigned_t): Likewise.
17623 * dse.c (find_shift_sequence): Likewise.
17624 * expmed.c (store_bit_field_1): Likewise.
17625 * expr.c (convert_move): Likewise.
17626 (store_field): Likewise.
17627 * internal-fn.c (expand_arith_overflow): Likewise.
17628 * optabs-query.c (get_best_extraction_insn): Likewise.
17629 * optabs.c (expand_twoval_binop_libfunc): Likewise.
17630 * stor-layout.c (layout_type): Likewise.
17631 (initialize_sizetypes): Likewise.
17632 * targhooks.c (default_get_mask_mode): Likewise.
17633 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
17635 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17636 Alan Hayward <alan.hayward@arm.com>
17637 David Sherwood <david.sherwood@arm.com>
17639 * machmode.h (opt_mode::else_blk): New function.
17640 (int_mode_for_mode): Declare.
17641 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
17642 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
17644 * cfgexpand.c (expand_debug_expr): Likewise.
17645 * combine.c (gen_lowpart_or_truncate): Likewise.
17646 (gen_lowpart_for_combine): Likewise.
17647 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
17648 * config/avr/avr.c (avr_to_int_mode): Likewise.
17649 (avr_out_plus_1): Likewise.
17650 (avr_out_plus): Likewise.
17651 (avr_out_round): Likewise.
17652 * config/i386/i386.c (ix86_split_to_parts): Likewise.
17653 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
17654 (s390_expand_vcond): Likewise.
17655 * config/spu/spu.c (spu_split_immediate): Likewise.
17656 (spu_expand_mov): Likewise.
17657 * dse.c (get_stored_val): Likewise.
17658 * expmed.c (store_bit_field_1): Likewise.
17659 (convert_extracted_bit_field): Use int_mode_for_mode instead of
17661 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
17662 (extract_low_bits): Likewise.
17663 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
17664 handling rather than repeating the check.
17665 (emit_group_store): Likewise.
17666 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
17667 * optabs.c (expand_absneg_bit): Likewise.
17668 (expand_copysign_absneg): Likewise.
17669 (expand_copysign_bit): Likewise.
17670 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
17671 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
17672 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
17673 * var-tracking.c (prepare_call_arguments): Likewise.
17674 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
17675 int_mode_for_mode instead of mode_for_size.
17676 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
17678 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17679 Alan Hayward <alan.hayward@arm.com>
17680 David Sherwood <david.sherwood@arm.com>
17682 * machmode.h (int_mode_for_size): New function.
17683 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
17684 instead of mode_for_size.
17685 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
17687 * combine.c (expand_field_assignment): Use int_mode_for_size
17688 instead of mode_for_size.
17689 (make_extraction): Likewise.
17690 (simplify_shift_const_1): Likewise.
17691 (simplify_comparison): Likewise.
17692 * dojump.c (do_jump): Likewise.
17693 * dwarf2out.c (mem_loc_descriptor): Likewise.
17694 * emit-rtl.c (init_derived_machine_modes): Likewise.
17695 * expmed.c (flip_storage_order): Likewise.
17696 (convert_extracted_bit_field): Likewise.
17697 * expr.c (copy_blkmode_from_reg): Likewise.
17698 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
17699 * internal-fn.c (expand_mul_overflow): Likewise.
17700 * lower-subreg.c (simple_move): Likewise.
17701 * optabs-libfuncs.c (init_optabs): Likewise.
17702 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17703 * tree.c (vector_type_mode): Likewise.
17704 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
17705 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
17706 * tree-vect-generic.c (expand_vector_parallel): Likewise.
17707 * tree-vect-stmts.c (vectorizable_load): Likewise.
17708 (vectorizable_store): Likewise.
17710 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17711 Alan Hayward <alan.hayward@arm.com>
17712 David Sherwood <david.sherwood@arm.com>
17714 * coretypes.h (pod_mode): New type.
17715 (scalar_int_mode_pod): New typedef.
17716 * machmode.h (pod_mode): New class.
17717 (int_n_data_t::m): Change type to scalar_int_mode_pod.
17718 * genmodes.c (emit_mode_int_n): Update accordingly.
17719 * lower-subreg.h (target_lower_subreg): Change type to
17720 scalar_int_mode_pod.
17721 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
17722 scalar_int_mode_pod.
17724 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17725 Alan Hayward <alan.hayward@arm.com>
17726 David Sherwood <david.sherwood@arm.com>
17728 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
17729 machine_mode to scalar_int_mode.
17730 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
17731 (rs6000_option_override_internal): Remove cast to int.
17732 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
17733 machine_mode to scalar_int_mode.
17734 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
17735 (rs6000_option_override_internal): Remove cast to int.
17736 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
17737 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
17739 * config/s390/s390.c (s390_expand_builtin): Likewise.
17740 * coretypes.h (scalar_int_mode): New type.
17741 (opt_scalar_int_mode): New typedef.
17742 * machmode.h (scalar_int_mode): New class.
17743 (scalar_int_mode::includes_p): New function.
17744 (byte_mode): Change type to scalar_int_mode.
17745 (word_mode): Likewise.
17746 (ptr_mode): Likewise.
17747 * emit-rtl.c (byte_mode): Likewise.
17748 (word_mode): Likewise.
17749 (ptr_mode): Likewise.
17750 (init_derived_machine_modes): Update accordingly.
17751 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
17752 and MODE_PARTIAL_INT.
17753 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
17754 opt_scalar_int_mode.
17756 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17757 Alan Hayward <alan.hayward@arm.com>
17758 David Sherwood <david.sherwood@arm.com>
17760 * target.def (libgcc_floating_mode_supported_p): Take a
17762 * doc/tm.texi: Regenerate.
17763 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
17765 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
17766 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
17769 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17770 Alan Hayward <alan.hayward@arm.com>
17771 David Sherwood <david.sherwood@arm.com>
17773 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
17774 * doc/tm.texi: Regenerate.
17775 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
17776 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
17777 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
17778 * targhooks.h (default_floatn_mode): Likewise.
17779 * targhooks.c (default_floatn_mode): Likewise.
17780 * tree.c (build_common_tree_nodes): Update accordingly.
17782 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17783 Alan Hayward <alan.hayward@arm.com>
17784 David Sherwood <david.sherwood@arm.com>
17786 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
17787 (mode_iterator::iterate_p): Likewise.
17788 (mode_iterator::get_wider): Likewise.
17789 * expr.c (init_expr_target): Use opt_scalar_float_mode.
17791 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17792 Alan Hayward <alan.hayward@arm.com>
17793 David Sherwood <david.sherwood@arm.com>
17795 * coretypes.h (opt_scalar_float_mode): New typedef.
17796 * machmode.h (float_mode_for_size): New function.
17797 * emit-rtl.c (double_mode): Delete.
17798 (init_emit_once): Use float_mode_for_size.
17799 * stor-layout.c (layout_type): Likewise.
17800 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
17802 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17803 Alan Hayward <alan.hayward@arm.com>
17804 David Sherwood <david.sherwood@arm.com>
17806 * output.h (assemble_real): Take a scalar_float_mode.
17807 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
17808 * config/arm/arm.md (consttable_4): Likewise.
17809 (consttable_8): Likewise.
17810 (consttable_16): Likewise.
17811 * config/mips/mips.md (consttable_float): Likewise.
17812 * config/s390/s390.c (s390_output_pool_entry): Likewise.
17813 * varasm.c (assemble_real): Take a scalar_float_mode.
17814 (output_constant_pool_2): Update accordingly.
17815 (output_constant): Likewise.
17817 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17818 Alan Hayward <alan.hayward@arm.com>
17819 David Sherwood <david.sherwood@arm.com>
17821 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
17822 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
17823 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
17824 (native_encode_real): Likewise.
17825 (native_interpret_real): Likewise.
17826 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
17827 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
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 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
17834 <scalar_float_mode>. Simplify.
17835 (gen_extend_conv_libfunc): Likewise.
17837 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17838 Alan Hayward <alan.hayward@arm.com>
17839 David Sherwood <david.sherwood@arm.com>
17841 * coretypes.h (scalar_float_mode): New type.
17842 * machmode.h (mode_traits::from_int): Use machine_mode if
17843 USE_ENUM_MODES is defined.
17844 (is_a): New function.
17846 (dyn_cast): Likewise.
17847 (scalar_float_mode): New class.
17848 (scalar_float_mode::includes_p): New function.
17849 (is_float_mode): Likewise.
17850 * gdbhooks.py (MachineModePrinter): New class.
17851 (build_pretty_printer): Use it for scalar_float_mode.
17852 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
17853 (format_helper::format_helper): Turn into a template.
17854 * genmodes.c (get_mode_class): New function.
17855 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
17856 or machine_mode if none.
17857 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
17858 as_a <scalar_float_mode>.
17859 * dwarf2out.c (mem_loc_descriptor): Likewise.
17860 (insert_float): Likewise.
17861 (add_const_value_attribute): Likewise.
17862 * simplify-rtx.c (simplify_immed_subreg): Likewise.
17863 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
17864 (expand_unop): Update accordingly.
17865 (expand_abs_nojump): Likewise.
17866 (expand_copysign_absneg): Take a scalar_float_mode.
17867 (expand_copysign_bit): Likewise.
17868 (expand_copysign): Update accordingly.
17870 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17871 Alan Hayward <alan.hayward@arm.com>
17872 David Sherwood <david.sherwood@arm.com>
17874 * coretypes.h (opt_mode): New class.
17875 * machmode.h (opt_mode): Likewise.
17876 (opt_mode::else_void): New function.
17877 (opt_mode::require): Likewise.
17878 (opt_mode::exists): Likewise.
17879 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
17880 (GET_MODE_2XWIDER_MODE): Likewise.
17881 (mode_iterator::get_wider): Update accordingly.
17882 (mode_iterator::get_2xwider): Likewise.
17883 (mode_iterator::get_known_wider): Likewise, turning into a template.
17884 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
17885 forcing a wider mode to exist.
17886 * config/cr16/cr16.h (LONG_REG_P): Likewise.
17887 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
17888 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
17889 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
17890 * lower-subreg.c (init_lower_subreg): Likewise.
17891 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
17892 on the final iteration.
17893 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
17894 a wider mode exists before asking for a move pattern.
17895 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
17896 forcing a wider mode to exist.
17897 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
17898 returning false if no such mode exists.
17899 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
17900 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
17901 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
17902 Avoid checking for a MODE_INT if we already know the mode is not a
17904 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
17905 forcing a wider mode to exist.
17906 (expmed_mult_highpart_optab): Likewise.
17907 (expmed_mult_highpart): Likewise.
17908 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
17910 * lto-streamer-in.c (lto_input_mode_table): Likewise.
17911 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
17912 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
17913 * internal-fn.c (expand_mul_overflow): Update use of
17914 GET_MODE_2XWIDER_MODE.
17915 * omp-low.c (omp_clause_aligned_alignment): Likewise.
17916 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
17917 GET_MODE_WIDER_MODE.
17918 (convert_plusminus_to_widen): Likewise.
17919 * tree-switch-conversion.c (array_value_type): Likewise.
17920 * var-tracking.c (emit_note_insn_var_location): Likewise.
17921 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
17922 Return false inside rather than outside the loop if no wider mode
17924 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
17925 and GET_MODE_2XWIDER_MODE
17926 (can_compare_p): Use else_void.
17927 * gdbhooks.py (OptMachineModePrinter): New class.
17928 (build_pretty_printer): Use it for opt_mode.
17930 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17931 Alan Hayward <alan.hayward@arm.com>
17932 David Sherwood <david.sherwood@arm.com>
17934 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
17935 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
17937 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17938 Alan Hayward <alan.hayward@arm.com>
17939 David Sherwood <david.sherwood@arm.com>
17941 * machmode.h (mode_traits): New structure.
17942 (get_narrowest_mode): New function.
17943 (mode_iterator::start): Likewise.
17944 (mode_iterator::iterate_p): Likewise.
17945 (mode_iterator::get_wider): Likewise.
17946 (mode_iterator::get_known_wider): Likewise.
17947 (mode_iterator::get_2xwider): Likewise.
17948 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
17949 (FOR_EACH_MODE): Likewise.
17950 (FOR_EACH_MODE_FROM): Likewise.
17951 (FOR_EACH_MODE_UNTIL): Likewise.
17952 (FOR_EACH_WIDER_MODE): Likewise.
17953 (FOR_EACH_2XWIDER_MODE): Likewise.
17954 * builtins.c (expand_builtin_strlen): Use new mode iterators.
17955 * combine.c (simplify_comparison): Likewise
17956 * config/i386/i386.c (type_natural_mode): Likewise.
17957 * cse.c (cse_insn): Likewise.
17958 * dse.c (find_shift_sequence): Likewise.
17959 * emit-rtl.c (init_derived_machine_modes): Likewise.
17960 (init_emit_once): Likewise.
17961 * explow.c (hard_function_value): Likewise.
17962 * expmed.c (extract_fixed_bit_field_1): Likewise.
17963 (extract_bit_field_1): Likewise.
17964 (expand_divmod): Likewise.
17965 (emit_store_flag_1): Likewise.
17966 * expr.c (init_expr_target): Likewise.
17967 (convert_move): Likewise.
17968 (alignment_for_piecewise_move): Likewise.
17969 (widest_int_mode_for_size): Likewise.
17970 (emit_block_move_via_movmem): Likewise.
17971 (copy_blkmode_to_reg): Likewise.
17972 (set_storage_via_setmem): Likewise.
17973 (compress_float_constant): Likewise.
17974 * omp-low.c (omp_clause_aligned_alignment): Likewise.
17975 * optabs-query.c (get_best_extraction_insn): Likewise.
17976 * optabs.c (expand_binop): Likewise.
17977 (expand_twoval_unop): Likewise.
17978 (expand_twoval_binop): Likewise.
17979 (widen_leading): Likewise.
17980 (widen_bswap): Likewise.
17981 (expand_parity): Likewise.
17982 (expand_unop): Likewise.
17983 (prepare_cmp_insn): Likewise.
17984 (prepare_float_lib_cmp): Likewise.
17985 (expand_float): Likewise.
17986 (expand_fix): Likewise.
17987 (expand_sfix_optab): Likewise.
17988 * postreload.c (move2add_use_add2_insn): Likewise.
17989 * reg-stack.c (reg_to_stack): Likewise.
17990 * reginfo.c (choose_hard_reg_mode): Likewise.
17991 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
17992 * stor-layout.c (mode_for_size): Likewise.
17993 (smallest_mode_for_size): Likewise.
17994 (mode_for_vector): Likewise.
17995 (finish_bitfield_representative): Likewise.
17996 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
17997 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
17998 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17999 * var-tracking.c (prepare_call_arguments): Likewise.
18001 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18002 Alan Hayward <alan.hayward@arm.com>
18003 David Sherwood <david.sherwood@arm.com>
18005 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
18006 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
18007 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
18008 * machmode.h (mode_size): Move earlier in file.
18009 (mode_precision): Likewise.
18010 (mode_inner): Likewise.
18011 (mode_nunits): Likewise.
18012 (mode_unit_size): Likewise.
18013 (unit_unit_precision): Likewise.
18014 (mode_wider): Likewise.
18015 (mode_2xwider): Likewise.
18016 (machine_mode): New class.
18017 (mode_to_bytes): New function.
18018 (mode_to_bits): Likewise.
18019 (mode_to_precision): Likewise.
18020 (mode_to_inner): Likewise.
18021 (mode_to_unit_size): Likewise.
18022 (mode_to_unit_precision): Likewise.
18023 (mode_to_nunits): Likewise.
18024 (GET_MODE_SIZE): Use mode_to_bytes.
18025 (GET_MODE_BITSIZE): Use mode_to_bits.
18026 (GET_MODE_PRECISION): Use mode_to_precision.
18027 (GET_MODE_INNER): Use mode_to_inner.
18028 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
18029 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
18030 (GET_MODE_NUNITS): Use mode_to_nunits.
18031 * system.h (ALWAYS_INLINE): New macro.
18032 * config/powerpcspe/powerpcspe-c.c
18033 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
18034 int for arg1_mode and arg2_mode.
18036 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18037 Alan Hayward <alan.hayward@arm.com>
18038 David Sherwood <david.sherwood@arm.com>
18040 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
18041 Prefix mode names with E_ in case statements.
18042 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18043 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
18044 (aarch64_split_simd_move): Likewise.
18045 (aarch64_gen_storewb_pair): Likewise.
18046 (aarch64_gen_loadwb_pair): Likewise.
18047 (aarch64_gen_store_pair): Likewise.
18048 (aarch64_gen_load_pair): Likewise.
18049 (aarch64_get_condition_code_1): Likewise.
18050 (aarch64_constant_pool_reload_icode): Likewise.
18051 (get_rsqrte_type): Likewise.
18052 (get_rsqrts_type): Likewise.
18053 (get_recpe_type): Likewise.
18054 (get_recps_type): Likewise.
18055 (aarch64_gimplify_va_arg_expr): Likewise.
18056 (aarch64_simd_container_mode): Likewise.
18057 (aarch64_emit_load_exclusive): Likewise.
18058 (aarch64_emit_store_exclusive): Likewise.
18059 (aarch64_expand_compare_and_swap): Likewise.
18060 (aarch64_gen_atomic_cas): Likewise.
18061 (aarch64_emit_bic): Likewise.
18062 (aarch64_emit_atomic_swap): Likewise.
18063 (aarch64_emit_atomic_load_op): Likewise.
18064 (aarch64_evpc_trn): Likewise.
18065 (aarch64_evpc_uzp): Likewise.
18066 (aarch64_evpc_zip): Likewise.
18067 (aarch64_evpc_ext): Likewise.
18068 (aarch64_evpc_rev): Likewise.
18069 (aarch64_evpc_dup): Likewise.
18070 (aarch64_gen_ccmp_first): Likewise.
18071 (aarch64_gen_ccmp_next): Likewise.
18072 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
18073 (alpha_emit_xfloating_libcall): Likewise.
18074 (emit_insxl): Likewise.
18075 (alpha_arg_type): Likewise.
18076 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
18077 (arc_preferred_simd_mode): Likewise.
18078 (arc_secondary_reload): Likewise.
18079 (get_arc_condition_code): Likewise.
18080 (arc_print_operand): Likewise.
18081 (arc_legitimate_constant_p): Likewise.
18082 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18083 * config/arc/arc.md (casesi_load): Likewise.
18084 (casesi_compact_jump): Likewise.
18085 * config/arc/predicates.md (proper_comparison_operator): Likewise.
18086 (cc_use_register): Likewise.
18087 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18088 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
18089 (arm_init_iwmmxt_builtins): Likewise.
18090 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
18091 (neon_expand_vector_init): Likewise.
18092 (arm_attr_length_move_neon): Likewise.
18093 (maybe_get_arm_condition_code): Likewise.
18094 (arm_emit_vector_const): Likewise.
18095 (arm_preferred_simd_mode): Likewise.
18096 (arm_output_iwmmxt_tinsr): Likewise.
18097 (thumb1_output_casesi): Likewise.
18098 (thumb2_output_casesi): Likewise.
18099 (arm_emit_load_exclusive): Likewise.
18100 (arm_emit_store_exclusive): Likewise.
18101 (arm_expand_compare_and_swap): Likewise.
18102 (arm_evpc_neon_vuzp): Likewise.
18103 (arm_evpc_neon_vzip): Likewise.
18104 (arm_evpc_neon_vrev): Likewise.
18105 (arm_evpc_neon_vtrn): Likewise.
18106 (arm_evpc_neon_vext): Likewise.
18107 (arm_validize_comparison): Likewise.
18108 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
18109 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
18110 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
18111 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
18112 (c6x_preferred_simd_mode): Likewise.
18113 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
18114 (epiphany_rtx_costs): Likewise.
18115 * config/epiphany/predicates.md (proper_comparison_operator):
18117 * config/frv/frv.c (condexec_memory_operand): Likewise.
18118 (frv_emit_move): Likewise.
18119 (output_move_single): Likewise.
18120 (output_condmove_single): Likewise.
18121 (frv_hard_regno_mode_ok): Likewise.
18122 (frv_matching_accg_mode): Likewise.
18123 * config/h8300/h8300.c (split_adds_subs): Likewise.
18124 (h8300_rtx_costs): Likewise.
18125 (h8300_print_operand): Likewise.
18126 (compute_mov_length): Likewise.
18127 (output_logical_op): Likewise.
18128 (compute_logical_op_length): Likewise.
18129 (compute_logical_op_cc): Likewise.
18130 (h8300_shift_needs_scratch_p): Likewise.
18131 (output_a_shift): Likewise.
18132 (compute_a_shift_length): Likewise.
18133 (compute_a_shift_cc): Likewise.
18134 (expand_a_rotate): Likewise.
18135 (output_a_rotate): Likewise.
18136 * config/i386/i386.c (classify_argument): Likewise.
18137 (function_arg_advance_32): Likewise.
18138 (function_arg_32): Likewise.
18139 (function_arg_64): Likewise.
18140 (function_value_64): Likewise.
18141 (ix86_gimplify_va_arg): Likewise.
18142 (ix86_legitimate_constant_p): Likewise.
18143 (put_condition_code): Likewise.
18144 (split_double_mode): Likewise.
18145 (ix86_avx256_split_vector_move_misalign): Likewise.
18146 (ix86_expand_vector_logical_operator): Likewise.
18147 (ix86_split_idivmod): Likewise.
18148 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
18149 (ix86_build_const_vector): Likewise.
18150 (ix86_build_signbit_mask): Likewise.
18151 (ix86_match_ccmode): Likewise.
18152 (ix86_cc_modes_compatible): Likewise.
18153 (ix86_expand_branch): Likewise.
18154 (ix86_expand_sse_cmp): Likewise.
18155 (ix86_expand_sse_movcc): Likewise.
18156 (ix86_expand_int_sse_cmp): Likewise.
18157 (ix86_expand_vec_perm_vpermi2): Likewise.
18158 (ix86_expand_vec_perm): Likewise.
18159 (ix86_expand_sse_unpack): Likewise.
18160 (ix86_expand_int_addcc): Likewise.
18161 (ix86_split_to_parts): Likewise.
18162 (ix86_vectorize_builtin_gather): Likewise.
18163 (ix86_vectorize_builtin_scatter): Likewise.
18164 (avx_vpermilp_parallel): Likewise.
18165 (inline_memory_move_cost): Likewise.
18166 (ix86_tieable_integer_mode_p): Likewise.
18167 (x86_maybe_negate_const_int): Likewise.
18168 (ix86_expand_vector_init_duplicate): Likewise.
18169 (ix86_expand_vector_init_one_nonzero): Likewise.
18170 (ix86_expand_vector_init_one_var): Likewise.
18171 (ix86_expand_vector_init_concat): Likewise.
18172 (ix86_expand_vector_init_interleave): Likewise.
18173 (ix86_expand_vector_init_general): Likewise.
18174 (ix86_expand_vector_set): Likewise.
18175 (ix86_expand_vector_extract): Likewise.
18176 (emit_reduc_half): Likewise.
18177 (ix86_emit_i387_round): Likewise.
18178 (ix86_mangle_type): Likewise.
18179 (ix86_expand_round_sse4): Likewise.
18180 (expand_vec_perm_blend): Likewise.
18181 (canonicalize_vector_int_perm): Likewise.
18182 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
18183 (expand_vec_perm_1): Likewise.
18184 (expand_vec_perm_interleave3): Likewise.
18185 (expand_vec_perm_even_odd_pack): Likewise.
18186 (expand_vec_perm_even_odd_1): Likewise.
18187 (expand_vec_perm_broadcast_1): Likewise.
18188 (ix86_vectorize_vec_perm_const_ok): Likewise.
18189 (ix86_expand_vecop_qihi): Likewise.
18190 (ix86_expand_mul_widen_hilo): Likewise.
18191 (ix86_expand_sse2_abs): Likewise.
18192 (ix86_expand_pextr): Likewise.
18193 (ix86_expand_pinsr): Likewise.
18194 (ix86_preferred_simd_mode): Likewise.
18195 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
18196 * config/i386/sse.md (*andnot<mode>3): Likewise.
18197 (<mask_codefor><code><mode>3<mask_name>): Likewise.
18198 (*<code><mode>3): Likewise.
18199 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
18200 (ia64_expand_atomic_op): Likewise.
18201 (ia64_arg_type): Likewise.
18202 (ia64_mode_to_int): Likewise.
18203 (ia64_scalar_mode_supported_p): Likewise.
18204 (ia64_vector_mode_supported_p): Likewise.
18205 (expand_vec_perm_broadcast): Likewise.
18206 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
18207 (iq2000_function_arg_advance): Likewise.
18208 (iq2000_function_arg): Likewise.
18209 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
18210 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
18211 (m68k_libcall_value): Likewise.
18212 (m68k_function_value): Likewise.
18213 (sched_attr_op_type): Likewise.
18214 * config/mcore/mcore.c (mcore_output_move): Likewise.
18215 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
18217 (microblaze_function_arg): Likewise.
18218 * config/mips/mips.c (mips16_build_call_stub): Likewise.
18219 (mips_print_operand): Likewise.
18220 (mips_mode_ok_for_mov_fmt_p): Likewise.
18221 (mips_vector_mode_supported_p): Likewise.
18222 (mips_preferred_simd_mode): Likewise.
18223 (mips_expand_vpc_loongson_even_odd): Likewise.
18224 (mips_expand_vec_unpack): Likewise.
18225 (mips_expand_vi_broadcast): Likewise.
18226 (mips_expand_vector_init): Likewise.
18227 (mips_expand_vec_reduc): Likewise.
18228 (mips_expand_msa_cmp): Likewise.
18229 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
18230 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
18231 (cc_flags_for_mode): Likewise.
18232 * config/msp430/msp430.c (msp430_print_operand): Likewise.
18233 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
18234 (nds32_output_casesi_pc_relative): Likewise.
18235 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18236 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
18237 (nvptx_gen_unpack): Likewise.
18238 (nvptx_gen_pack): Likewise.
18239 (nvptx_gen_shuffle): Likewise.
18240 (nvptx_gen_wcast): Likewise.
18241 (nvptx_preferred_simd_mode): Likewise.
18242 * config/pa/pa.c (pa_secondary_reload): Likewise.
18243 * config/pa/predicates.md (base14_operand): Likewise.
18244 * config/powerpcspe/powerpcspe-c.c
18245 (altivec_resolve_overloaded_builtin): Likewise.
18246 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
18248 (rs6000_preferred_simd_mode): Likewise.
18249 (output_vec_const_move): Likewise.
18250 (rs6000_expand_vector_extract): Likewise.
18251 (rs6000_split_vec_extract_var): Likewise.
18252 (reg_offset_addressing_ok_p): Likewise.
18253 (rs6000_legitimate_offset_address_p): Likewise.
18254 (rs6000_legitimize_address): Likewise.
18255 (rs6000_emit_set_const): Likewise.
18256 (rs6000_const_vec): Likewise.
18257 (rs6000_emit_move): Likewise.
18258 (spe_build_register_parallel): Likewise.
18259 (rs6000_darwin64_record_arg_recurse): Likewise.
18260 (swap_selector_for_mode): Likewise.
18261 (spe_init_builtins): Likewise.
18262 (paired_init_builtins): Likewise.
18263 (altivec_init_builtins): Likewise.
18264 (do_load_for_compare): Likewise.
18265 (rs6000_generate_compare): Likewise.
18266 (rs6000_expand_float128_convert): Likewise.
18267 (emit_load_locked): Likewise.
18268 (emit_store_conditional): Likewise.
18269 (rs6000_output_function_epilogue): Likewise.
18270 (rs6000_handle_altivec_attribute): Likewise.
18271 (rs6000_function_value): Likewise.
18272 (emit_fusion_gpr_load): Likewise.
18273 (emit_fusion_p9_load): Likewise.
18274 (emit_fusion_p9_store): Likewise.
18275 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
18276 (fusion_gpr_mem_load): Likewise.
18277 (fusion_addis_mem_combo_load): Likewise.
18278 (fusion_addis_mem_combo_store): Likewise.
18279 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
18280 (fusion_gpr_mem_load): Likewise.
18281 (fusion_addis_mem_combo_load): Likewise.
18282 (fusion_addis_mem_combo_store): Likewise.
18283 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18285 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
18286 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
18287 (rs6000_preferred_simd_mode): Likewise.
18288 (output_vec_const_move): Likewise.
18289 (rs6000_expand_vector_extract): Likewise.
18290 (rs6000_split_vec_extract_var): Likewise.
18291 (reg_offset_addressing_ok_p): Likewise.
18292 (rs6000_legitimate_offset_address_p): Likewise.
18293 (rs6000_legitimize_address): Likewise.
18294 (rs6000_emit_set_const): Likewise.
18295 (rs6000_const_vec): Likewise.
18296 (rs6000_emit_move): Likewise.
18297 (rs6000_darwin64_record_arg_recurse): Likewise.
18298 (swap_selector_for_mode): Likewise.
18299 (paired_init_builtins): Likewise.
18300 (altivec_init_builtins): Likewise.
18301 (rs6000_expand_float128_convert): Likewise.
18302 (emit_load_locked): Likewise.
18303 (emit_store_conditional): Likewise.
18304 (rs6000_output_function_epilogue): Likewise.
18305 (rs6000_handle_altivec_attribute): Likewise.
18306 (rs6000_function_value): Likewise.
18307 (emit_fusion_gpr_load): Likewise.
18308 (emit_fusion_p9_load): Likewise.
18309 (emit_fusion_p9_store): Likewise.
18310 * config/rx/rx.c (rx_gen_move_template): Likewise.
18311 (flags_from_mode): Likewise.
18312 * config/s390/predicates.md (s390_alc_comparison): Likewise.
18313 (s390_slb_comparison): Likewise.
18314 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
18315 (s390_vector_mode_supported_p): Likewise.
18316 (s390_cc_modes_compatible): Likewise.
18317 (s390_match_ccmode_set): Likewise.
18318 (s390_canonicalize_comparison): Likewise.
18319 (s390_emit_compare_and_swap): Likewise.
18320 (s390_branch_condition_mask): Likewise.
18321 (s390_rtx_costs): Likewise.
18322 (s390_secondary_reload): Likewise.
18323 (__SECONDARY_RELOAD_CASE): Likewise.
18324 (s390_expand_cs): Likewise.
18325 (s390_preferred_simd_mode): Likewise.
18326 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
18327 * config/sh/sh.c (sh_print_operand): Likewise.
18328 (dump_table): Likewise.
18329 (sh_secondary_reload): Likewise.
18330 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18331 * config/sh/sh.md (casesi_worker_1): Likewise.
18332 (casesi_worker_2): Likewise.
18333 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
18334 (fcc_comparison_operator): Likewise.
18335 * config/sparc/sparc.c (sparc_expand_move): Likewise.
18336 (emit_soft_tfmode_cvt): Likewise.
18337 (sparc_preferred_simd_mode): Likewise.
18338 (output_cbranch): Likewise.
18339 (sparc_print_operand): Likewise.
18340 (sparc_expand_vec_perm_bmask): Likewise.
18341 (vector_init_bshuffle): Likewise.
18342 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
18343 (spu_vector_mode_supported_p): Likewise.
18344 (spu_expand_insv): Likewise.
18345 (spu_emit_branch_or_set): Likewise.
18346 (spu_handle_vector_attribute): Likewise.
18347 (spu_builtin_splats): Likewise.
18348 (spu_builtin_extract): Likewise.
18349 (spu_builtin_promote): Likewise.
18350 (spu_expand_sign_extend): Likewise.
18351 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
18352 (tilegx_simd_int): Likewise.
18353 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
18354 (tilepro_simd_int): Likewise.
18355 * config/v850/v850.c (const_double_split): Likewise.
18356 (v850_print_operand): Likewise.
18357 (ep_memory_offset): Likewise.
18358 * config/vax/vax.c (vax_rtx_costs): Likewise.
18359 (vax_output_int_move): Likewise.
18360 (vax_output_int_add): Likewise.
18361 (vax_output_int_subtract): Likewise.
18362 * config/visium/predicates.md (visium_branch_operator): Likewise.
18363 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
18364 (visium_print_operand_address): Likewise.
18365 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18366 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
18367 (xtensa_expand_conditional_branch): Likewise.
18368 (xtensa_copy_incoming_a7): Likewise.
18369 (xtensa_output_literal): Likewise.
18370 * dfp.c (decimal_real_maxval): Likewise.
18371 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
18373 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18374 Alan Hayward <alan.hayward@arm.com>
18375 David Sherwood <david.sherwood@arm.com>
18377 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
18378 (mode_nunits_inline): Likewise.
18379 (mode_inner_inline): Likewise.
18380 (mode_unit_size_inline): Likewise.
18381 (mode_unit_precision_inline): Likewise.
18382 (emit_insn_modes_h): Likewise. Also emit a #define of the
18384 (emit_mode_wider): Add an E_ prefix to mode names.
18385 (emit_mode_complex): Likewise.
18386 (emit_mode_inner): Likewise.
18387 (emit_mode_adjustments): Likewise.
18388 (emit_mode_int_n): Likewise.
18389 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
18390 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
18391 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
18392 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
18393 (CRC32_BUILTIN, ENTRY): Likewise.
18394 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
18395 (aarch64_pop_regs): Likewise.
18396 (aarch64_process_components): Likewise.
18397 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
18398 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
18399 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
18400 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
18401 * config/arm/arm.c (arm_init_libfuncs): Likewise.
18402 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
18404 * config/i386/i386-builtin.def (pcmpestr): Likewise.
18405 (pcmpistr): Likewise.
18406 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
18407 * config/mmix/mmix.c (mmix_output_condition): Likewise.
18408 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
18410 * config/rl78/rl78.c (mduc_regs): Likewise.
18411 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
18412 (htm_expand_builtin): Likewise.
18413 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
18414 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
18415 * config/xtensa/xtensa.c (print_operand): Likewise.
18416 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
18417 (NUM_MODE_VECTOR_INT): Likewise.
18418 * genoutput.c (null_operand): Likewise.
18419 (output_operand_data): Likewise.
18420 * genrecog.c (print_parameter_value): Likewise.
18421 * lra.c (debug_operand_data): Likewise.
18423 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18425 * dwarf2out.c (scompare_loc_descriptor_wide)
18426 (scompare_loc_descriptor_narrow): New functions, split out from...
18427 (scompare_loc_descriptor): ...here.
18428 * expmed.c (emit_store_flag_int): New function, split out from...
18429 (emit_store_flag): ...here.
18431 2017-08-30 Richard Biener <rguenther@suse.de>
18433 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
18434 (dwarf2out_early_finish): Move setting of AT_pubnames from
18435 early debug output to early finish.
18437 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
18439 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
18440 and -mdata-region to the assembler.
18442 2017-08-30 Richard Biener <rguenther@suse.de>
18444 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
18446 (gen_subprogram_die): Add DW_AT_object_pointer only early.
18447 (dwarf2out_early_global_decl): Only generate a DIE for the
18448 abstract origin if it doesn't already exist or is a declaration DIE.
18449 (resolve_addr): Do not add the linkage name twice when
18450 generating a stub DIE for the DW_TAG_GNU_call_site target.
18452 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18454 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18455 Use machine_mode rather than int for arg1_mode.
18457 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
18460 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
18461 that the second argument of the built-in functions to unpack
18462 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
18463 switch statement instead a lot of if statements.
18464 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
18465 Allow 64-bit values to be in Altivec registers as well as
18466 traditional floating point registers.
18467 (pack<mode>, FMOVE128_VSX iterator): Likewise.
18469 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
18471 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
18472 MAX_REGS_PER_ADDRESS == 1.
18474 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
18476 * config/i386/i386.opt (flag_fentry): Do not init to -1.
18477 * config/i386/i386.c (ix86_option_override_internal): Simplify
18478 setting of opts->x_flag_entry.
18480 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18481 Jakub Jelinek <jakub@redhat.com>
18482 Richard Biener <rguenther@suse.de>
18484 PR tree-optimization/81503
18485 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
18486 folded constant fits in the target type; reorder tests for clarity.
18488 2017-08-29 Martin Liska <mliska@suse.cz>
18490 * passes.def: Include pass_lower_switch.
18491 * stmt.c (dump_case_nodes): Remove and move to
18492 tree-switch-conversion.
18493 (case_values_threshold): Likewise.
18494 (expand_switch_as_decision_tree_p): Likewise.
18495 (emit_case_decision_tree): Likewise.
18496 (expand_case): Likewise.
18497 (balance_case_nodes): Likewise.
18498 (node_has_low_bound): Likewise.
18499 (node_has_high_bound): Likewise.
18500 (node_is_bounded): Likewise.
18501 (emit_case_nodes): Likewise.
18502 (struct simple_case_node): New struct.
18503 (add_case_node): Remove.
18504 (emit_case_dispatch_table): Use vector instead of case_list.
18505 (reset_out_edges_aux): Remove.
18506 (compute_cases_per_edge): Likewise.
18507 (expand_case): Build list of simple_case_node.
18508 (expand_sjlj_dispatch_table): Use it.
18509 * tree-switch-conversion.c (struct case_node): Moved from
18510 stmt.c and adjusted.
18511 (emit_case_nodes): Likewise.
18512 (node_has_low_bound): Likewise.
18513 (node_has_high_bound): Likewise.
18514 (node_is_bounded): Likewise.
18515 (case_values_threshold): Likewise.
18516 (reset_out_edges_aux): Likewise.
18517 (compute_cases_per_edge): Likewise.
18518 (add_case_node): Likewise.
18519 (dump_case_nodes): Likewise.
18520 (balance_case_nodes): Likewise.
18521 (expand_switch_as_decision_tree_p): Likewise.
18522 (emit_jump): Likewise.
18523 (emit_case_decision_tree): Likewise.
18524 (try_switch_expansion): Likewise.
18525 (do_jump_if_equal): Likewise.
18526 (emit_cmp_and_jump_insns): Likewise.
18527 (fix_phi_operands_for_edge): New function.
18528 (record_phi_operand_mapping): Likewise.
18529 (class pass_lower_switch): New pass.
18530 (pass_lower_switch::execute): New function.
18531 (make_pass_lower_switch): Likewise.
18532 (conditional_probability):
18533 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
18534 * tree-pass.h: Add make_pass_lower_switch.
18536 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
18539 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
18542 2017-08-29 Richard Biener <rguenther@suse.de>
18544 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
18545 we do not add a DW_AT_inline attribute twice.
18546 (gen_subprogram_die): Remove code setting DW_AT_inline on
18547 DECL_ABSTRACT_P nodes.
18549 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
18551 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
18552 calls to internal functions.
18553 (gimplify_modify_expr): Likewise.
18554 * tree-call-cdce.c (use_internal_fn): Likewise.
18555 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
18556 (convert_to_divmod): Set the nothrow flag.
18557 * tree-if-conv.c (predicate_mem_writes): Likewise.
18558 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
18559 (vectorizable_call): Likewise.
18560 (vectorizable_store): Likewise.
18561 (vectorizable_load): Likewise.
18562 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
18563 (vect_recog_mask_conversion_pattern): Likewise.
18565 2017-08-29 Martin Liska <mliska@suse.cz>
18568 * gcc.c (driver_handle_option): Add new argument.
18569 * opts-common.c (handle_option): Pass
18570 target_option_override_hook.
18571 * opts-global.c (lang_handle_option): Add new option.
18572 (set_default_handlers): Add new argument.
18573 (decode_options): Likewise.
18574 * opts.c (target_handle_option): Likewise.
18575 (common_handle_option): Call target_option_override_hook.
18576 * opts.h (struct cl_option_handler_func): Add hook for
18577 target option override.
18578 (struct cl_option_handlers): Likewise.
18579 (set_default_handlers): Add new argument.
18580 (decode_options): Likewise.
18581 (common_handle_option): Likewise.
18582 (target_handle_option): Likewise.
18583 * toplev.c (toplev::main): Pass targetm.target_option.override
18586 2017-08-29 Richard Biener <rguenther@suse.de>
18587 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
18589 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
18590 life to the active subtree.
18592 2017-08-28 Jeff Law <law@redhat.com>
18594 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
18595 derive_equivalences.
18596 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
18597 Code moved into....
18598 (edge_info::derive_equivalences): New private member function
18600 * tree-ssa-dom.c (class edge_info): Changed from a struct
18601 to a class. Add ctor/dtor, methods and data members.
18602 (edge_info::edge_info): Renamed from allocate_edge_info.
18603 Initialize additional members.
18604 (edge_info::~edge_info): New.
18605 (free_dom_edge_info): Delete the edge info.
18606 (record_edge_info): Use new class & associated member functions.
18607 Tighten forms for testing for edge equivalences.
18608 (record_temporary_equivalences): Iterate over the simple
18609 equivalences rather than assuming there's only one per edge.
18610 (cprop_into_successor_phis): Iterate over the simple
18611 equivalences rather than assuming there's only one per edge.
18612 (optimize_stmt): Use operand_equal_p rather than pointer
18613 equality for mini-DSE code.
18615 2017-08-28 Nathan Sidwell <nathan@acm.org>
18617 * gcc.c (execute): Fold SIGPIPE handling into switch
18618 statement. Adjust internal error message.
18620 2017-08-28 Richard Biener <rguenther@suse.de>
18623 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
18624 Do nothing for removed DIEs.
18626 2017-08-28 Richard Biener <rguenther@suse.de>
18628 PR tree-optimization/81977
18629 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
18632 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
18635 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
18636 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
18637 using targetm.gen_mem_thread_fence.
18639 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
18642 * config/i386/i386.md (*<btsc><mode>): Change operand 2
18643 predicate to register_operand. Reorder operands.
18644 (*btr<mode>): Ditto.
18645 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
18646 (*btr<mode>_mask): Ditto.
18648 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
18650 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
18651 * config/rs6000/xmmintrin.h: New file.
18652 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
18654 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18657 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
18658 parameter and_insn and return it.
18659 (recombine_lvx_pattern): Insert a copy to ensure availability of
18660 the base register of the copied masking operation at the point of
18661 the instruction replacement.
18662 (recombine_stvx_pattern): Likewise.
18664 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
18666 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
18667 undocumented switches.
18668 (-mpower9-dform-vector): Likewise.
18669 (-mpower9-dform): Likewise.
18670 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
18671 comments to delete references to -mpower9-dform* switches.
18672 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
18673 Delete reference to -mpower9-dform* switches, test for
18674 -mpower9-vector instead.
18675 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
18676 (OTHER_P9_VECTOR_MASKS): Likewise.
18677 (POWERPC_MASKS): Likewise.
18678 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
18679 tests against -mpower9-dform* to -mpower9-vector. Delete code
18680 that checked for -mpower9-dform* consistancy with other options.
18681 Add test for -mpower9-misc to enable other power9 switches.
18682 (rs6000_init_hard_regno_mode_ok): Likewise.
18683 (rs6000_option_override_internal): Likewise.
18684 (rs6000_emit_prologue): Likewise.
18685 (rs6000_emit_epilogue): Likewise.
18686 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
18687 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
18688 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
18690 (emit_fusion_p9_store): Likewise.
18691 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
18692 resetting these macros if the assembler does not support ISA 3.0
18694 (TARGET_P9_DFORM_VECTOR): Likewise.
18695 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
18696 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
18698 2017-08-25 Alan Modra <amodra@gmail.com>
18700 PR rtl-optimization/81747
18701 * cse.c (cse_extended_basic_block): Don't attempt to record
18702 equivalences for degenerate conditional branches that branch
18703 to their fall-through.
18705 2017-08-24 Martin Sebor <msebor@redhat.com>
18707 PR middle-end/81908
18708 * gimple-fold.c (size_must_be_zero_p): New function.
18709 (gimple_fold_builtin_memory_op): Call it.
18711 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
18713 * config/rs6000/mm_malloc.h: New file.
18715 2017-08-24 Bin Cheng <bin.cheng@arm.com>
18717 PR tree-optimization/81913
18718 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
18719 analysis when either IVs in condition can wrap.
18721 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
18723 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
18724 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
18726 2017-08-24 Richard Biener <rguenther@suse.de>
18729 * targhooks.c (default_target_can_inline_p): Properly
18730 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
18731 is present and always compare.
18732 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
18733 infer -mfpmath=sse from TARGET_SSE_P.
18734 (ix86_can_inline_p): Properly use target_option_default_node when
18735 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
18737 2017-08-24 Richard Biener <rguenther@suse.de>
18740 * dwarf2out.c (output_die): Handle flag_generate_offload like
18742 (output_comp_unit): Likewise.
18743 (gen_array_type_die): Likewise.
18744 (dwarf2out_early_finish): Likewise.
18745 (note_variable_value_in_expr): Likewise.
18746 (dwarf2out_finish): Likewise. Adjust assert.
18747 * cgraphunit.c (symbol_table::compile): Move setting of
18748 flag_generate_offload earlier ...
18749 (symbol_table::finalize_compilation_unit): ... here, before
18750 early debug finalization.
18752 2017-08-24 Richard Biener <rguenther@suse.de>
18754 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
18755 and ipa-fnsummary.h.
18756 (ix86_can_inline_p): When ix86_fpmath flags do not match
18757 check whether the callee uses FP math at all.
18759 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
18761 PR middle-end/81931
18762 * tree-ssanames.c (get_nonzero_bits): Use element_precision
18763 instead of TYPE_PRECISION.
18765 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
18766 Alan Hayward <alan.hayward@arm.com>
18767 David Sherwood <david.sherwood@arm.com>
18769 * combine.c (make_extraction): Use subreg_offset_from_lsb.
18771 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
18773 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
18775 (ix86_frame::stack_realign_allocate): New field.
18776 (struct machine_frame_state): Modify comments.
18777 (machine_frame_state::sp_realigned_fp_end): New field.
18778 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
18779 layout calculation.
18780 (sp_valid_at): Add assertion to assure no attempt to access invalid
18781 offset of a realigned stack.
18782 (fp_valid_at): Likewise.
18783 (choose_baseaddr): Modify comments.
18784 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
18785 ix86_expand_prologue.
18786 (ix86_expand_prologue): Modify stack realignment and allocation.
18787 (ix86_expand_epilogue): Modify comments.
18788 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
18789 avx2_runtime, avx512f, and avx512f_runtime.
18791 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
18793 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
18794 (mstackrealign): Do not init to -1.
18795 * config/i386/i386.c (ix86_option_override_internal):
18796 Check opts_set, not opts when setting default value of
18797 opts->x_ix86_force_align_arg_pointer.
18799 2017-08-23 Richard Biener <rguenther@suse.de>
18801 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
18802 lang_hooks.decl_printable_name.
18803 * print-rtl-function.c (print_rtx_function): Likewise.
18804 * tree-pretty-print.c (dump_function_header): Likewise.
18806 2017-08-23 Richard Biener <rguenther@suse.de>
18809 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
18810 -g0 at compile-time.
18812 2017-08-23 Tamar Christina <tamar.christina@arm.com>
18814 PR middle-end/19706
18815 * doc/sourcebuild.texi (Other hardware attributes):
18818 2017-08-23 Tamar Christina <tamar.christina@arm.com>
18820 PR middle-end/19706
18821 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
18822 Fix single-use check.
18824 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
18826 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
18828 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
18830 * doc/install.texi: Modify to add more details on running selected
18833 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
18835 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
18836 is combined with -mabi=ms.
18837 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
18840 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18842 PR tree-optimization/81488
18843 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
18844 and cached_basis fields.
18845 (MAX_SPREAD): New constant.
18846 (alloc_cand_and_find_basis): Initialize new fields.
18847 (clear_visited): New function.
18848 (create_phi_basis_1): Rename from create_phi_basis, set visited
18849 and cached_basis fields.
18850 (create_phi_basis): New wrapper function.
18851 (phi_add_costs_1): Rename from phi_add_costs, add spread
18852 parameter, set visited field, short-circuit when limits reached.
18853 (phi_add_costs): New wrapper function.
18854 (record_phi_increments_1): Rename from record_phi_increments, set
18856 (record_phi_increments): New wrapper function.
18857 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
18858 (phi_incr_cost): New wrapper function.
18859 (all_phi_incrs_profitable_1): Rename from
18860 all_phi_incrs_profitable, set visited field.
18861 (all_phi_incrs_profitable): New wrapper function.
18863 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
18864 Alan Hayward <alan.hayward@arm.com>
18865 David Sherwood <david.sherwood@arm.com>
18867 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
18868 that takes the outer and inner modes.
18869 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
18870 comparison as the canonical test for a paradoxical subreg.
18871 * combine.c (simplify_set): Use paradoxical_subreg_p.
18872 (make_extraction): Likewise.
18873 (force_to_mode): Likewise.
18874 (rtx_equal_for_field_assignment_p): Likewise.
18875 (gen_lowpart_for_combine): Likewise.
18876 (simplify_comparison): Likewise.
18877 * cse.c (equiv_constant): Likewise.
18878 * expmed.c (store_bit_field_1): Likewise.
18879 * final.c (alter_subreg): Likewise.
18880 * fwprop.c (propagate_rtx): Likewise.
18881 (forward_propagate_subreg): Likewise.
18882 * ira-conflicts.c (ira_build_conflicts): Likewise.
18883 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
18884 * lra-constraints.c (curr_insn_transform): Likewise.
18885 (split_reg): Likewise.
18886 * lra-eliminations.c (move_plus_up): Likewise.
18887 (lra_eliminate_regs_1): Likewise.
18888 * recog.c (general_operand): Likewise.
18889 * ree.c (combine_reaching_defs): Likewise.
18890 * reload.c (push_reload): Likewise.
18891 (find_reloads): Likewise.
18892 * reload1.c (elimination_effects): Likewise.
18893 (compute_reload_subreg_offset): Likewise.
18894 (choose_reload_regs): Likewise.
18895 * rtlanal.c (subreg_lsb_1): Likewise.
18896 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18897 (simplify_subreg): Likewise.
18898 * var-tracking.c (track_loc_p): Likewise.
18899 * emit-rtl.c (byte_lowpart_offset): Likewise.
18900 (paradoxical_subreg_p): Delete out-of-line definition.
18902 2017-08-22 Jeff Law <law@redhat.com>
18904 PR tree-optimization/81741
18905 PR tree-optimization/71947
18906 * tree-ssa-dom.c: Include tree-inline.h.
18907 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
18908 equivalences if one is more expensive to compute than the other.
18909 * tree-ssa-scopedtables.h (class const_or_copies): Make
18910 record_const_or_copy_raw method private.
18911 (class avail_exprs_stack): New method simplify_binary_operation.
18912 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
18913 avail_exprs_stack::simplify_binary_operation as needed.
18914 (avail_exprs_stack::simplify_binary_operation): New function.
18916 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
18918 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
18919 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
18920 (DOT_SYMBOLS): Likewise.
18921 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
18922 (RELOCATABLE_NEEDS_FIXUP): Likewise.
18923 (RS6000_ABI_NAME): Likewise.
18924 (TARGET_CMODEL): Likewise.
18925 (TOC_SECTION_ASM_OP): Likewise.
18926 (SET_CMODEL): New macro.
18927 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
18929 2017-08-22 Richard Biener <rguenther@suse.de>
18931 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
18932 to allow for free-lang-data replacements similar to verify_type_variant.
18934 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
18935 Alan Hayward <alan.hayward@arm.com>
18936 David Sherwood <david.sherwood@arm.com>
18938 * config/aarch64/aarch64.md (casesi): Use DImode rather than
18939 VOIDmode for the LABEL_REF.
18941 2017-08-22 Richard Biener <rguenther@suse.de>
18943 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
18945 2017-08-22 Richard Biener <rguenther@suse.de>
18947 * common.opt (feliminate-dwarf2-dups): Ignore.
18948 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
18949 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
18950 same_die_p_wrap, compute_section_prefix,
18951 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
18952 (comdat_symbol_id, comdat_symbol_number): Likewise.
18953 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
18955 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
18956 (output_die): Mark unreachable path unreachable.
18957 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
18958 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
18959 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
18960 (dwarf2out_early_finish): Likewise.
18962 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
18964 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
18966 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
18969 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
18970 not VAR_P. Filter attribute warnings with OPT_Wattributes.
18971 (avr_attribute_table) <io, io_low, address>: Initialize
18972 .decl_required with true.
18974 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
18976 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
18977 undocumented debugging options.
18978 (-mvsx-scalar-double): Likewise.
18979 (-mallow-df-permute): Likewise.
18980 (-mvectorize-builtins): Likewise.
18981 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
18982 (rs6000_builtin_vectorized_function): Likewise.
18983 (rs6000_builtin_md_vectorized_function): Likewise.
18984 (rs6000_opt_vars): Likewise.
18986 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
18989 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
18990 (*btrq_imm): Rename from *btrq.
18991 (*btcq_imm): Rename from *btcq.
18992 (btsc): New code attribute.
18993 (*<btsc><mode>): New insn pattern.
18994 (*btr<mode>): Ditto.
18995 (*<btsc><mode>_mask): New insn_and_split pattern.
18996 (*btr<mode>_mask): Ditto.
18998 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
18999 Alan Hayward <alan.hayward@arm.com>
19000 David Sherwood <david.sherwood@arm.com>
19002 * function.c (pad_below): Simplify padding calculation.
19004 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19005 Alan Hayward <alan.hayward@arm.com>
19006 David Sherwood <david.sherwood@arm.com>
19008 * target.def (function_prologue): Remove frame size argument.
19009 (function_epilogue): Likewise.
19010 * doc/tm.texi: Regenerate.
19011 * final.c (final_start_function): Update call to function_prologue.
19012 (final_end_function): Update call to function_epilogue.
19013 (default_function_pro_epilogue): Remove frame size argument.
19014 * output.h (default_function_pro_epilogue): Likewise.
19015 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
19016 (arm_output_function_prologue): Likewise.
19017 * config/frv/frv.c (frv_function_prologue): Likewise.
19018 (frv_function_epilogue): Likewise.
19019 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
19020 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
19021 (ia64_output_function_epilogue): Likewise.
19022 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
19023 (m32r_output_function_epilogue): Likewise.
19024 * config/microblaze/microblaze.c (microblaze_function_prologue)
19025 (microblaze_function_epilogue): Likewise.
19026 * config/mips/mips.c (mips_output_function_prologue): Likewise.
19027 (mips_output_function_epilogue): Likewise.
19028 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
19029 (mmix_target_asm_function_epilogue): Likewise.
19030 * config/msp430/msp430.c (msp430_start_function): Likewise.
19031 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
19032 (nds32_asm_function_epilogue): Likewise.
19033 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
19034 * config/pa/pa.c (pa_output_function_prologue): Likewise.
19035 (pa_output_function_epilogue): Likewise.
19036 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
19037 (rs6000_output_function_epilogue): Likewise.
19038 * config/rl78/rl78.c (rl78_start_function): Likewise.
19039 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
19040 (rs6000_output_function_epilogue): Likewise.
19041 * config/rx/rx.c (rx_output_function_prologue): Likewise.
19042 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
19043 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
19044 (sparc_asm_function_epilogue): Likewise.
19046 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19048 * tree.h (type_has_mode_precision_p): New function.
19049 * convert.c (convert_to_integer_1): Use it.
19050 * expr.c (expand_expr_real_2): Likewise.
19051 (expand_expr_real_1): Likewise.
19052 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
19053 * match.pd: Likewise.
19054 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
19055 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
19056 * tree-tailcall.c (process_assignment): Likewise.
19057 * tree-vect-loop.c (vectorizable_reduction): Likewise.
19058 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
19059 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
19060 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
19061 (vectorizable_assignment): Likewise.
19062 (vectorizable_shift): Likewise.
19063 (vectorizable_operation): Likewise.
19064 * tree-vrp.c (register_edge_assert_for_2): Likewise.
19066 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
19068 * match.pd: Add pow (C, x) simplification.
19070 2017-08-21 Richard Biener <rguenther@suse.de>
19072 PR tree-optimization/81900
19073 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
19074 for blocks with abnormal predecessors.
19075 (compute_antic): Do not set visited flag prematurely.
19077 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
19080 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
19082 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19084 * stor-layout.h (vector_type_mode): Move to...
19085 * tree.h (vector_type_mode): ...here.
19086 * stor-layout.c (vector_type_mode): Move to...
19087 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
19089 2017-08-21 Richard Biener <rguenther@suse.de>
19091 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
19092 register_external_die hooks.
19093 (debug_false_tree_charstarstar_uhwistar): Declare.
19094 (debug_nothing_tree_charstar_uhwi): Likewise.
19095 * debug.c (do_nothing_debug_hooks): Adjust.
19096 (debug_false_tree_charstarstar_uhwistar): New do nothing.
19097 (debug_nothing_tree_charstar_uhwi): Likewise.
19098 * dbxout.c (dbx_debug_hooks): Adjust.
19099 (xcoff_debug_hooks): Likewise.
19100 * sdbout.c (sdb_debug_hooks): Likewise.
19101 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
19102 * dwarf2out.c (macinfo_label_base): New global.
19103 (dwarf2out_register_external_die): New function for the
19104 register_external_die hook.
19105 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
19106 (dwarf2_debug_hooks): Use them.
19107 (dwarf2_lineno_debug_hooks): Adjust.
19108 (struct die_struct): Add with_offset flag.
19109 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
19110 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
19111 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
19112 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
19113 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
19114 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
19115 defining section names for the early LTO debug variants.
19116 (reset_indirect_string): New helper.
19117 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
19118 (print_dw_val): Add support for offsetted symbol references.
19119 (get_ultimate_context): Split out from is_cxx.
19120 (is_cxx): Use get_ultimate_context.
19121 (is_fortran): Add decl overload.
19122 (compute_comp_unit_symbol): Split out worker from
19123 compute_section_prefix.
19124 (compute_section_prefix): Call compute_comp_unit_symbol and
19125 set comdat_type_p here.
19126 (output_die): Skip DIE symbol output for the LTO added one.
19127 Handle DIE symbol references with offset.
19128 (output_comp_unit): Guard section name mangling properly.
19129 For LTO debug sections emit a symbol at the section beginning
19130 which we use to refer to its DIEs.
19131 (add_abstract_origin_attribute): For DIEs registered via
19132 dwarf2out_register_external_die directly refer to the early
19133 DIE rather than indirectly through the shadow one we created.
19134 Remove obsolete call to dwarf2out_abstract_function for
19135 non-function/block origins.
19136 (gen_array_type_die): When generating early LTO debug do
19137 not emit DW_AT_string_length.
19138 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
19139 late when in LTO. As suggested place a gcc_unreachable for
19140 the DECL_ABSTRACT_P case.
19141 (gen_subprogram_die): Avoid another specification DIE
19142 for early built declarations/definitions for the late LTO case.
19143 (gen_variable_die): Add type references for late duplicated VLA dies
19145 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
19146 we have the abstract instance already.
19147 (process_scope_var): Adjust decl DIE contexts in LTO which
19148 first puts them in limbo.
19149 (gen_decl_die): Do not generate type DIEs late apart from
19150 types for VLAs or for decls we do not yet have a DIE. Do not
19151 call dwarf2out_abstract_function late.
19152 (dwarf2out_early_global_decl): Make sure to create DIEs
19153 for abstract instances of a decl first.
19154 (dwarf2out_late_global_decl): Adjust comment.
19155 (output_macinfo_op): With multiple macro sections use
19156 macinfo_label_base to distinguish labels.
19157 (output_macinfo): Likewise. Update macinfo_label_base.
19158 Pass in the line info label.
19159 (note_variable_value_in_expr): When generating LTO resolve
19160 all variable values here by generating DIEs as needed.
19161 (init_sections_and_labels): Add early LTO debug flag parameter
19162 and generate different sections and names if set. Add generation
19163 counter for the labels so we can have multiple of them.
19164 (reset_dies): Helper to allow DIEs to be output multiple times.
19165 (dwarf2out_finish): When outputting DIEs to the fat part of an
19166 LTO object first reset DIEs.
19167 (dwarf2out_early_finish): Output early DIEs when generating LTO.
19168 (modified_type_die): Check for decl_ultimate_origin being self
19170 (gen_type_die_with_usage): Likewise.
19171 (gen_typedef_die): Allow decl_ultimate_origin being self.
19172 (set_decl_abstract_flags): Remove.
19173 (set_block_abstract_flags): Likewise.
19174 (dwarf2out_abstract_function): Treat the early generated DIEs
19175 as the abstract copy and only add DW_AT_inline and
19176 DW_AT_artificial here and call set_decl_origin_self.
19177 If the DIE has an abstract origin don't do anything.
19178 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
19179 if we have none yet (Go fails to build one, PR78628).
19180 (variably_modified_type_p): Prevent endless recursion for Ada
19181 cyclic pointer types.
19182 * lto-streamer-in.c: Include debug.h.
19183 (dref_queue): New global.
19184 (lto_read_tree_1): Stream in DIE references.
19185 (lto_input_tree): Register DIE references.
19186 (input_function): Stream DECL_DEBUG_ARGS.
19187 * lto-streamer-out.c: Include debug.h.
19188 (lto_write_tree_1): Output DIE references.
19189 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
19190 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
19191 (output_function): Stream DECL_DEBUG_ARGS.
19192 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
19193 Stream DECL_ABSTRACT_ORIGIN.
19194 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
19195 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
19196 DECL_CONTEXT for file-scope decls.
19197 * lto-streamer.h (struct dref_entry): Declare.
19198 (dref_queue): Likewise.
19199 * cfgexpand.c (pass_expand::execute): Do not call the
19200 outlining_inline_function hook here.
19201 * lto-wrapper.c (debug_obj): New global.
19202 (tool_cleanup): Unlink it if required.
19203 (debug_objcopy): New function.
19204 (run_gcc): Handle early debug sections in the IL files by
19205 extracting them to separate files, partially linkin them and
19206 feeding the result back as result to the linker.
19207 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
19208 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
19209 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
19210 sections into a separate segment.
19211 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
19213 (darwin_asm_dwarf_section): Likewise.
19214 (darwin_asm_output_dwarf_offset): Likewise.
19215 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
19217 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19218 Alan Hayward <alan.hayward@arm.com>
19219 David Sherwood <david.sherwood@arm.com>
19221 * read-md.h (md_reader::record_potential_iterator_use): Replace
19222 pointer argument with an rtx and an index.
19223 * read-rtl.c (iterator_group::apply_iterator): Likewise.
19224 (apply_mode_iterator): Likewise.
19225 (apply_code_iterator): Likewise.
19226 (apply_int_iterator): Likewise.
19227 (apply_subst_iterator): Likewise.
19228 (record_iterator_use): Likewise.
19229 (record_attribute_use): Likewise.
19230 (md_reader::record_potential_iterator_use): Likewise. Update calls
19231 to record_iterator_use and apply_iterator.
19232 (iterator_use): Replace ptr with x and index.
19233 (attribute_use): Likewise.
19234 (apply_attribute_uses): Update calls to apply_iterator.
19235 (apply_iterators): Likewise. Update initialization of iterator_use.
19236 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
19237 and record_potential_iterator_use.
19238 (rtx_reader::read_rtx_operand): Likewise.
19240 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19241 Alan Hayward <alan.hayward@arm.com>
19242 David Sherwood <david.sherwood@arm.com>
19244 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
19247 2017-08-21 Richard Biener <rguenther@suse.de>
19249 PR middle-end/81884
19250 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
19251 at struct end conservatively when comparing common bases.
19253 2017-08-21 Richard Biener <rguenther@suse.de>
19255 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
19256 (mem_ref_in_stmt): Remove.
19257 (determine_max_movement): Use ref index to get at the reference.
19258 (invariantness_dom_walker::before_dom_children): Deal with
19259 lim data already initialized.
19260 (gather_mem_refs_stmt): Initialize lim data and record ref index.
19262 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
19264 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
19265 (TARGET_ISA_ROUND): Ditto.
19266 (TARGET_ROUND): Ditto.
19267 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
19268 * config/i386/i386.md: Ditto.
19269 * config/i386/sse.md: Ditto.
19270 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
19271 with OPTION_MASK_ISA_SSE4_1.
19273 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
19276 * doc/extend.texi (x86 Built-in Functions): Correct the name of
19277 __builtin_ia32_lzcnt_u16.
19279 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
19282 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
19283 (rs6000_set_current_function): Rewrite function to use it.
19285 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
19288 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
19289 and TYPE_WARN_IF_NOT_ALIGN.
19290 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
19291 (handle_warn_if_not_align): New.
19292 (place_union_field): Call handle_warn_if_not_align.
19293 (place_field): Call handle_warn_if_not_align.
19294 Copy TYPE_WARN_IF_NOT_ALIGN.
19295 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
19296 (layout_type): Likewise.
19297 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
19299 (tree_decl_common): Add warn_if_not_align.
19300 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
19301 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
19302 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
19303 (DECL_WARN_IF_NOT_ALIGN): Likewise.
19304 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
19305 * doc/extend.texi: Document warn_if_not_aligned attribute.
19306 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
19308 2017-08-17 Martin Liska <mliska@suse.cz>
19311 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
19312 (get_data_dependence): Use it as pointer type.
19313 (distribute_loop): Likewise.
19315 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19317 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
19318 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
19319 (p8_vmrgow_<mode>_direct): New define_insn.
19320 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
19321 handle endianness for vmrgew and vmrgow permute patterns.
19323 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
19325 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
19326 * config/rs6000/rs6000-cpus.def: Remove comment.
19327 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
19328 (POWERPC_MASKS): Likewise.
19329 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
19330 use of TARGET_VSX_TIMODE.
19331 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
19332 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
19333 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
19334 (rs6000_option_override_internal): Remove dead code.
19335 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
19336 (rs6000_legitimize_reload_address): Likewise.
19337 (rs6000_legitimate_address_p): Likewise.
19338 (rs6000_opt_masks): Delete "vsx-timode".
19339 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
19340 from function comment.
19341 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
19342 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
19343 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
19345 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
19346 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
19347 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
19348 (VSX_TI): Likewise.
19350 (define_peephole2): Likewise.
19352 2017-08-17 Martin Sebor <msebor@redhat.com>
19355 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
19356 past the end of an array.
19357 (test_pp_format): Add test cases.
19359 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
19361 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
19362 missing ECF_NOTHROW flags.
19364 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
19367 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
19368 operands residing in integer registers.
19369 (*vsx_le_perm_load_<mode>): Likewise.
19370 (*vsx_le_perm_store_<mode>): Likewise.
19371 (define_peephole2): Add peepholes to optimize the above.
19373 2017-08-17 Marek Polacek <polacek@redhat.com>
19375 PR middle-end/81814
19376 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
19377 to mimic what shorten_compare did. Change the return type to bool.
19378 (fold_cond_expr_with_comparison): Update call to
19379 operand_equal_for_comparison_p.
19380 (fold_ternary_loc): Likewise.
19382 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
19384 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
19386 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
19388 2017-08-17 Richard Biener <rguenther@suse.de>
19390 * tree-ssa-structalias.c (solve_graph): When propagating
19391 to successors update the graphs succ edges and avoid duplicate work.
19393 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
19396 * config/i386/i386.c (ix86_option_override_internal): Save target
19397 specific options after ix86_stack_protector_guard_reg was changed.
19399 2017-08-17 Richard Biener <rguenther@suse.de>
19401 PR tree-optimization/81827
19402 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
19403 (new_var_info): Initialize it conservatively.
19404 (get_call_vi): Mark register vars.
19405 (new_scalar_tmp_constraint_exp): Likewise.
19406 (handle_rhs_call): Likewise.
19407 (handle_const_call): Likewise.
19408 (create_function_info_for): Likewise.
19409 (solve_constraints): Sort varinfos to separate register from
19410 non-register vars to pack points-to solution bitmaps during
19413 2017-08-17 Marek Polacek <polacek@redhat.com>
19415 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
19417 2017-08-17 Richard Biener <rguenther@suse.de>
19419 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
19420 to true when overflow is undefined and we saturated the result.
19422 2017-08-17 Alan Modra <amodra@gmail.com>
19425 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
19426 Don't use store multiple if only one reg needs saving.
19427 (interesting_frame_related_regno): New function.
19428 (rs6000_frame_related): Don't emit frame info for regs that
19430 (rs6000_emit_epilogue): Likewise.
19432 2017-08-16 Nathan Sidwell <nathan@acm.org>
19434 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
19435 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
19436 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
19437 * tree.c (free_lang_data_in_type): Use else-if chain. Always
19438 clear TYPE_LANG_1. Remove obsolete member-function stripping.
19439 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
19440 (verify_type): Adjust for TYPE_BINFO move.
19441 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
19442 process TYPE_BINFO directly.
19443 (hash_tree): Likewise.
19444 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
19446 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
19449 2017-08-16 David Malcolm <dmalcolm@redhat.com>
19451 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
19453 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
19456 * config/i386/i386.md (*anddi_1_btr): Change predicates of
19457 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
19458 Add ix86_binary_operator_ok to insn constraint.
19459 (*iordi_1_bts): Ditto.
19460 (*xordi_1_btc): Ditto.
19461 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
19462 Update corresponding peephole2 pattern.
19466 2017-08-16 Bin Cheng <bin.cheng@arm.com>
19468 PR tree-optimization/81832
19469 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
19470 copy loop header which has IFN_LOOP_DIST_ALIAS call.
19472 2017-08-16 Marek Polacek <polacek@redhat.com>
19475 * fold-const.c (fold_indirect_ref_1): Restore original behavior
19476 regarding size_zero_node.
19478 2017-08-16 Martin Liska <mliska@suse.cz>
19481 * config.gcc: Respect previously set extra_objs in case
19484 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
19486 PR tree-optimization/81835
19487 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
19488 the phi SSA_NAME. Check that the condition in a COND_EXPR does
19489 not depend on the phi.
19491 2017-08-16 Alan Modra <amodra@gmail.com>
19493 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
19496 2017-08-16 Alan Modra <amodra@gmail.com>
19498 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
19499 (save_reg_p): ..into this. Update all callers.
19500 (first_reg_to_save): Simplify.
19502 2017-08-16 Alan Modra <amodra@gmail.com>
19504 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
19507 2017-08-15 Joseph Myers <joseph@codesourcery.com>
19511 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
19512 constant count if that count is less than 32.
19514 2017-08-15 Nathan Sidwell <nathan@acm.org>
19516 * gcc.c (execute): Emit friendlier message if inferior is killed
19517 by an external cause.
19519 2017-08-15 Richard Biener <rguenther@suse.de>
19521 PR tree-optimization/81790
19522 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
19523 CONSTRUCTORs from simplifying and VN.
19525 2017-08-14 Martin Sebor <msebor@redhat.com>
19527 * builtin-attrs.def: Add comments.
19529 2017-08-14 Martin Sebor <msebor@redhat.com>
19532 * doc/extend.texi (attribute nonstring): Document new attribute.
19534 2017-08-14 Martin Sebor <msebor@redhat.com>
19537 * tree-diagnostic.c (default_tree_printer): Handle %G.
19538 * gimple-pretty-print.h (percent_G_format): Declare new function.
19539 * gimple-pretty-print.c (percent_G_format): Define.
19540 * tree-pretty-print.c (percent_K_format): Add argument.
19542 2017-08-14 Martin Sebor <msebor@redhat.com>
19544 PR translation/79998
19545 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
19546 Remove a stray space.
19548 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
19551 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
19552 (*iordi_1_bts): Ditto.
19553 (*xordi_1_btc): Ditto.
19555 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19558 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
19559 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19561 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
19562 quoted strings, and make more translator-friendly.
19563 (darwin_rs6000_override_options): Likewise.
19564 (rs6000_option_override_internal): Likewise.
19565 (rs6000_return_in_memory): Fix overlong line.
19566 (init_cmulative_args): Use quoted strings, and make more
19567 translator-friendly.
19568 (rs6000_pass_by_reference): Fix overlong line.
19569 (def_builtin): Use quoted strings.
19570 (altivec_expand_predicate_builtin): Use quoted strings, and make
19571 more translator-friendly.
19572 (htm_expand_builtin): Use quoted strings.
19573 (cpu_expand_builtin): Use quoted strings, and make more
19574 translator-friendly.
19575 (altivec_expand_builtin): Likewise.
19576 (paired_expand_predicate_builtin): Likewise.
19577 (rs6000_invalid_builtin): Likewise.
19578 (builtin_function_type): Use quoted strings.
19579 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
19580 more translator-friendly.
19581 (rs6000_trampoline_init): Likewise.
19582 (rs6000_handle_altivec_attribute): Likewise.
19583 (rs6000_inner_target_options): Use quoted strings.
19584 (rs6000_disable_incompatible_switches): Likewise.
19585 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
19586 strings, and make more translator-friendly.
19587 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
19589 2017-08-14 Bin Cheng <bin.cheng@arm.com>
19591 PR tree-optimization/81799
19592 * tree-loop-distribution.c (version_loop_by_alias_check): Force
19593 cond_expr to simple gimple operand.
19595 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
19597 PR middle-end/46932
19598 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
19600 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
19604 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
19605 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
19606 TARGET_GASISR_PROLOGUES.
19607 * config/avr/avr.c (avr_option_override): Same.
19608 (avr_pass_pre_proep::execute): Same.
19610 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
19613 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
19614 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
19616 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
19618 * config/i386/i386.md (*load_tp_<mode>): Redefine as
19619 define_insn_and_split. Split to a memory load from 0 in
19620 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
19621 using PTR mode iterator.
19622 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
19623 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
19624 (*add_tp_<mode>): Redefine as define_insn_and_split.
19625 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
19626 address space. Merge with *add_tp_x32 using PTR mode iterator.
19627 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
19628 Split to an add with a memory load from 0 in
19629 DEFAULT_TLS_SEG_REG address space.
19631 2017-08-12 Andrew Pinski <apinski@cavium.com>
19633 * config/aarch64/aarch64-option-extensions.def (rdma):
19634 Fix feature string to what Linux prints out in /proc/cpuinfo.
19636 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
19639 * dwarf2out.c (modified_type_die): For C typedef types that have
19640 an ultimate origin, process the ultimate origin instead of the
19642 (gen_typedef_die): Assert that input DECLs have no ultimate
19644 (gen_type_die_with_usage): For typedef variants that have an
19645 ultimate origin, just call gen_decl_die on the original DECL.
19646 (process_scope_var): Avoid creating DIEs for local typedefs and
19647 concrete static variables.
19649 2017-08-12 Alan Modra <amodra@gmail.com>
19653 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
19654 match gnu-user.h startfile.
19655 (ENDFILE_LINUX_SPEC): Similarly.
19657 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
19660 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
19662 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
19664 2017-08-11 Tamar Christina <tamar.christina@arm.com>
19665 * config/aarch64/aarch64.md (mov<mode>): Change.
19666 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
19667 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
19668 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
19670 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
19672 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
19673 for storage order barriers.
19675 2017-08-11 Martin Liska <mliska@suse.cz>
19678 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
19679 variables of void type.
19681 2017-08-11 Martin Liska <mliska@suse.cz>
19683 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
19684 TARGET_SUPPORTS_ALIASES.
19685 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
19686 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
19687 (optimize_weakref): Likewise.
19688 * symtab.c (symtab_node::noninterposable_alias): Likewise.
19689 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
19690 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
19692 2017-08-11 Martin Liska <mliska@suse.cz>
19695 * config/i386/i386.c (make_resolver_func): Do complete
19696 refactoring of the function.
19698 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
19701 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
19702 * config/i386/i386.c (ix86_stack_protect_guard): Use
19703 ix86_stack_protect_guard_symbol_str to generate varible declaration.
19704 * doc/invoke.texi (x86 Options): Document
19705 -mstack-protector-guard-symbol= option.
19707 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
19709 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
19710 * config/i386/i386.c (ix86_split_stack_guard): New function.
19711 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
19712 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
19713 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
19714 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
19715 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
19716 (split_stack_space_check): Call ix86_split_stack_guard.
19718 2017-08-10 Martin Sebor <msebor@redhat.com>
19720 * print-tree.c (print_node): Print location using the established
19722 Replace spaces with colons.
19723 (debug_raw, debug): Ditto.
19725 2017-08-10 Martin Sebor <msebor@redhat.com>
19728 * pretty-print.c (pp_format): Correct the handling of %s precision.
19730 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
19733 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
19735 (ix86_finalize_stack_frame_flags): This. Also clear
19736 frame_pointer_needed if -fno-omit-frame-pointer is used without
19738 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
19739 with ix86_finalize_stack_frame_flags.
19740 (ix86_expand_epilogue): Likewise.
19741 (ix86_expand_split_stack_prologue): Likewise.
19742 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
19744 2017-08-10 Martin Liska <mliska@suse.cz>
19747 * c-attribs.c (handle_target_attribute):
19748 Report warning for an empty string argument of target attribute.
19750 2017-08-09 Jakub Jelinek <jakub@redhat.com>
19753 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
19755 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
19756 or DECL_NONLOCAL labels.
19757 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
19758 or DECL_NONLOCAL labels here.
19760 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
19762 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
19763 to indicate when early gimple folding has been disabled.
19764 (rs6000_gimple_fold_builtin): Add debug content.
19765 (rs6000_invalid_builtin): Fix whitespace.
19766 (rs6000_expand_builtin): Fix whitespace.
19767 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
19769 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
19772 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
19773 SAVE_MULTIPLE if not all the registers that saves, should be saved.
19775 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
19777 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
19778 (qdf24xx): Likewise.
19779 * config/aarch64/aarch64.md: Include falkor.md.
19780 * config/aarch64/falkor.md: New.
19782 2017-08-09 Marek Polacek <polacek@redhat.com>
19785 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
19786 * diagnostic.c (emit_diagnostic): Add a comment.
19787 (emit_diagnostic_valist): New function.
19789 2017-08-09 Marek Polacek <polacek@redhat.com>
19792 * input.c (make_location): New overload.
19793 * input.h (make_location): Declare.
19795 2017-08-08 Alan Modra <amodra@gmail.com>
19796 H.J. Lu <hongjiu.lu@intel.com>
19799 * gcc.c (NO_PIE_SPEC): Delete.
19800 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
19802 (LINK_PIE_SPEC): ..to here.
19803 (LINK_COMMAND_SPEC): Support -no-pie.
19804 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
19805 chain of crtbegin*.o selection, update for PIE_SPEC changes and
19807 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
19808 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
19809 (ENDFILE_CRTEND_SPEC): Similarly.
19811 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
19814 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
19815 (mstack-protector-guard-offset=): Ditto.
19816 * config/i386/i386.c (ix86_option_override): Handle
19817 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
19819 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
19820 ix86_stack_protect_guard_offset variables.
19821 (TARGET_STACK_PROTECT_GUARD): Always define.
19822 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
19823 and -mstack-protector-guard-offset= options.
19825 2017-08-08 Bin Cheng <bin.cheng@arm.com>
19827 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
19828 boundary case for the last candidate.
19830 2017-08-08 Bin Cheng <bin.cheng@arm.com>
19832 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
19833 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
19835 2017-08-08 Tamar Christina <tamar.christina@arm.com>
19837 PR middle-end/19706
19838 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
19839 * config/aarch64/aarch64-builtins.c
19840 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
19841 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
19842 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
19844 2017-08-08 Tamar Christina <tamar.christina@arm.com>
19845 Andrew Pinski <pinskia@gmail.com>
19847 PR middle-end/19706
19848 * internal-fn.def (XORSIGN): New.
19849 * optabs.def (xorsign_optab): New.
19850 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
19851 (convert_expand_mult_copysign): New.
19852 (pass_optimize_widening_mul::execute): Call
19853 convert_expand_mult_copysign.
19855 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19857 PR tree-optimization/81354
19858 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
19859 Insert on edges rather than explicitly creating landing pads.
19860 (analyze_candidates_and_replace): Commit edge inserts.
19862 2017-08-08 Richard Biener <rguenther@suse.de>
19864 PR middle-end/81719
19865 * tree-ssa-loop-niter.c: Include tree-dfa.h.
19866 (expand_simple_operations): Also look through ADDR_EXPRs with
19867 MEM_REF bases treating them as POINTER_PLUS_EXPR.
19869 2017-08-08 Richard Biener <rguenther@suse.de>
19871 PR tree-optimization/81723
19872 * tree-vect-slp.c (struct bst_traits): New hash traits.
19873 (bst_fail): New global.
19874 (vect_build_slp_tree_2): New worker, split out from ...
19875 (vect_build_slp_tree): ... this now wrapping it with using
19876 bst_fail set to cache SLP tree build fails. Properly handle
19878 (vect_analyze_slp_instance): Allocate and free bst_fail.
19880 2017-08-08 Martin Liska <mliska@suse.cz>
19883 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
19884 LABEL_DECLs that can be from a different function.
19886 2017-08-08 Bin Cheng <bin.cheng@arm.com>
19888 PR tree-optimization/81744
19889 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
19890 loop's number of iterations.
19892 2017-08-08 Martin Liska <mliska@suse.cz>
19894 * asan.c: Include header files.
19895 * attribs.c (build_decl_attribute_variant): New function moved
19897 (build_type_attribute_qual_variant): Likewise.
19898 (cmp_attrib_identifiers): Likewise.
19899 (simple_cst_list_equal): Likewise.
19900 (omp_declare_simd_clauses_equal): Likewise.
19901 (attribute_value_equal): Likewise.
19902 (comp_type_attributes): Likewise.
19903 (build_type_attribute_variant): Likewise.
19904 (lookup_ident_attribute): Likewise.
19905 (remove_attribute): Likewise.
19906 (merge_attributes): Likewise.
19907 (merge_type_attributes): Likewise.
19908 (merge_decl_attributes): Likewise.
19909 (merge_dllimport_decl_attributes): Likewise.
19910 (handle_dll_attribute): Likewise.
19911 (attribute_list_equal): Likewise.
19912 (attribute_list_contained): Likewise.
19913 * attribs.h (lookup_attribute): New function moved from tree.[ch].
19914 (lookup_attribute_by_prefix): Likewise.
19915 * bb-reorder.c: Include header files.
19916 * builtins.c: Likewise.
19917 * calls.c: Likewise.
19918 * cfgexpand.c: Likewise.
19919 * cgraph.c: Likewise.
19920 * cgraphunit.c: Likewise.
19921 * convert.c: Likewise.
19922 * dwarf2out.c: Likewise.
19923 * final.c: Likewise.
19924 * fold-const.c: Likewise.
19925 * function.c: Likewise.
19926 * gimple-expr.c: Likewise.
19927 * gimple-fold.c: Likewise.
19928 * gimple-pretty-print.c: Likewise.
19929 * gimple.c: Likewise.
19930 * gimplify.c: Likewise.
19931 * hsa-common.c: Likewise.
19932 * hsa-gen.c: Likewise.
19933 * internal-fn.c: Likewise.
19934 * ipa-chkp.c: Likewise.
19935 * ipa-cp.c: Likewise.
19936 * ipa-devirt.c: Likewise.
19937 * ipa-fnsummary.c: Likewise.
19938 * ipa-inline.c: Likewise.
19939 * ipa-visibility.c: Likewise.
19941 * lto-cgraph.c: Likewise.
19942 * omp-expand.c: Likewise.
19943 * omp-general.c: Likewise.
19944 * omp-low.c: Likewise.
19945 * omp-offload.c: Likewise.
19946 * omp-simd-clone.c: Likewise.
19947 * opts-global.c: Likewise.
19948 * passes.c: Likewise.
19949 * predict.c: Likewise.
19950 * sancov.c: Likewise.
19951 * sanopt.c: Likewise.
19952 * symtab.c: Likewise.
19953 * toplev.c: Likewise.
19954 * trans-mem.c: Likewise.
19955 * tree-chkp.c: Likewise.
19956 * tree-eh.c: Likewise.
19957 * tree-into-ssa.c: Likewise.
19958 * tree-object-size.c: Likewise.
19959 * tree-parloops.c: Likewise.
19960 * tree-profile.c: Likewise.
19961 * tree-ssa-ccp.c: Likewise.
19962 * tree-ssa-live.c: Likewise.
19963 * tree-ssa-loop.c: Likewise.
19964 * tree-ssa-sccvn.c: Likewise.
19965 * tree-ssa-structalias.c: Likewise.
19966 * tree-ssa.c: Likewise.
19967 * tree-streamer-in.c: Likewise.
19968 * tree-vectorizer.c: Likewise.
19969 * tree-vrp.c: Likewise.
19970 * tsan.c: Likewise.
19971 * ubsan.c: Likewise.
19972 * varasm.c: Likewise.
19973 * varpool.c: Likewise.
19974 * tree.c: Remove functions moved to attribs.[ch].
19975 * tree.h: Likewise.
19976 * config/aarch64/aarch64.c: Add attrs.h header file.
19977 * config/alpha/alpha.c: Likewise.
19978 * config/arc/arc.c: Likewise.
19979 * config/arm/arm.c: Likewise.
19980 * config/avr/avr.c: Likewise.
19981 * config/bfin/bfin.c: Likewise.
19982 * config/c6x/c6x.c: Likewise.
19983 * config/cr16/cr16.c: Likewise.
19984 * config/cris/cris.c: Likewise.
19985 * config/darwin.c: Likewise.
19986 * config/epiphany/epiphany.c: Likewise.
19987 * config/fr30/fr30.c: Likewise.
19988 * config/frv/frv.c: Likewise.
19989 * config/ft32/ft32.c: Likewise.
19990 * config/h8300/h8300.c: Likewise.
19991 * config/i386/winnt.c: Likewise.
19992 * config/ia64/ia64.c: Likewise.
19993 * config/iq2000/iq2000.c: Likewise.
19994 * config/lm32/lm32.c: Likewise.
19995 * config/m32c/m32c.c: Likewise.
19996 * config/m32r/m32r.c: Likewise.
19997 * config/m68k/m68k.c: Likewise.
19998 * config/mcore/mcore.c: Likewise.
19999 * config/microblaze/microblaze.c: Likewise.
20000 * config/mips/mips.c: Likewise.
20001 * config/mmix/mmix.c: Likewise.
20002 * config/mn10300/mn10300.c: Likewise.
20003 * config/moxie/moxie.c: Likewise.
20004 * config/msp430/msp430.c: Likewise.
20005 * config/nds32/nds32-isr.c: Likewise.
20006 * config/nds32/nds32.c: Likewise.
20007 * config/nios2/nios2.c: Likewise.
20008 * config/nvptx/nvptx.c: Likewise.
20009 * config/pa/pa.c: Likewise.
20010 * config/pdp11/pdp11.c: Likewise.
20011 * config/powerpcspe/powerpcspe.c: Likewise.
20012 * config/riscv/riscv.c: Likewise.
20013 * config/rl78/rl78.c: Likewise.
20014 * config/rx/rx.c: Likewise.
20015 * config/s390/s390.c: Likewise.
20016 * config/sh/sh.c: Likewise.
20017 * config/sol2.c: Likewise.
20018 * config/sparc/sparc.c: Likewise.
20019 * config/spu/spu.c: Likewise.
20020 * config/stormy16/stormy16.c: Likewise.
20021 * config/tilegx/tilegx.c: Likewise.
20022 * config/tilepro/tilepro.c: Likewise.
20023 * config/v850/v850.c: Likewise.
20024 * config/vax/vax.c: Likewise.
20025 * config/visium/visium.c: Likewise.
20026 * config/xtensa/xtensa.c: Likewise.
20028 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
20031 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
20032 constraints since the -mupper-regs-* switches have been
20034 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
20035 into a vector from a double word element that was extracted from
20036 another vector, and eliminate extra XXPERMDI instructions.
20037 (vsx_concat_<mode>_2): Likewise.
20038 (vsx_concat_<mode>_3): Likewise.
20039 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
20040 concat to allow optimizing inserts from previous extracts.
20042 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
20044 * config/i386/i386.c (ix86_stack_protect_guard): Generate
20045 memory reference to a SSP offset in TLS address space.
20046 (ix86_print_operand) <case '@'>: Remove.
20047 (ix86_print_operand_punct_valid_p): Remove '@' code.
20048 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
20049 UNSPEC_SP_TLS_TEST.
20050 (stack_tls_protect_set_<mode>): Remove.
20051 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
20052 (stack_tls_protect_test_<mode>): Remove.
20053 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
20055 2017-08-07 Olivier Hainque <hainque@adacore.com>
20058 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
20060 2017-08-07 Douglas Rupp <rupp@adacore.com>
20062 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
20063 variable was referenced as multidir in command.
20065 2017-08-07 Jakub Jelinek <jakub@redhat.com>
20068 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
20071 2017-08-07 Martin Liska <mliska@suse.cz>
20073 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
20074 * config/rl78/rl78.c: Add include of attribs.h.
20075 * config/sh/sh.c: Likewise.
20076 * config/v850/v850.c: Likewise.
20078 2017-08-07 Tom de Vries <tom@codesourcery.com>
20080 PR middle-end/78266
20081 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
20083 2017-08-07 Martin Liska <mliska@suse.cz>
20085 * config/mips/mips.c: Include attribs.h.
20087 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
20090 * doc/invoke.texi: Document change in behvaior for -Ofast for
20093 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
20095 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
20097 (aarch64_pop_regs): Likewise.
20098 (aarch64_gen_load_pair): Likewise.
20099 (aarch64_save_callee_saves): Likewise.
20100 (aarch64_restore_callee_saves): Likewise.
20102 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
20104 * config/i386/i386.c: Revert the last change.
20106 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
20109 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
20111 (ix86_finalize_stack_frame_flags): This. Also clear
20112 frame_pointer_needed if -fno-omit-frame-pointer is used without
20114 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
20115 with ix86_finalize_stack_frame_flags.
20116 (ix86_expand_epilogue): Likewise.
20117 (ix86_expand_split_stack_prologue): Likewise.
20119 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
20122 * config/i386/i386.c (get_builtin_code_for_version): Set priority
20123 to P_AES for Westmere.
20125 2017-08-07 Jonathan Yong <10walls@gmail.com>
20127 * config/i386/mingw.opt (fset-stack-executable): Removed.
20128 * config/i386/cygming.opt (fset-stack-executable): Moved
20130 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
20132 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
20134 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
20136 2017-08-07 Marek Polacek <polacek@redhat.com>
20138 PR middle-end/81737
20139 * fold-const.c (fold_indirect_ref_1): Check type_domain.
20141 2017-08-07 Martin Liska <mliska@suse.cz>
20143 * attribs.h (canonicalize_attr_name): New function.
20144 (cmp_attribs): Move from c-format.c and adjusted.
20145 (is_attribute_p): Moved from tree.h.
20146 * tree-inline.c: Add new includes.
20147 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
20148 (private_is_attribute_p): Remove.
20149 (private_lookup_attribute): Likewise.
20150 (private_lookup_attribute_by_prefix): Simplify.
20151 (remove_attribute): Use is_attribute_p.
20152 * tree.h: Remove removed declarations.
20154 2017-08-07 Jakub Jelinek <jakub@redhat.com>
20156 PR middle-end/81698
20157 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
20158 instead of computing it in the function. Formatting fix.
20159 (expand_case): Don't rely on default_edge being the first edge,
20160 clear it if removing it, pass default_edge to
20161 emit_case_dispatch_table.
20162 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
20165 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
20167 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
20168 insn in the function, emit NOP after the insn.
20170 2017-08-06 Tom de Vries <tom@codesourcery.com>
20172 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
20175 2017-08-06 Tom de Vries <tom@codesourcery.com>
20177 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
20180 2017-08-04 Yury Gribov <tetra2005@gmail.com>
20182 PR tree-optimization/57371
20183 * match.pd: New pattern.
20185 2017-08-04 Marek Polacek <polacek@redhat.com>
20187 PR middle-end/81695
20188 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
20189 perform the computation in offset_int.
20191 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
20193 PR tree-optimization/81136
20194 * tree-vectorizer.h: Include tree-hash-traits.h.
20195 (vec_base_alignments): New typedef.
20196 (vec_info): Add a base_alignments field.
20197 (vect_record_base_alignments): Declare.
20198 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
20200 (DR_IS_CONDITIONAL_IN_STMT): New macro.
20201 (create_data_ref): Add an is_conditional_in_stmt argument.
20202 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
20203 the is_conditional_in_stmt field.
20204 (data_ref_loc): Add an is_conditional_in_stmt field.
20205 (get_references_in_stmt): Set the is_conditional_in_stmt field.
20206 (find_data_references_in_stmt): Update call to create_data_ref.
20207 (graphite_find_data_references_in_stmt): Likewise.
20208 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
20209 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
20210 (vect_record_base_alignment): New function.
20211 (vect_record_base_alignments): Likewise.
20212 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
20213 for nested statements even if we fail to compute a misalignment.
20214 Use pooled base alignments for unconditional references.
20215 (vect_find_same_alignment_drs): Compare base addresses instead
20217 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
20218 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
20220 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
20222 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
20223 Add an explicit name for the enum. Use auto_vec for slp_instances
20224 and grouped_stores.
20225 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
20227 (_bb_vec_info): Add a constructor and destructor.
20228 (vinfo_for_stmt): Return NULL for uids of -1 as well.
20229 (destroy_loop_vec_info): Delete.
20230 (vect_destroy_datarefs): Likewise.
20231 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
20232 (vec_info::vec_info): New function.
20233 (vec_info::~vec_info): Likewise.
20234 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
20235 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
20236 destroy_loop_vec_info.
20237 * tree-vect-loop.c (new_loop_vec_info): Replace with...
20238 (_loop_vec_info::_loop_vec_info): ...this.
20239 (destroy_loop_vec_info): Replace with...
20240 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
20241 the stmt_vec_infos. Leave handling of vec_info information to its
20242 destructor. Remove explicit vector releases.
20243 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
20244 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
20245 * tree-vect-slp.c (new_bb_vec_info): Replace with...
20246 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
20247 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
20248 (destroy_bb_vec_info): Replace with...
20249 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
20250 information to its destructor.
20251 (vect_slp_analyze_bb_1): Use new and delete instead of
20252 new_bb_vec_info and destroy_bb_vec_info.
20253 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
20256 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
20258 * tree-data-ref.h (subscript): Add access_fn field.
20259 (data_dependence_relation): Add could_be_independent_p.
20260 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
20261 (same_access_functions): Move to tree-data-ref.c.
20262 * tree-data-ref.c (ref_contains_union_access_p): New function.
20263 (access_fn_component_p): Likewise.
20264 (access_fn_components_comparable_p): Likewise.
20265 (dr_analyze_indices): Add a reference to access_fn_component_p.
20266 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
20268 (constant_access_functions): Likewise.
20269 (add_other_self_distances): Likewise.
20270 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
20271 (initialize_data_dependence_relation): Use XCNEW and remove
20272 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
20273 of access functions that have the same type. Allow the
20274 subsequence to end with different bases in some circumstances.
20275 Record the chosen access functions in SUB_ACCESS_FN.
20276 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
20277 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
20278 (subscript_dependence_tester_1): Likewise dra and drb.
20279 (build_classic_dist_vector): Update calls accordingly.
20280 (subscript_dependence_tester): Likewise.
20281 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
20282 DDR_COULD_BE_INDEPENDENT_P.
20283 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
20284 comp_alias_ddrs instead of may_alias_ddrs.
20285 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
20287 (vect_analyze_data_ref_dependence): Use it if
20288 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
20289 distance vectors if that fails.
20290 (dependence_distance_ge_vf): New function.
20291 (vect_prune_runtime_alias_test_list): Use it. Don't clear
20292 LOOP_VINFO_MAY_ALIAS_DDRS.
20294 2017-08-04 Richard Biener <rguenther@suse.de>
20296 PR middle-end/81705
20297 * fold-const.c (fold_binary_loc): Properly restrict
20298 minus_var0 && minus_var1 case when associating undefined overflow
20301 2017-08-04 Tom de Vries <tom@codesourcery.com>
20303 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
20305 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20307 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20308 Don't start diagnostic messages with a capital letter.
20309 * config/rs6000/rs6000.c (rs6000_option_override_internal):
20311 (rs6000_invalid_builtin): Likewise.
20312 (rs6000_trampoline_init): Likewise.
20314 2017-08-03 Jakub Jelinek <jakub@redhat.com>
20317 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
20318 after setting changeable df flags.
20320 2017-08-03 Richard Biener <rguenther@suse.de>
20322 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
20323 up if the use is in USE - X.
20325 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
20327 * toplev.c (dumpfile.h): New include.
20328 (internal_error_reentered): New static function. Use it...
20329 (internal_error_function): ...here to handle reentered internal_error.
20331 2017-08-03 Richard Biener <rguenther@suse.de>
20333 PR middle-end/81148
20334 * fold-const.c (split_tree): Add minus_var and minus_con
20335 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
20336 here but always use minus_*.
20337 (associate_trees): Assert we never associate with MINUS_EXPR
20338 and NULL first operand. Do not recurse for PLUS_EXPR operands
20339 when associating as MINUS_EXPR either.
20340 (fold_binary_loc): Track minus_var and minus_con.
20342 2017-08-03 Tom de Vries <tom@codesourcery.com>
20345 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
20346 ACCEL_COMPILER, apply finish_options on
20347 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
20349 2017-08-03 Tom de Vries <tom@codesourcery.com>
20352 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
20353 function_entry_patch_area_size > 0.
20355 2017-08-03 Jakub Jelinek <jakub@redhat.com>
20358 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
20359 instead of 10??LU, perform unit multiplication in wide_int,
20360 don't change alloc_object_size_limit if the limit is larger
20363 PR tree-optimization/81655
20364 PR tree-optimization/81588
20365 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
20366 the case when ranges[i].low and high are 1 for unsigned type with
20369 PR middle-end/81052
20370 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
20371 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
20373 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20375 * tree-vrp.h: Add include guard.
20377 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
20380 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
20381 (ud2): New insn pattern.
20382 * config/i386/i386.c (ix86_expand_epilogue):
20383 For naked functions, generate ud2 instead of trap insn.
20385 2017-08-02 Marek Polacek <polacek@redhat.com>
20388 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
20390 2017-08-02 Tom de Vries <tom@codesourcery.com>
20391 Cesar Philippidis <cesar@codesourcery.com>
20393 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
20394 Add missing edge probabilities.
20396 2017-08-02 Tamar Christina <tamar.christina@arm.com>
20398 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
20399 Correct endianness.
20401 2017-08-02 Jakub Jelinek <jakub@redhat.com>
20403 PR middle-end/79499
20404 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
20405 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
20406 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
20407 prologue_seq sequences - if any.
20409 2017-08-02 Richard Biener <rguenther@suse.de>
20411 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
20412 via vectors if supported, integer extracts via punning if supported
20413 or otherwise vector extracts.
20415 2017-08-02 Richard Biener <rguenther@suse.de>
20417 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
20419 (bitmap_insert_into_set): ... this.
20421 2017-08-02 Richard Biener <rguenther@suse.de>
20423 PR tree-optimization/81633
20425 2015-08-17 Alan Hayward <alan.hayward@arm.com>
20427 PR tree-optimization/71752
20428 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
20430 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
20432 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
20433 (machine_function::call_ms2sysv_pad_out): Remove field.
20434 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
20435 (ix86_compute_frame_layout): Likewise.
20437 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
20440 * config/i386/i386.c (ix86_set_func_type): Disallow naked
20441 attribute with interrupt attribute.
20443 2017-08-01 Andrew Pinski <apinski@cavium.com>
20445 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
20446 BIT_INSERT_EXPR's operand 1
20447 to see if the types precision matches.
20449 2017-08-01 Martin Liska <mliska@suse.cz>
20451 PR middle-end/70140
20452 * builtins.c (expand_builtin_memcpy_args): Remove.
20453 (expand_builtin_memcpy): Call newly added function
20454 expand_builtin_memory_copy_args.
20455 (expand_builtin_memcpy_with_bounds): Likewise.
20456 (expand_builtin_mempcpy): Remove last argument.
20457 (expand_builtin_mempcpy_with_bounds): Likewise.
20458 (expand_builtin_memory_copy_args): New function created from
20459 expand_builtin_mempcpy_args with small modifications.
20460 (expand_builtin_mempcpy_args): Remove.
20461 (expand_builtin_stpcpy): Remove unused argument.
20462 (expand_builtin): Likewise.
20463 (expand_builtin_with_bounds): Likewise.
20465 2017-08-01 Martin Liska <mliska@suse.cz>
20468 Make mempcpy more optimal (PR middle-end/70140).
20470 2017-08-01 Jakub Jelinek <jakub@redhat.com>
20473 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
20474 __builtin_vec_cmpne verify both arguments are compatible vectors
20475 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
20476 verify arg1_type is a pointer or array type. For __builtin_vec_st,
20477 move computation of aligned to after checking the argument types.
20481 * config/rs6000/vsx.md (vextract_fp_from_shorth,
20482 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
20485 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
20486 Doug Rupp <rupp@adacore.com>
20487 Olivier Hainque <hainque@adacore.com>
20489 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
20490 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
20492 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
20493 for TARGET_OS_CPP_BUILTIN.
20494 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
20495 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
20497 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
20498 passing required abi options to the assembler for EABI configurations.
20499 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
20500 of .text.hot and .text.unlikely sections for kernel modules when
20501 using ARM style exceptions.
20502 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
20503 options. Add EXTRA_CC1_SPEC.
20504 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
20506 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
20508 (ARM_TARGET2_DWARF_FORMAT): Define.
20509 * config/arm/t-vxworks: Adjust multilib control to removal of the
20510 Diab command line options.
20512 2017-08-01 Martin Liska <mliska@suse.cz>
20514 PR gcov-profile/81561
20515 * gcov.c (unblock): Make unblocking safe as we need to preserve
20516 index correspondence of blocks and block_lists.
20518 2017-08-01 Richard Biener <rguenther@suse.de>
20520 PR tree-optimization/81181
20521 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
20522 (compute_antic): ... end of iteration here.
20524 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
20526 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
20527 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
20528 (ftree-slp-vectorize): Likewise.
20529 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
20530 can no longer be set independent of flag_tree_loop_vectorize.
20531 * omp-general.c (emp_max_vf): Likewise.
20532 * opts.c (enable_fdo_optimizations): Remove references to
20533 flag_tree_vectorize, these are now implicit.
20534 (common_handle_option): Remove handling for OPT_ftree_vectorize,
20535 and leave it for the options machinery.
20537 2017-08-01 Martin Liska <mliska@suse.cz>
20539 PR middle-end/70140
20540 * builtins.c (expand_builtin_memcpy_args): Remove.
20541 (expand_builtin_memcpy): Call newly added function
20542 expand_builtin_memory_copy_args.
20543 (expand_builtin_memcpy_with_bounds): Likewise.
20544 (expand_builtin_mempcpy): Remove last argument.
20545 (expand_builtin_mempcpy_with_bounds): Likewise.
20546 (expand_builtin_memory_copy_args): New function created from
20547 expand_builtin_mempcpy_args with small modifications.
20548 (expand_builtin_mempcpy_args): Remove.
20549 (expand_builtin_stpcpy): Remove unused argument.
20550 (expand_builtin): Likewise.
20551 (expand_builtin_with_bounds): Likewise.
20553 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
20556 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
20557 print "ds:" only for immediates in generic address space.
20559 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
20562 * config/i386/i386.c (ix86_funciton_naked): New prototype.
20563 (ix86_function_ok_for_sibcall): Return false for naked functions.
20565 2017-08-01 Richard Biener <rguenther@suse.de>
20567 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
20568 (compute_antic): Seed worklist with exit block predecessors.
20569 * cfganal.c (dfs_find_deadend): For a cycle return the source
20570 of the edge closing it.
20572 2017-08-01 Tamar Christina <tamar.christina@arm.com>
20574 * config/aarch64/aarch64.c
20575 (aarch64_can_const_movi_rtx_p): Move 0 check.
20577 2017-08-01 Bin Cheng <bin.cheng@arm.com>
20579 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
20580 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
20582 * match.pd: Ditto in address comparison pattern.
20584 2017-08-01 Bin Cheng <bin.cheng@arm.com>
20586 PR tree-optimization/81627
20587 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
20588 closed ssa form for store-store chain.
20590 2017-08-01 Bin Cheng <bin.cheng@arm.com>
20592 PR tree-optimization/81620
20593 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
20594 for store-store chain.
20596 2017-08-01 Jakub Jelinek <jakub@redhat.com>
20598 PR tree-optimization/81588
20599 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
20600 ranges[i].in_p, invert comparison code ccode. For >/>=,
20601 swap rhs1 and rhs2 and comparison code unconditionally,
20602 for </<= don't do that. Don't swap rhs1/rhs2 again if
20603 ranges[i].in_p, instead invert comparison code ccode if
20604 opcode or oe->rank is BIT_IOR_EXPR.
20607 * optabs.def (vec_extract_optab, vec_init_optab): Change from
20608 a direct optab to conversion optab.
20609 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
20610 with GET_MODE_INNER as last argument instead of optab_handler.
20611 * expmed.c (extract_bit_field_1): Likewise. Use vector from
20612 vector extraction if possible and optab is available.
20613 * expr.c (store_constructor): Use convert_optab_handler instead
20614 of optab_handler. Use vector initialization from smaller
20615 vectors if possible and optab is available.
20616 * tree-vect-stmts.c (vectorizable_load): Likewise.
20617 * doc/md.texi (vec_extract, vec_init): Document that the optabs
20618 now have two modes.
20619 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
20620 of vec_init from half-sized vectors with the same element mode.
20621 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
20622 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
20623 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
20624 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
20625 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
20626 after mode in gen_vec_extract* calls.
20627 (vec_extract<mode>): Renamed to ...
20628 (vec_extract<mode><ssescalarmodelower>): ... this.
20629 (vec_extract<mode><ssehalfvecmodelower>): New expander.
20630 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
20631 element mode after mode in gen_vec_init* calls.
20632 (VEC_INIT_HALF_MODE): New mode iterator.
20633 (vec_init<mode>): Renamed to ...
20634 (vec_init<mode><ssescalarmodelower>): ... this.
20635 (vec_init<mode><ssehalfvecmodelower>): New expander.
20636 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
20637 (vec_extractv2sfsf): ... this.
20638 (vec_initv2sf): Renamed to ...
20639 (vec_initv2sfsf): ... this.
20640 (vec_extractv2si): Renamed to ...
20641 (vec_extractv2sisi): ... this.
20642 (vec_initv2si): Renamed to ...
20643 (vec_initv2sisi): ... this.
20644 (vec_extractv4hi): Renamed to ...
20645 (vec_extractv4hihi): ... this.
20646 (vec_initv4hi): Renamed to ...
20647 (vec_initv4hihi): ... this.
20648 (vec_extractv8qi): Renamed to ...
20649 (vec_extractv8qiqi): ... this.
20650 (vec_initv8qi): Renamed to ...
20651 (vec_initv8qiqi): ... this.
20652 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
20653 (vec_init<mode>): Renamed to ...
20654 (vec_init<mode><VEC_base_l>): ... this.
20655 (vec_extract<mode>): Renamed to ...
20656 (vec_extract<mode><VEC_base_l>): ... this.
20657 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
20658 (vec_initv2sfsf): ... this.
20659 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
20660 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
20661 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
20662 element mode after mode in gen_vec_init* calls.
20663 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
20664 (vec_init<mode><Vel>): ... this.
20665 (vec_extract<mode>): Renamed to ...
20666 (vec_extract<mode><Vel>): ... this.
20667 * config/aarch64/iterators.md (Vel): New mode attribute.
20668 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
20669 Add element mode after mode in gen_vec_extract* calls.
20670 * config/s390/vector.md (non_vec_l): New mode attribute.
20671 (vec_extract<mode>): Renamed to ...
20672 (vec_extract<mode><non_vec_l>): ... this.
20673 (vec_init<mode>): Renamed to ...
20674 (vec_init<mode><non_vec_l>): ... this.
20675 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
20676 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
20678 * config/arm/iterators.md (V_elem_l): New mode attribute.
20679 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
20680 (vec_extract<mode><V_elem_l>): ... this.
20681 (vec_extractv2di): Renamed to ...
20682 (vec_extractv2didi): ... this.
20683 (vec_init<mode>): Renamed to ...
20684 (vec_init<mode><V_elem_l>): ... this.
20685 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
20686 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
20687 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
20688 Add element mode after gen_vec_extract* calls.
20689 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
20690 (vec_init<mode><unitmode>): ... this.
20691 (vec_extract<mode>): Renamed to ...
20692 (vec_extract<mode><unitmode>): ... this.
20693 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
20694 (vec_init<mode><unitmode>): ... this.
20695 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
20696 (vec_initv2sfsf): ... this.
20697 (vec_extractv2sf): Renamed to ...
20698 (vec_extractv2sfsf): ... this.
20699 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
20700 Add element mode after gen_vec_extract* calls.
20701 * config/mips/mips.md (unitmode): New mode iterator.
20702 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
20703 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
20704 * config/spu/spu.md (inner_l): New mode attribute.
20705 (vec_init<mode>): Renamed to ...
20706 (vec_init<mode><inner_l>): ... this.
20707 (vec_extract<mode>): Renamed to ...
20708 (vec_extract<mode><inner_l>): ... this.
20709 * config/sparc/sparc.md (veltmode): New mode iterator.
20710 (vec_init<VMALL:mode>): Renamed to ...
20711 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
20712 * config/ia64/vect.md (vec_initv2si): Renamed to ...
20713 (vec_initv2sisi): ... this.
20714 (vec_initv2sf): Renamed to ...
20715 (vec_initv2sfsf): ... this.
20716 (vec_extractv2sf): Renamed to ...
20717 (vec_extractv2sfsf): ... this.
20718 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
20719 (vec_init<mode>): Renamed to ...
20720 (vec_init<mode><VEC_base_l>): ... this.
20721 (vec_extract<mode>): Renamed to ...
20722 (vec_extract<mode><VEC_base_l>): ... this.
20723 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
20724 (vec_initv2sfsf): ... this.
20725 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
20726 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
20727 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
20728 gen_vec_init* calls.
20730 2017-08-01 Richard Biener <rguenther@suse.de>
20732 PR tree-optimization/81297
20733 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
20734 TREE_OVERFLOW from INTEGER_CSTs.
20736 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
20738 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
20740 2017-07-31 Carl Love <cel@us.ibm.com>
20742 * config/rs6000/rs6000-c: Add support for built-in functions
20743 vector signed char vec_xl_be (signed long long, signed char *);
20744 vector unsigned char vec_xl_be (signed long long, unsigned char *);
20745 vector signed int vec_xl_be (signed long long, signed int *);
20746 vector unsigned int vec_xl_be (signed long long, unsigned int *);
20747 vector signed long long vec_xl_be (signed long long, signed long long *);
20748 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
20749 vector signed short vec_xl_be (signed long long, signed short *);
20750 vector unsigned short vec_xl_be (signed long long, unsigned short *);
20751 vector double vec_xl_be (signed long long, double *);
20752 vector float vec_xl_be (signed long long, float *);
20753 * config/rs6000/altivec.h (vec_xl_be): Add #define.
20754 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
20755 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
20757 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
20758 (altivec_expand_builtin): Add switch statement to call
20759 altivec_expand_xl_be for each builtin.
20760 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
20761 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
20762 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
20763 __builtin_vsx_le_be_v16qi.
20764 * doc/extend.texi: Update the built-in documentation file for the
20765 new built-in functions.
20767 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
20770 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
20772 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
20774 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20776 * config.gcc: Add z14.
20777 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
20778 CPU model numbers for z13s and z14.
20779 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
20781 * config/s390/s390-opts.h (enum processor_type): Rename
20782 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
20783 * config/s390/s390.c (processor_table): Add field for CPU name to
20784 be passed to Binutils.
20785 (s390_asm_output_machine_for_arch): Use the new field in
20786 processor_table for Binutils.
20787 (s390_expand_builtin): Replace arch12 with z14.
20788 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
20789 (s390_get_sched_attrmask): Likewise.
20790 (s390_get_unit_mask): Likewise.
20791 * config/s390/s390.opt: Add z14 to processor_type enum.
20793 2017-07-31 Martin Jambor <mjambor@suse.cz>
20796 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
20797 regardless of optimization level.
20799 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
20800 Martin Liska <mliska@suse.cz>
20802 * predict.def: Remove old comment and adjust probability.
20803 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
20804 PREDICT statements.
20806 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
20809 * config/i386/i386.c (ix86_function_naked): New function.
20810 (ix86_can_use_return_insn_p): Return false for naked functions.
20811 (ix86_expand_prologue): Skip prologue for naked functions.
20812 (ix86_expand_epilogue): Skip epilogue for naked functions
20813 and emit trap instruction.
20814 (ix86_warn_func_return): New function.
20815 (ix86_attribute_table): Add "naked" attribute specification.
20816 (TARGET_WARN_FUNC_RETURN): Define.
20817 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
20819 2017-07-31 Martin Liska <mliska@suse.cz>
20821 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
20822 (dump_gimple_bb_header): Always dump BB info.
20823 (pp_cfg_jump): Do not append info about BB when dumping a jump.
20825 2017-07-31 Martin Liska <mliska@suse.cz>
20828 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
20829 also with current_function_decl non-null equality.
20831 2017-07-31 Jakub Jelinek <jakub@redhat.com>
20834 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
20835 change type to the element type, instead add eltype variable and
20836 use it where we are interested in the element type.
20838 PR tree-optimization/81603
20839 * ipa-polymorphic-call.c
20840 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
20841 offset arithmetic in offset_int, bail out if the resulting bit offset
20842 doesn't fit into shwi.
20844 2017-07-31 Martin Liska <mliska@suse.cz>
20846 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
20847 (gimplify_save_expr): Fix comment.
20849 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
20852 * config/i386/i386.c (ix86_function_arg): Update arguments for
20854 (ix86_compute_frame_layout): Set the initial stack offset to
20855 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
20856 INCOMING_FRAME_SP_OFFSET.
20857 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
20858 stack before exception handler returns.
20859 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
20860 the 'ERROR_CODE' for exception handler.
20862 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
20864 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
20865 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
20866 (ASM_OUTPUT_REG_POP): Ditto.
20867 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
20868 instead of asm_fprintf to output pure string.
20870 2017-07-29 Jakub Jelinek <jakub@redhat.com>
20872 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
20873 to imported_module_or_decl hook.
20874 (debug_nothing_tree_tree_tree_bool): Remove.
20875 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
20876 * debug.c (do_nothing_debug_hooks): Use
20877 debug_nothing_tree_tree_tree_bool_bool instead of
20878 debug_nothing_tree_tree_tree_bool.
20879 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
20880 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
20881 * sdbout.c (sdb_debug_hooks): Likewise.
20882 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
20883 (gen_namespace_die): Add DW_AT_export_symbols attribute if
20885 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
20886 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
20887 attribute, don't add anything.
20889 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20891 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
20892 (fold_build2_stat_loc): Likewise.
20893 (fold_build3_stat_loc): Likewise.
20894 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
20895 (fold_build1_loc): Remove macro.
20896 (fold_build2_loc): Likewise.
20897 (fold_build3_loc): Likewise.
20899 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20901 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
20902 (gimple_build_debug_bind_source_stat): Likewise.
20903 * gimple.h (gimple_build_debug_bind): Remove macro.
20904 (gimple_build_debug_bind_source): Likewise.
20906 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20908 * bitmap.c (bitmap_alloc): Adjust.
20909 (bitmap_gc_alloc): Likewise.
20910 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
20912 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20914 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
20915 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
20916 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
20917 (bitmap_gc_alloc_stat): Likewise.
20918 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
20920 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20922 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
20923 * rtl.h (shallow_copy_rtx): Remove macro.
20925 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20927 * emit-rtl.c (gen_raw_REG): Adjust.
20928 * gengenrtl.c (gendef): Likewise.
20929 * rtl.c (rtx_alloc_stat): Remove _stat from name.
20930 * rtl.h (rtx_alloc): Remove macro.
20932 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20934 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
20935 (build_tree_list_stat): Likewise.
20936 * tree.h (build_tree_list): Remove macro.
20937 (build_tree_list_vec): Likewise.
20939 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20941 * tree.c (make_vector_stat): Remove _stat from name.
20942 (build_vector_stat): Likewise.
20943 * tree.h (make_vector_stat): Remove macro.
20944 (build_vector_stat): Likewise.
20946 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20948 * tree.h (build_var_debug_value): Remove prototype.
20950 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20952 * tree.c (tree_cons_stat): Remove _stat from name.
20953 * tree.h (tree_cons): Remove macro.
20955 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20957 * tree.c (build_vl_exp_stat): Remove _stat from name.
20958 * tree.h (build_vl_exp): Remove macro.
20960 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20962 * tree.c (build_decl_stat): Remove _stat from name.
20963 * tree.h (build_decl): Remove macro.
20965 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20967 * gimple.c (gimple_build_with_ops_stat): Adjust.
20968 (gimple_alloc_stat): Remove _stat from name.
20969 * gimple.h (gimple_alloc): Remove macro.
20971 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20973 * tree.c (make_tree_vec_stat): Remove _stat from name.
20974 (grow_tree_vec_stat): Likewise.
20975 * tree.h (make_tree_vec_stat): Adjust prototype.
20976 (grow_tree_vec_stat): Likewise.
20977 (make_tree_vec): Remove macro.
20978 (grow_tree_vec): Likewise.
20980 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20982 * fold-const.c (fold_build1_stat_loc): Adjust.
20983 (fold_build2_stat_loc): Likewise.
20984 (fold_build3_stat_loc): Likewise.
20985 * tree.c (build0_stat): Remove _stat from name.
20986 (build1_stat): Likewise.
20987 (build2_stat): Likewise.
20988 (build3_stat): Likewise.
20989 (build4_stat): Likewise.
20990 (build5_stat): Likewise.
20991 * tree.h (build1_loc): Remove macro, and rename _stat function
20993 (build2_loc): Likewise.
20994 (build3_loc): Likewise.
20995 (build4_loc): Likewise.
20996 (build5_loc): Likewise.
20998 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21000 * tree.c (make_int_cst_stat): Remove _stat from name.
21001 * tree.h (make_int_cst_stat): Adjust prototype.
21002 (make_int_cst): Remove macro.
21004 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21006 * tree.c (make_tre_binfo_stat): Remove _stat from name.
21007 * tree.h (make_tree_binfo_stat): Adjust prototype.
21008 (make_tree_binfo): Remove.
21010 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21012 * tree.c (copy_node_stat): Rename to copy_node.
21013 (build_distinct_type_copy): Adjust.
21014 * tree.h (copy_node_stat): Adjust prototype.
21015 (copy_node): Remove macro.
21017 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21019 * tree.c (make_node_stat): rename to make_node.
21020 (build_tree_list_stat): Adjust.
21021 (build0_stat): Likewise.
21022 (build2_stat): Likewise.
21023 (build3_stat): Likewise.
21024 (build4_stat): Likewise.
21025 (build5_stat): Likewise.
21026 (build_decl_stat): Likewise.
21027 * tree.h (make_node_stat): Adjust prototype.
21028 (make_node): remove macro.
21030 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
21032 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
21033 (PPC_FEATURE2_SCV): Likewise.
21034 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
21036 2017-07-28 Tamar Christina <tamar.christina@arm.com>
21038 * config/aarch64/aarch64.c
21039 (aarch64_internal_mov_immediate): Add new special pattern.
21040 * config/aarch64/aarch64.md (*movdi_aarch64):
21041 Add reg/32bit const mov case.
21043 2017-07-28 Tamar Christina <tamar.christina@arm.com>
21044 Richard Sandiford <richard.sandiford@linaro.org>
21046 * config/aarch64/aarch64.md (mov<mode>): Generalize.
21047 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
21048 Add integer and movi cases.
21049 (movi-split-hf-df-sf split, fp16): New.
21050 (enabled): Added TARGET_FP_F16INST.
21051 * config/aarch64/iterators.md (GPF_HF): New.
21052 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
21054 2017-07-28 Tamar Christina <tamar.christina@arm.com>
21056 * config/aarch64/aarch64.c
21057 (aarch64_simd_container_mode): Add prototype.
21058 (aarch64_expand_mov_immediate): Add HI support.
21059 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
21060 (aarch64_can_const_movi_rtx_p): New.
21061 (aarch64_preferred_reload_class):
21062 Remove restrictions of using FP registers for certain SIMD operations.
21063 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
21064 (aarch64_valid_floating_const): Add integer move validation.
21065 (aarch64_simd_imm_scalar_p): Remove.
21066 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
21067 (aarch64_legitimate_constant_p): Expand list of supported cases.
21068 * config/aarch64/aarch64-protos.h
21069 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
21070 (aarch64_reinterpret_float_as_int): New.
21071 (aarch64_simd_imm_scalar_p): Remove.
21072 * config/aarch64/constraints.md (Uvi): New.
21073 (Dd): Split into Ds and new Dd.
21074 * config/aarch64/aarch64.md (*movsi_aarch64):
21076 (*movdi_aarch64): Add SIMD mov case.
21078 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21080 * tree-predcom.c: (struct chain): Handle store-store chain in which
21081 stores for elimination only store loop invariant values.
21082 (execute_pred_commoning_chain): Ditto.
21083 (prepare_initializers_chain_store_elim): Ditto.
21084 (prepare_finalizers): Ditto.
21085 (is_inv_store_elimination_chain): New function.
21086 (initialize_root_vars_store_elim_1): New function.
21088 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21090 * tree-predcom.c: Revise general description of the pass.
21091 (enum chain_type): New enum type for store elimination.
21092 (struct chain): New field supporting store elimination.
21093 (struct component): Ditto.
21094 (dump_chain): Dump store-stores chain.
21095 (release_chain): Release resources.
21096 (split_data_refs_to_components): Compute and create component
21097 contains only stores for elimination.
21098 (get_chain_last_ref_at): New function.
21099 (make_invariant_chain): Initialization.
21100 (make_rooted_chain): Specify chain type in parameter and record it.
21101 (add_looparound_copies): Skip for store-stores chain.
21102 (determine_roots_comp): Compute type of chain and pass it to
21104 (initialize_root_vars_store_elim_2): New function.
21105 (finalize_eliminated_stores): New function.
21106 (remove_stmt): Handle store for elimination.
21107 (execute_pred_commoning_chain): Execute predictive commoning on
21108 store-store chains.
21109 (determine_unroll_factor): Skip unroll for store-stores chain.
21110 (prepare_initializers_chain_store_elim): New function.
21111 (prepare_initializers_chain): Hanlde store-store chain.
21112 (prepare_finalizers_chain, prepare_finalizers): New function.
21113 (tree_predictive_commoning_loop): Return integer value indicating
21114 if loop is unrolled or lcssa form is corrupted.
21115 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
21117 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21119 * tree-predcom.c (initialize_root): Delete.
21120 (execute_pred_commoning_chain): Initialize root vars and replace
21121 reference of non-combined chain directly, rather than call above
21124 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21126 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
21127 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
21129 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21131 * tree-predcom.c (struct chain): New field init_seq.
21132 (release_chain): Release init_seq.
21133 (prepare_initializers_chain): Record intialization stmts in above
21135 (insert_init_seqs): New function.
21136 (tree_predictive_commoning_loop): Call insert_init_seqs.
21138 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21140 * tree-predcom.c (determine_roots_comp): Skip trivial components.
21142 2017-07-28 Richard Biener <rguenther@suse.de>
21144 * match.pd: Remove superfluous :c.
21145 * genmatch.c (simplify::id): Add member.
21146 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
21148 (current_id): New global.
21149 (dt_node::parent): Move from ...
21150 (dt_operand::parent): ... here. Add for_id member.
21151 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
21152 (decision_tree::find_node): Relax order requirement when
21153 merging DT_TRUE nodes to ones inbetween the current simplify
21154 and the one we try to merge with. Add diagnostic whenever
21155 we need to enforce pattern order by not merging.
21156 (decision_tree::insert): Set current_id.
21157 (decision_tree::print_node): Dump parent node and for_id.
21158 (parser::last_id): Add member.
21159 (parser::push_simplify): Assign unique id.
21160 (parser::parser): Initialize last_id.
21162 2017-07-28 Martin Liska <mliska@suse.cz>
21165 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
21166 gimple_build_debug_bind.
21168 2017-07-28 Richard Biener <rguenther@suse.de>
21170 PR tree-optimization/81502
21171 * match.pd: Add pattern combining BIT_INSERT_EXPR with
21173 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
21175 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
21176 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
21177 for BIT_FIELD_REF args.
21178 * fold-const.c (make_bit_field_ref): Likewise.
21179 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
21181 2017-07-28 Jakub Jelinek <jakub@redhat.com>
21184 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
21185 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
21186 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
21187 Or it into SANITIZER_UNDEFINED.
21188 * ubsan.c: Include gimple-fold.h and varasm.h.
21189 (ubsan_expand_ptr_ifn): New function.
21190 (instrument_pointer_overflow): New function.
21191 (maybe_instrument_pointer_overflow): New function.
21192 (instrument_object_size): Formatting fix.
21193 (pass_ubsan::execute): Call instrument_pointer_overflow
21194 and maybe_instrument_pointer_overflow.
21195 * internal-fn.c (expand_UBSAN_PTR): New function.
21196 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
21197 * sanitizer.def (__ubsan_handle_pointer_overflow,
21198 __ubsan_handle_pointer_overflow_abort): New builtins.
21199 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
21200 * internal-fn.def (UBSAN_PTR): New internal function.
21201 * opts.c (sanitizer_opts): Add pointer-overflow.
21202 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
21203 * fold-const.c (build_range_check): Compute pointer range check in
21204 integral type if pointer arithmetics would be needed. Formatting
21207 2017-07-28 Martin Liska <mliska@suse.cz>
21210 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
21211 parameters that are of a variable-length.
21213 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
21215 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
21217 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
21218 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
21219 (CRT_CALL_STATIC_FUNCTION): Likewise.
21220 (ASM_DEFAULT_SPEC): New define.
21221 (ASM_SPEC32): Likewise.
21222 (ASM_SPEC64): Likewise.
21223 (ASM_SPEC_COMMON): Likewise.
21224 (ASM_SPEC): Likewise.
21225 (INVALID_64BIT): Likewise.
21226 (LINK_OS_DEFAULT_SPEC): Likewise.
21227 (LINK_OS_SPEC32): Likewise.
21228 (LINK_OS_SPEC64): Likewise.
21229 (POWERPC_LINUX): Likewise.
21230 (PTRDIFF_TYPE): Likewise.
21231 (RESTORE_FP_PREFIX): Likewise.
21232 (RESTORE_FP_SUFFIX): Likewise.
21233 (SAVE_FP_PREFIX): Likewise.
21234 (SAVE_FP_SUFFIX): Likewise.
21235 (SIZE_TYPE): Likewise.
21236 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
21237 (TARGET_64BIT): Likewise.
21238 (TARGET_64BIT): Likewise.
21239 (TARGET_AIX): Likewise.
21240 (WCHAR_TYPE_SIZE): Likewise.
21241 (WCHAR_TYPE): Undefine.
21242 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
21243 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
21244 (CPP_OS_RTEMS_SPEC): Delete.
21245 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
21246 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
21248 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
21250 2017-07-28 Jakub Jelinek <jakub@redhat.com>
21252 PR tree-optimization/81578
21253 * tree-parloops.c (build_new_reduction): Bail out if
21254 reduction_code isn't one of the standard OpenMP reductions.
21255 Move the details printing after that decision.
21257 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
21259 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
21260 related to reload_in_progress.
21261 (splat_input_operand): Likewise.
21262 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
21264 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
21266 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
21267 (TARGET_INSTANTIATE_DECLS): Likewise.
21268 (legitimate_indexed_address_p): Delete reload_in_progress code.
21269 (rs6000_debug_legitimate_address_p): Likewise.
21270 (rs6000_eliminate_indexed_memrefs): Likewise.
21271 (rs6000_emit_le_vsx_store): Likewise.
21272 (rs6000_emit_move_si_sf_subreg): Likewise.
21273 (rs6000_emit_move): Likewise.
21274 (register_to_reg_type): Likewise.
21275 (rs6000_pre_atomic_barrier): Likewise.
21276 (rs6000_machopic_legitimize_pic_address): Likewise.
21277 (rs6000_allocate_stack_temp): Likewise.
21278 (rs6000_address_for_fpconvert): Likewise.
21279 (rs6000_address_for_altivec): Likewise.
21280 (rs6000_secondary_memory_needed_rtx): Delete function.
21281 (rs6000_check_sdmode): Likewise.
21282 (rs6000_alloc_sdmode_stack_slot): Likewise.
21283 (rs6000_instantiate_decls): Likewise.
21284 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
21285 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
21286 Delete reload_in_progress.
21287 (*vec_reload_and_plus_<mptrsize>): Likewise.
21288 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
21289 (vsx_div_v2di): Likewise.
21290 (vsx_udiv_v2di): Likewise.
21292 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
21294 * config/rs6000/rs6000.opt (mlra): Replace with stub.
21295 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
21296 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
21297 (rs6000_debug_reg_global): Delete print of LRA status.
21298 (rs6000_option_override_internal): Delete dead LRA related code.
21299 (rs6000_lra_p): Delete function.
21300 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
21302 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
21304 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
21305 * config/riscv/rtems.h: New file.
21307 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21308 Sudakshina Das <sudi.das@arm.com>
21310 * config/aarch64/aarch64.md
21311 (define_split for and<mode>3nr_compare): Move
21312 non aarch64_logical_operand to a register.
21313 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
21314 register immediate operand to a register.
21315 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
21317 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
21319 PR middle-end/81564
21320 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
21322 2017-07-27 Richard Biener <rguenther@suse.de>
21324 PR tree-optimization/81573
21325 PR tree-optimization/81494
21326 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
21327 multi defuse cycle case.
21329 2017-07-27 Richard Biener <rguenther@suse.de>
21331 PR tree-optimization/81571
21332 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
21335 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
21337 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
21338 earlier and only if MASK_FPU is set. Adjust formatting.
21340 2017-07-27 Martin Liska <mliska@suse.cz>
21342 * opt-functions.awk: Add validation of value of Init.
21343 * optc-gen.awk: Pass new argument.
21345 2017-07-27 Martin Liska <mliska@suse.cz>
21347 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
21348 Fix wrong condition.
21350 2017-07-27 Martin Liska <mliska@suse.cz>
21352 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
21353 BBs and edges seen by autoFDO.
21355 2017-07-27 Richard Biener <rguenther@suse.de>
21357 PR tree-optimization/81502
21358 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
21359 with incompatible but same sized type.
21360 (execute_update_addresses_taken): Likewise.
21362 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
21364 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
21365 flag_tree_loop_vectorize rather than flag_tree_vectorize.
21367 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21370 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
21371 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
21372 Change s_operand to memory_operand.
21374 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
21376 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
21377 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
21378 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
21379 Emit instructions rather than returning an expression. Handle TFmode
21380 and KFmode by casting to TImode.
21381 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
21382 (rs6000_emit_le_vsx_store): Likewise.
21383 * config/rs6000/vsx.md (VSX_TI): New iterator.
21384 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
21385 (*vsx_le_undo_permute_<mode>): Likewise.
21386 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
21387 emit the split sequence.
21388 (*vsx_le_perm_store_<mode>): Likewise.
21390 2017-07-27 Jakub Jelinek <jakub@redhat.com>
21392 PR tree-optimization/81555
21393 PR tree-optimization/81556
21394 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
21395 if true, force CHANGED for the recursive invocation.
21396 (reassociate_bb): Remember original length of ops array, pass
21397 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
21399 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
21400 attributes for noipa attribute. For naked attribute use
21401 lookup_attribute first before lookup_attribute_spec.
21402 * final.c (rest_of_handle_final): Disable IPA RA for functions with
21404 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
21405 for functions with noipa attribute.
21406 (cgraph_externally_visible_p): Return true for functions with noipa
21408 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
21409 for functions with noipa attribute.
21410 * doc/extend.texi: Document noipa function attribute.
21411 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
21412 also for functions with noipa attribute.
21413 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
21415 2017-07-26 Andrew Pinski <apinski@cavium.com>
21417 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
21418 vec_unalign_load_cost and vec_unalign_store_cost.
21420 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
21422 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
21423 -mvsx-small-integer option.
21424 (ISA_3_0_MASKS_IEEE): Likewise.
21425 (OTHER_VSX_VECTOR_MASKS): Likewise.
21426 (POWERPC_MASKS): Likewise.
21427 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
21428 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
21429 code, only testing for DImode being allowed in non-VSX floating
21431 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
21432 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
21434 (rs6000_option_override_internal): Delete -mvsx-small-integer.
21435 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
21436 TARGET_P8_VECTOR test.
21437 (rs6000_secondary_reload_simple_move): Likewise.
21438 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
21439 since TARGET_P9_VECTOR was already tested.
21440 (rs6000_opt_masks): Remove -mvsx-small-integer.
21441 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
21442 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
21444 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
21445 test for TARGET_VEXTRACTUB was used, and that uses
21447 (p9 extract splitter): Likewise.
21448 (vsx_extract_<mode>_di_p9): Likewise.
21449 (vsx_extract_<mode>_store_p9): Likewise.
21450 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
21451 for TARGET_P9_VECTOR was used. Delete code that is now dead with
21452 the elimination of TARGET_VSX_SMALL_INTEGER.
21453 (vsx_extract_<mode>_p8): Likewise.
21454 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
21455 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
21456 (vsx_set_<mode>_p9): Likewise.
21457 (vsx_set_v4sf_p9): Likewise.
21458 (vsx_set_v4sf_p9_zero): Likewise.
21459 (vsx_insert_extract_v4sf_p9): Likewise.
21460 (vsx_insert_extract_v4sf_p9_2): Likewise.
21461 * config/rs6000/rs6000.md (sign extend splitter): Change
21462 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
21463 (floatsi<mode>2_lfiwax_mem): Likewise.
21464 (floatunssi<mode>2_lfiwzx_mem): Likewise.
21465 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
21466 since a test for TARGET_P9_VECTOR was used.
21467 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
21468 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
21469 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
21470 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
21471 TARGET_P8_VECTOR test.
21472 (fix_trunc<mode>si2_stfiwx): Likewise.
21473 (fix_trunc<mode>si2_internal): Likewise.
21474 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
21475 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
21477 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
21478 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
21479 TARGET_P8_VECTOR test.
21480 (fixuns_trunc<mode>si2_stfiwx): Likewise.
21481 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
21482 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
21484 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
21485 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
21486 since a test for TARGET_P9_VECTOR was used.
21487 (splitter for loading small constants): Likewise.
21489 2017-07-26 Andrew Pinski <apinski@cavium.com>
21491 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
21494 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
21497 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
21498 (fp_valid_at): Likewise.
21500 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
21502 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
21503 (qdf24xx_addrcost_table): Likewise.
21504 (cortexa57_tunings): Update to use generic_branch_cost.
21505 (cortexa72_tunings): Likewise.
21506 (cortexa73_tunings): Likewise.
21507 (qdf24xx_tunings): Likewise.
21509 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
21511 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
21512 (thunderx2t99_branch_cost): Likewise.
21513 (cortexa35_tunings): Update to use generic_branch_cost.
21514 (cortexa53_tunings): Likewise.
21515 (cortexa57_tunings): Likewise.
21516 (cortexa72_tunings): Likewise.
21517 (cortexa73_tunings): Likewise.
21518 (thunderx2t99_tunings): Likewise.
21520 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
21522 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
21523 (sparc_option_override): Honour MASK_FSMULD.
21524 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
21525 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
21526 * config/sparc/sparc.opt (mfsmuld): New option.
21527 * doc/invoke.texi (mfsmuld): Document option.
21529 2017-07-26 Marek Polacek <polacek@redhat.com>
21531 PR middle-end/70992
21532 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
21534 2017-07-26 Richard Biener <rguenther@suse.de>
21536 * gimple-match-head.c (do_valueize): Return OP if valueize
21538 (get_def): New helper to get at the def stmt of a SSA name
21539 if valueize allows.
21540 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
21541 do_valueize to get at the def stmt.
21542 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
21544 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
21546 PR middle-end/46932
21547 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
21549 2017-07-26 Martin Liska <mliska@suse.cz>
21552 * function.c (expand_function_start): Make expansion of
21553 nonlocal_goto_save_area after parm_birth_insn.
21555 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
21557 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
21558 from all CPU target flags enable members.
21560 2017-07-26 Richard Biener <rguenther@suse.de>
21562 * genmatch.c (dt_simplify::gen): Make iterator vars const.
21563 (decision_tree::gen): Make 'type' const.
21564 (write_predicate): Likewise.
21566 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
21568 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
21569 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
21570 (rs6000_option_override_internal): Likewise.
21571 (rs6000_expand_vector_set): Likewise.
21572 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
21573 (TARGET_UPPER_REGS_SF): Likewise.
21574 (TARGET_UPPER_REGS_DI): Likewise.
21575 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
21576 (TARGET_DIRECT_MOVE_64BIT): Likewise.
21577 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
21578 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
21579 (Splitters for DI constants in Altivec registers): Likewise.
21580 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
21581 (vsx_set_v4sf_p9): Likewise.
21582 (vsx_set_v4sf_p9_zero): Likewise.
21583 (vsx_insert_extract_v4sf_p9): Likewise.
21584 (vsx_insert_extract_v4sf_p9_2): Likewise.
21586 2017-07-25 Carl Love <cel@us.ibm.com>
21588 * doc/extend.texi: Update the built-in documentation file for the
21589 existing built-in functions
21590 vector signed char vec_cnttz (vector signed char);
21591 vector unsigned char vec_cnttz (vector unsigned char);
21592 vector signed short vec_cnttz (vector signed short);
21593 vector unsigned short vec_cnttz (vector unsigned short);
21594 vector signed int vec_cnttz (vector signed int);
21595 vector unsigned int vec_cnttz (vector unsigned int);
21596 vector signed long long vec_cnttz (vector signed long long);
21597 vector unsigned long long vec_cnttz (vector unsigned long long);
21599 2017-07-25 Andrew Pinski <apinski@cavium.com>
21601 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
21602 accesses where the use is for the first operand of a BIT_INSERT.
21604 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
21607 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
21608 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
21610 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
21612 * config/i386/gstabs.h: Delete.
21613 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
21615 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
21617 * config/i386/i386.c (ix86_decompose_address): Do not check for
21618 register RTX when looking at index_reg or base_reg.
21619 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
21621 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
21623 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
21624 to update EH info here.
21626 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
21628 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
21630 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
21632 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
21634 2017-07-25 Torsten Duwe <duwe@suse.de>
21636 * common.opt: Introduce -fpatchable-function-entry
21637 command line option, and its variables function_entry_patch_area_size
21638 and function_entry_patch_area_start.
21639 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
21640 including a two-value parser.
21641 * target.def (print_patchable_function_entry): New target hook.
21642 * targhooks.h (default_print_patchable_function_entry): New function.
21643 * targhooks.c (default_print_patchable_function_entry): Likewise.
21644 * toplev.c (process_options): Switch off IPA-RA if
21645 patchable function entries are being generated.
21646 * varasm.c (assemble_start_function): Look at the
21647 patchable-function-entry command line switch and current
21648 function attributes and maybe generate NOP instructions by
21649 calling the print_patchable_function_entry hook.
21650 * doc/extend.texi: Document patchable_function_entry attribute.
21651 * doc/invoke.texi: Document -fpatchable_function_entry
21652 command line option.
21653 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
21655 * doc/tm.texi: Re-generate.
21657 2017-07-25 Jakub Jelinek <jakub@redhat.com>
21660 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
21661 TARGET_AVX512DQ rather than TARGET_AVX512BW.
21663 2017-07-25 Tamar Christina <tamar.christina@arm.com>
21665 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
21667 2017-07-25 Richard Biener <rguenther@suse.de>
21669 PR tree-optimization/81455
21670 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
21671 not walk in cycles when looking for guards.
21673 2017-07-25 Richard Biener <rguenther@suse.de>
21675 PR tree-optimization/81529
21676 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
21677 when optimizing backedge uses.
21679 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
21681 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
21683 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
21684 to dl_section_ref. On AIX, append an expression to subtract
21685 the size of the section length to dl_section_ref.
21687 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
21689 * configure.ac: If any of the config.* scripts fail, exit 1.
21690 * configure: Regenerate.
21692 2017-07-25 Richard Biener <rguenther@suse.de>
21694 PR middle-end/81546
21695 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
21696 of immediate uses, be more verbose on errors.
21698 2017-07-25 Richard Biener <rguenther@suse.de>
21700 PR tree-optimization/81510
21701 * tree-vect-loop.c (vect_is_simple_reduction): When the
21702 reduction stmt is not inside the loop bail out.
21704 2017-07-25 Richard Biener <rguenther@suse.de>
21706 PR tree-optimization/81303
21707 * tree-vect-loop-manip.c (vect_loop_versioning): Build
21708 profitability check against LOOP_VINFO_NITERSM1.
21710 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
21712 * domwalk.c (cmp_bb_postorder): Simplify.
21713 (sort_bbs_postorder): New function. Use it...
21714 (dom_walker::walk): ...here to optimize common cases.
21716 2017-07-25 Martin Liska <mliska@suse.cz>
21719 * ipa-visibility.c (function_and_variable_visibility): Make the
21720 redirection just on target that supports aliasing.
21721 Fix GNU coding style.
21723 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
21726 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
21728 * config/arm/rtems.h: Likewise.
21729 * config/bfin/rtems.h: Likewise.
21730 * config/i386/rtemself.h: Likewise.
21731 * config/lm32/rtems.h: Likewise.
21732 * config/m32c/rtems.h: Likewise.
21733 * config/m68k/rtemself.h: Likewise.
21734 * config/microblaze/rtems.h: Likewise.
21735 * config/mips/rtems.h: Likewise.
21736 * config/moxie/rtems.h: Likewise.
21737 * config/nios2/rtems.h: Likewise.
21738 * config/powerpcspe/rtems.h: Likewise.
21739 * config/rs6000/rtems.h: Likewise.
21740 * config/rtems.h: Likewise.
21741 * config/sh/rtems.h: Likewise.
21742 * config/sh/rtemself.h: Likewise.
21743 * config/sparc/rtemself.h: Likewise.
21745 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
21748 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
21749 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
21750 * tree-ssa-structalias.c (alias_get_name): Same.
21752 2017-07-25 Bin Cheng <bin.cheng@arm.com>
21755 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
21756 instructions if no du chain is found.
21758 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
21760 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
21762 2017-07-25 Richard Biener <rguenther@suse.de>
21764 PR middle-end/81505
21765 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
21768 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
21770 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
21771 upper-regs options.
21772 (ISA_2_7_MASKS_SERVER): Likewise.
21773 (ISA_3_0_MASKS_IEEE): Likewise.
21774 (OTHER_P8_VECTOR_MASKS): Likewise.
21775 (OTHER_VSX_VECTOR_MASKS): Likewise.
21776 (POWERPC_MASKS): Likewise.
21777 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
21778 duplicate list of options.
21779 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
21780 explicit -mupper-regs options.
21781 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
21782 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
21783 alias for -mupper-regs-df.
21784 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
21785 (rs6000_init_hard_regno_mode_ok): Likewise.
21786 (rs6000_option_override_internal): Likewise.
21787 (rs6000_opt_masks): Likewise.
21788 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
21789 options in terms of whether -mvsx or -mpower8-vector was used.
21790 (TARGET_UPPER_REGS_DI): Likewise.
21791 (TARGET_UPPER_REGS_SF): Likewise.
21792 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
21793 -mupper-regs-* options.
21795 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
21797 * passes.c (emergency_dump_function): Print some empty lines and a
21798 header before the RTL dump.
21800 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
21802 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
21804 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
21807 * config/aarch64/aarch64.c (aarch64_classify_symbol):
21808 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
21810 2017-07-24 Carl Love <cel@us.ibm.com>
21812 * config/rs6000/rs6000-c.c: Add support for built-in functions
21813 vector float vec_extract_fp32_from_shorth (vector unsigned short);
21814 vector float vec_extract_fp32_from_shortl (vector unsigned short);
21815 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
21816 vec_extract_fp_from_shortl): Add defines for the two builtins.
21817 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
21818 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
21820 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
21821 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
21822 * doc/extend.texi: Update the built-in documentation file for the
21823 new built-in function.
21825 2017-07-24 Jakub Jelinek <jakub@redhat.com>
21828 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
21830 * doc/generic.texi: Likewise.
21831 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
21832 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
21834 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
21836 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
21837 (aarch64_mls_elt_merge<mode>): Likewise.
21839 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
21841 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
21842 having __cxa_atexit.
21844 2017-07-23 Michael Collison <michael.collison@arm.com>
21846 * config/arm/arm.c (arm_option_override): Deprecate
21847 use of -mstructure-size-boundary.
21848 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
21849 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
21851 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21854 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
21855 Reduce cost estimate for direct moves.
21857 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
21860 * config/i386/i386.c (ix86_option_override_internal): Disable
21861 BMI, BMI2 and TBM instructions for -m16.
21863 2017-07-21 Carl Love <cel@us.ibm.com>
21865 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21866 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
21867 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
21868 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
21869 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
21870 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
21871 VMULOSW): New enum "unspec" values.
21872 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
21873 altivec_vmulosw): New patterns.
21874 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
21875 VMULOSW): Add definitions.
21877 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
21879 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
21880 (qdf24xx): Likewise.
21881 * config/aarch64/aarch64-options-extensions.def (rdma); New.
21882 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
21883 (AARCH64_FL_V8_1): Renumber.
21884 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
21885 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
21886 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
21887 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
21888 rdma to feature modifiers list.
21890 2017-07-21 Yury Gribov <tetra2005@gmail.com>
21892 PR middle-end/56727
21893 * ipa-visibility (function_and_variable_visibility): Convert
21894 recursive PLT call to direct call if appropriate.
21896 2017-07-21 Andrew Pinski <apinski@cavium.com>
21898 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
21899 operand 1 to see if the types precision matches.
21900 * fold-const.c (operand_equal_p): Likewise.
21902 2017-07-21 Richard Biener <rguenther@suse.de>
21904 PR tree-optimization/81303
21905 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
21906 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
21907 (vect_peeling_hash_get_lowest_cost): Adjust.
21908 (vect_enhance_data_refs_alignment): Likewise. Use
21909 vect_get_peeling_costs_all_drs to compute the penalty for no
21910 peeling to match up costs.
21912 2017-07-21 Richard Biener <rguenther@suse.de>
21914 PR tree-optimization/81500
21915 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
21916 we didn't identify a reduction path.
21918 2017-07-21 Tom de Vries <tom@codesourcery.com>
21919 Cesar Philippidis <cesar@codesourcery.com>
21921 PR gcov-profile/81442
21922 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
21925 2017-07-21 Tom de Vries <tom@codesourcery.com>
21928 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
21930 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
21931 nvptx_override_options_after_change.
21933 2017-07-21 Ulrich Drepper <drepper@redhat.com>
21935 * dwarf2out.c (output_file_names): Avoid double testing for
21936 dwarf_version >= 5.
21938 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
21940 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
21942 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
21944 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
21946 (try_crossjump_to_edge): Do not punt on partitioned functions.
21948 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
21950 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
21951 Put all BBs reachable only via paths crossing cold region to cold
21953 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
21955 2016-07-21 Richard Biener <rguenther@suse.de>
21957 PR tree-optimization/81303
21958 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
21959 into account prologue and epilogue iterations when raising
21960 min_profitable_iters to sth at least covering one vector iteration.
21962 2017-07-21 Tamar Christina <tamar.christina@arm.com>
21964 * config/arm/arm.c (arm_test_cpu_arch_dat):
21967 2017-07-20 Nathan Sidwell <nathan@acm.org>
21969 Remove TYPE_METHODS.
21970 * tree.h (TYPE_METHODS): Delete.
21971 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
21972 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
21973 (dbxout_type_methods): Scan TYPE_FIELDS.
21974 (dbxout_type): Don't check TYPE_METHODS here.
21975 * function.c (use_register_for_decl): Always ignore register for
21976 class types when not optimizing.
21977 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
21978 * tree.c (free_lang_data_in_type): Stitch out member functions and
21979 templates from TYPE_FIELDS.
21980 (build_distinct_type_copy, verify_type_variant,
21981 verify_type): Member fns are on TYPE_FIELDS.
21982 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
21983 * tree-pretty-print.c (dump_generic_node): Likewise.
21985 2017-07-20 Jakub Jelinek <jakub@redhat.com>
21988 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
21989 V2TImode and V4TImode.
21990 (ix86_expand_vector_extract): Likewise.
21991 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
21992 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
21993 (ssescalarmode): Handle V4TImode and V2TImode.
21994 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
21995 (*vec_extractv2ti, *vec_extractv4ti): New insns.
21996 (VEXTRACTI128_MODE): New mode iterator.
21997 (splitter for *vec_extractv?ti first element): New.
21998 (VEC_INIT_MODE): New mode iterator.
21999 (vec_init<mode>): Consolidate 3 expanders into one using
22000 VEC_INIT_MODE mode iterator.
22002 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
22004 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
22005 non_spilled_static_chain_regno_p.
22007 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
22009 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
22011 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
22013 * bb-reorder.c (connect_traces): Allow copying of blocks within
22016 2017-07-20 Richard Biener <rguenther@suse.de>
22018 * gimple.h (gimple_phi_result): Add gphi * overload.
22019 (gimple_phi_result_ptr): Likewise.
22020 (gimple_phi_arg): Likewise. Adjust index assert to only
22021 allow actual argument accesses rather than all slots available
22023 (gimple_phi_arg_def): Add gphi * overload.
22024 * tree-phinodes.c (make_phi_node): Initialize only actual
22026 (resize_phi_node): Clear memory not covered by old node,
22027 do not initialize excess argument slots.
22028 (reserve_phi_args_for_new_edge): Initialize new argument slot
22031 2017-07-20 Bin Cheng <bin.cheng@arm.com>
22033 PR tree-optimization/81388
22035 2016-07-21 Bin Cheng <bin.cheng@arm.com>
22037 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
22038 by removing computation of may_be_zero.
22040 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
22041 Tom de Vries <tom@codesourcery.com>
22043 PR middle-end/81030
22044 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
22045 when gimple level profile disagrees with what RTL expander did.
22047 2017-07-20 Richard Biener <rguenther@suse.de>
22049 PR tree-optimization/61171
22050 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
22051 (vect_analyze_stmt): Add slp instance parameter.
22052 (vectorizable_reduction): Likewise.
22053 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
22054 (vect_is_simple_reduction): Deal with chains not detected
22055 as SLP reduction chain, specifically not properly associated
22056 chains containing a mix of plus/minus.
22057 (get_reduction_op): Remove.
22058 (get_initial_defs_for_reduction): Simplify, pass in whether
22059 this is a reduction chain, pass in the SLP node for the PHIs.
22060 (vect_create_epilog_for_reduction): Get the SLP instance as
22062 (vectorizable_reduction): Get the SLP instance as arg.
22063 During analysis remember the SLP node with the PHIs in the
22064 instance. Simplify getting at the vectorized reduction PHIs.
22065 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
22066 through SLP instance.
22067 (vect_slp_analyze_operations): Likewise.
22068 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
22069 (vect_transform_stmt): Likewise.
22071 2017-07-20 Tom de Vries <tom@codesourcery.com>
22073 PR tree-optimization/81489
22074 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
22075 read of phi arg location to before loop that modifies phi.
22077 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22079 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
22082 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
22084 PR middle-end/81331
22085 * except.c (execute): Fix ordering issue.
22087 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
22089 PR rtl-optimization/81423
22090 * combine.c (make_compound_operation_int): Don't try to optimize
22091 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
22093 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
22095 PR rtl-optimization/81423
22096 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
22097 with a constant that is -1 in the truncated to mode.
22099 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
22101 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
22102 (determine_unlikely_bbs): ... here.
22103 * predict.h (propagate_unlikely_bbs_forward): Declare.
22104 * cfgexpand.c (pass_expand::execute): Use it.
22105 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
22107 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
22108 propagate_unlikely_bbs_forward.
22110 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
22112 PR middle-end/81331
22113 * except.c (maybe_add_nop_after_section_switch): New function.
22116 2017-07-19 Tom de Vries <tom@codesourcery.com>
22118 * gimple.h (gimple_phi_set_arg): Make assert more strict.
22120 2017-07-19 Tom de Vries <tom@codesourcery.com>
22122 * gimple.h (gimple_phi_arg): Make assert more strict.
22124 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
22126 * config.gcc (powerpc*-*-*): Add mmintrin.h.
22127 * config/rs6000/mmintrin.h: New file.
22128 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
22130 2017-07-19 Jakub Jelinek <jakub@redhat.com>
22132 PR tree-optimization/81346
22133 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
22135 2017-07-19 Tom de Vries <tom@codesourcery.com>
22137 * config/nvptx/nvptx.md (VECIM): Add V2DI.
22139 2017-07-19 Tom de Vries <tom@codesourcery.com>
22141 * config/nvptx/nvptx-modes.def: Add V2DImode.
22142 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
22143 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
22144 (nvptx_output_mov_insn): Handle lack of mov.b128.
22145 (nvptx_print_operand): Handle 'H' and 'L' codes.
22146 (nvptx_vector_mode_supported): Allow V2DImode.
22147 (nvptx_preferred_simd_mode): New function.
22148 (nvptx_data_alignment): New function.
22149 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
22150 nvptx_preferred_simd_mode.
22151 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
22153 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
22155 2017-07-19 Tom de Vries <tom@codesourcery.com>
22157 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
22158 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
22159 (nvptx_vector_mode_supported): New function. Allow V2SImode.
22160 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
22161 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
22162 (mov<VECIM>_insn): New define_insn.
22163 (define_expand "mov<VECIM>): New define_expand.
22165 2017-07-19 Tom de Vries <tom@codesourcery.com>
22167 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
22169 2017-07-19 Jakub Jelinek <jakub@redhat.com>
22171 PR tree-optimization/81346
22172 * fold-const.h (fold_div_compare, range_check_type): Declare.
22173 * fold-const.c (range_check_type): New function.
22174 (build_range_check): Use range_check_type.
22175 (fold_div_compare): No longer static, rewritten into
22176 a match.pd helper function.
22177 (fold_comparison): Don't call fold_div_compare here.
22178 * match.pd (X / C1 op C2): New optimization using fold_div_compare
22179 as helper function.
22181 2017-07-19 Nathan Sidwell <nathan@acm.org>
22183 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
22184 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
22185 * tree.c (find_decls_types_r, verify_type): Use
22186 TYPE_{MIN,MAX}_VALUE_RAW.
22187 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
22188 (hash_tree): Likewise.
22189 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
22191 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
22194 2017-07-18 Tom de Vries <tom@codesourcery.com>
22196 PR middle-end/81464
22197 * omp-expand.c (expand_omp_for_static_chunk): Handle
22198 equal-argument loop exit phi.
22200 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
22203 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
22204 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
22205 operand 2 predicate.
22206 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
22207 operand 2 predicate.
22208 (ror,rol -> rorx splitters): Use const_int_operand as
22209 operand 2 predicate.
22211 2017-06-18 Richard Biener <rguenther@suse.de>
22213 PR tree-optimization/81410
22214 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
22215 the gap in the ! slp_perm SLP case after each group.
22217 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
22219 PR middle-end/81463
22220 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
22223 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
22225 PR middle-end/81462
22226 * predict.c (set_even_probabilities): Cleanup; do not affect
22227 probabilities that are already known.
22228 (combine_predictions_for_bb): Call even when count is set.
22230 2017-07-18 Nathan Sidwell <nathan@acm.org>
22232 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
22235 2017-07-18 Bin Cheng <bin.cheng@arm.com>
22238 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
22239 optimization for loop niter analysis.
22241 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
22244 * config/avr/avr.c (avr_optimize_casesi): Don't use
22245 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
22247 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
22249 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
22250 body_cost_vec from _vect_peel_extended_info.
22251 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
22252 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
22255 2017-07-18 Bin Cheng <bin.cheng@arm.com>
22257 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
22259 2017-07-18 Richard Biener <rguenther@suse.de>
22261 PR tree-optimization/80620
22262 PR tree-optimization/81403
22263 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
22264 info when re-using a VN table entry.
22266 2017-07-18 Richard Biener <rguenther@suse.de>
22268 PR tree-optimization/81418
22269 * tree-vect-loop.c (vectorizable_reduction): Properly compute
22270 vectype_in. Verify that with lane-reducing reduction operations
22271 we have a single def-use cycle.
22273 2017-07-17 Carl Love <cel@us.ibm.com>
22275 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
22277 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22278 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
22279 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
22280 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
22281 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
22282 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
22283 VMULOSW): New enum "unspec" values.
22284 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
22285 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
22286 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
22287 altivec_vmulosw): New patterns.
22288 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
22289 VMULOSW): Add definitions.
22291 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
22293 * config/alpha/alpha.c: Include predict.h.
22295 2017-07-17 Yury Gribov <tetra2005@gmail.com>
22297 * tree-vrp.c (compare_assert_loc): Fix comparison function
22298 to return predictable results.
22300 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
22302 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
22304 (subdi3): Likewise.
22305 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
22306 * doc/invoke.texi (mexpand-adddi): Update text.
22308 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
22310 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
22311 that also clobbers the CC register. The old expand code is moved
22313 (*arc_clzsi2): ... here.
22314 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
22315 the CC register. The old expand code is moved to ...
22316 (arc_ctzsi2): ... here.
22318 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
22320 * config/arc/arc.opt (mindexed-loads): Use initial value
22321 TARGET_INDEXED_LOADS_DEFAULT.
22322 (mauto-modify-reg): Use initial value
22323 TARGET_AUTO_MODIFY_REG_DEFAULT.
22324 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
22325 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
22326 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
22327 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
22329 2017-07-17 Martin Liska <mliska@suse.cz>
22332 * opts.c (finish_options): Do not allow -fgnu-tm
22333 w/ -fsanitize={kernel-,}address. Say sorry.
22335 2017-07-17 Bin Cheng <bin.cheng@arm.com>
22338 * tree-loop-distribution.c (classify_partition): Only assert on
22339 numer of iterations.
22340 (merge_dep_scc_partitions): Delete prameter. Update function call.
22341 (distribute_loop): Remove code handling loop with unknown niters.
22342 (pass_loop_distribution::execute): Skip loop with unknown niters.
22344 2017-07-17 Bin Cheng <bin.cheng@arm.com>
22347 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
22348 function sort_partitions_by_post_order.
22350 2017-07-17 Bin Cheng <bin.cheng@arm.com>
22352 PR tree-optimization/81374
22353 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
22354 the max index of basic blocks, rather than number of basic blocks.
22356 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
22358 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
22360 (arc_legitimate_pic_operand_p): Likewise.
22361 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
22363 (arc_needs_pcl_p): Likewise.
22364 (arc_legitimate_pc_offset_p): Likewise.
22365 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
22366 function is also used in constrains.md.
22367 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
22368 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
22369 PLUS. Only return true/false in known cases, otherwise assert.
22370 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
22371 is already called in arc_legitimate_constant_p.
22372 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
22374 (LEGITIMATE_PIC_OPERAND_P): Use
22375 arc_raw_symbolic_reference_mentioned_p function.
22376 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
22381 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
22382 Andrew Burgess <andrew.burgess@embecosm.com>
22384 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
22385 (arc_return_address_register): New function.
22386 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
22387 (arc_handle_fndecl_attribute): Add naked attribute.
22388 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
22389 (TARGET_WARN_FUNC_RETURN): Likewise.
22390 (arc_allocate_stack_slots_for_args): New function.
22391 (arc_warn_func_return): Likewise.
22392 (machine_function): Change type fn_type.
22393 (arc_compute_function_type): Consider new naked function type,
22394 change function return type.
22395 (arc_must_save_register): Adapt to handle new
22396 arc_compute_function_type's return type.
22397 (arc_expand_prologue): Likewise.
22398 (arc_expand_epilogue): Likewise.
22399 (arc_return_address_regs): Delete.
22400 (arc_return_address_register): New function.
22401 (arc_epilogue_uses): Use above function.
22402 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
22403 (arc_function_type): Change encoding, add naked type.
22404 (ARC_INTERRUPT_P): Change to handle the new encoding.
22405 (ARC_FAST_INTERRUPT_P): Likewise.
22406 (ARC_NORMAL_P): Define.
22407 (ARC_NAKED_P): Likewise.
22408 (arc_compute_function_type): Delete prototype.
22409 * config/arc/arc.md (in_ret_delay_slot): Use
22410 arc_return_address_register function.
22411 (simple_return): Likewise.
22412 (p_return_i): Likewise.
22414 2017-07-17 Jakub Jelinek <jakub@redhat.com>
22416 PR tree-optimization/81428
22417 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
22418 can't be built for those types.
22420 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
22422 Remove stuff dead since r239246.
22424 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
22425 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
22426 (avr_inform_devices): Remove dead stuff.
22428 2017-07-17 Tamar Christina <tamar.christina@arm.com>
22430 * config/arm/arm_neon.h: Fix softp typo.
22432 2017-07-17 Jakub Jelinek <jakub@redhat.com>
22434 PR tree-optimization/81365
22435 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
22436 aggregate moves onto bb predecessor edges, make sure there are no
22437 loads that could alias the lhs in between the start of bb and the
22440 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
22443 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
22444 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
22445 [LSHIFTRT, outer_code = TRUNCATE]: Same.
22447 2017-07-17 Jakub Jelinek <jakub@redhat.com>
22449 PR tree-optimization/81396
22450 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
22451 (init_symbolic_number): Initialize it to 1.
22452 (perform_symbolic_merge): Add n_ops from both operands into the new
22454 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
22455 without base_addr as useless if they need more than one operation.
22456 (bswap_replace): Handle !bswap case for NULL base_addr.
22458 2017-07-17 Tom de Vries <tom@codesourcery.com>
22461 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
22464 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
22466 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
22467 conditional builtin define __FIX_LEON3FT_B2BST.
22469 2017-07-17 Daniel Cederman <cederman@gaisler.com>
22471 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
22472 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
22475 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
22477 PR rtl-optimization/81424
22478 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
22479 to remove potential trapping from operands if -fnon-call-exceptions.
22481 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
22483 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
22484 profile_proability for scalling.
22485 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
22487 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
22489 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
22491 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
22493 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
22494 fixpoint arithmetics.
22496 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
22498 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
22499 fixpoint arithmetics.
22501 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
22503 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
22504 fixpoint arithmetics.
22506 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
22508 * profile-count.h (profile_probability::from_reg_br_prob_note,
22509 profile_probability::to_reg_br_prob_note): New functions.
22510 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
22511 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
22512 * predict.c (probability_reliable_p): Update.
22513 (edge_probability_reliable_p): Update.
22514 (br_prob_note_reliable_p): Update.
22515 (invert_br_probabilities): Update.
22516 (add_reg_br_prob_note): New function.
22517 (combine_predictions_for_insn): Update.
22518 * asan.c (asan_clear_shadow): Update.
22519 * cfgbuild.c (compute_outgoing_frequencies): Update.
22520 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
22521 (update_br_prob_note): Update.
22522 (rtl_verify_edges): Update.
22523 (purge_dead_edges): Update.
22524 (fixup_reorder_chain): Update.
22525 * emit-rtl.c (try_split): Update.
22526 * ifcvt.c (cond_exec_process_insns): Update.
22527 (cond_exec_process_if_block): Update.
22528 (dead_or_predicable): Update.
22529 * internal-fn.c (expand_addsub_overflow): Update.
22530 (expand_neg_overflow): Update.
22531 (expand_mul_overflow): Update.
22532 * loop-doloop.c (doloop_modify): Update.
22533 * loop-unroll.c (compare_and_jump_seq): Update.
22534 * optabs.c (emit_cmp_and_jump_insn_1): Update.
22535 * predict.h: Update.
22536 * reorg.c (mostly_true_jump): Update.
22538 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
22539 * config/alpha/alpha.c (emit_unlikely_jump): Update.
22540 * config/arc/arc.c: (emit_unlikely_jump): Update.
22541 * config/arm/arm.c: (emit_unlikely_jump): Update.
22542 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
22543 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
22544 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
22545 (ix86_print_operand): Update.
22546 (ix86_split_fp_branch): Update.
22547 (predict_jump): Update.
22548 * config/ia64/ia64.c (ia64_print_operand): Update.
22549 * config/mmix/mmix.c (mmix_print_operand): Update.
22550 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
22551 (rs6000_expand_split_stack_prologue): Update.
22552 * config/rs6000/rs6000.c: Update.
22553 * config/s390/s390.c (s390_expand_vec_strlen): Update.
22554 (s390_expand_vec_movstr): Update.
22555 (s390_expand_cs_tdsi): Update.
22556 (s390_expand_split_stack_prologue): Update.
22557 * config/sh/sh.c (sh_print_operand): Update.
22558 (expand_cbranchsi4): Update.
22559 (expand_cbranchdi4): Update.
22560 * config/sparc/sparc.c (output_v9branch): Update.
22561 * config/spu/spu.c (get_branch_target): Update.
22562 (ea_load_store_inline): Update.
22563 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
22564 * config/tilepro/tilepro.c: Update.
22566 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
22568 * gimplify.c (mostly_copy_tree_r): Revert latest change.
22569 (gimplify_save_expr): Likewise.
22571 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
22573 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
22575 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
22577 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
22579 * timevar.def (TV_IPA_FNSUMMARY): Define.
22581 2017-07-16 Daniel Cederman <cederman@gaisler.com>
22583 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
22584 to back store errata sensitive sequence from being generated.
22585 (sqrtdf2_fix): Likewise.
22587 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
22589 * tree-ssa-threadupdate.c (compute_path_counts,
22590 update_joiner_offpath_counts): Use profile_probability.
22592 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
22595 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
22597 * config/arm/arm-c.c (arm_cpu_builtins): Define
22598 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
22600 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
22602 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22603 array entries to represent __ieee128 versions of the
22604 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
22605 scalar_extract_sig, and scalar_insert_exp built-in functions.
22606 (altivec_resolve_overloaded_builtin): Add special case handling
22607 for the __builtin_scalar_insert_exp function, as represented by
22608 the P9V_BUILTIN_VEC_VSIEDP constant.
22609 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
22610 exponent support for __ieee128 argument.
22611 (VSESQP): Add scalar extract signature support for __ieee128
22613 (VSTDCNQP): Add scalar test negative support for __ieee128
22615 (VSIEQP): Add scalar insert exponent support for __int128 argument
22616 with __ieee128 result.
22617 (VSIEQPF): Add scalar insert exponent support for __ieee128
22618 argument with __ieee128 result.
22619 (VSTDCQP): Add scalar test data class support for __ieee128
22621 (VSTDCNQP): Add overload support for scalar test negative with
22622 __ieee128 argument.
22623 (VSTDCQP): Add overload support for scalar test data class
22624 __ieee128 argument.
22625 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
22626 UNSPEC_VSX_SXSIGDP.
22627 (UNSPEC_VSX_SIEXPQP): New constant.
22628 (xsxexpqp): New insn for VSX scalar extract exponent quad
22630 (xsxsigqp): New insn for VSX scalar extract significand quad
22632 (xsiexpqpf): New insn for VSX scalar insert exponent quad
22633 precision with floating point argument.
22634 (xststdcqp): New expand for VSX scalar test data class quad
22636 (xststdcnegqp): New expand for VSX scalar test negative quad
22638 (xststdcqp): New insn to match expansions for VSX scalar test data
22639 class quad precision and VSX scalar test negative quad precision.
22640 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
22641 special case operand checking to enforce that second operand of
22642 VSX scalar test data class with quad precision argument is a 7-bit
22644 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
22645 prototypes and descriptions of __ieee128 versions of
22646 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
22647 scalar_test_data_class, and scalar_test_neg built-in functions.
22649 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22651 PR tree-optimization/81162
22652 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
22653 replace a negate with an add.
22655 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
22657 * doc/invoke.texi (arm/-mcpu): Document +crypto.
22659 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
22661 * config/arm/arm-c.c (arm_cpu_builtins): Define
22662 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
22664 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
22666 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
22667 (armv8-r): Set ARM Cortex-R52 as default CPU.
22668 * config/arm/arm-tables.opt: Regenerate.
22669 * config/arm/arm-tune.md: Regenerate.
22670 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
22672 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
22673 extension for -mcpu=cortex-r52.
22675 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
22677 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
22678 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
22679 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
22680 (fp-armv8): Define it as FP_ARMv8 only.
22681 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
22682 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
22684 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
22685 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
22686 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
22687 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
22688 than TARGET_FPU_ARMV8.
22689 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
22690 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
22691 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
22693 * config/arm/neon.md (neon_vrint): Likewise.
22694 (neon_vcvt): Likewise.
22695 (neon_<fmaxmin_op><mode>): Likewise.
22696 (<fmaxmin><mode>3): Likewise.
22697 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
22698 * config/arm/predicates.md (arm_cond_move_operator): Check against
22699 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
22701 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
22703 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
22704 to top of function.
22706 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22708 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
22709 loop in comment with memset.
22711 2017-07-14 Martin Liska <mliska@suse.cz>
22713 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
22714 * dwarf2out.c (is_java): Remove the function.
22715 (output_pubname): Remove usage of the function.
22716 (lower_bound_default): Remove usage of DW_LANG_Java.
22717 (gen_compile_unit_die): Likewise.
22718 * gcc.c: Remove compiler defaults for .java and .zip files.
22719 * gimple-expr.c (remove_suffix): Change as there's no longer
22720 extension than 4-letter one.
22721 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
22722 (gimplify_save_expr): Likewise.
22723 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
22724 as it's possible even for other languages than Java.
22725 * langhooks.h (struct lang_hooks): Remove Java from a comment.
22726 * lto-opts.c (lto_write_options): Remove reference to Java.
22727 * opts.c (strip_off_ending): Update file extension handling.
22728 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
22729 * tree-eh.c (lower_resx): Likewise.
22730 * tree.c (free_lang_data_in_type): Remove dead code.
22731 (find_decls_types_r): Likewise.
22732 (build_common_builtin_nodes): Remove Java from a comment.
22733 (verify_type): Remove dead code.
22734 * varasm.c (assemble_external): Remove Java from a comment.
22736 2017-07-14 Martin Liska <mliska@suse.cz>
22738 * opts.c (finish_options): Add quotes.
22739 (common_handle_option): Likewise.
22741 2017-07-14 Martin Liska <mliska@suse.cz>
22743 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
22744 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
22745 Remove N_SO_PASCAL.
22746 * dwarf2out.c (lower_bound_default): Do not handle
22748 (gen_compile_unit_die): Likewise.
22749 * gcc.c: Remove default extension binding for GNU Pascal.
22750 * stmt.c: Remove Pascal language from a comment.
22751 * xcoffout.c: Likewise.
22753 2017-07-13 David Malcolm <dmalcolm@redhat.com>
22756 * diagnostic-show-locus.c (fixit_cmp): New function.
22757 (layout::layout): Sort m_fixit_hints.
22758 (column_range::column_range): Assert that the values are valid.
22759 (struct char_span): New struct.
22760 (correction::overwrite): New method.
22761 (struct source_line): New struct.
22762 (line_corrections::add_hint): Add assertions. Reimplement memcpy
22763 calls in terms of classes source_line and char_span, and
22764 correction::overwrite.
22765 (selftest::test_overlapped_fixit_printing_2): New function.
22766 (selftest::diagnostic_show_locus_c_tests): Call it.
22768 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
22770 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
22771 early if there is no lhs.
22773 2017-07-13 Martin Liska <mliska@suse.cz>
22775 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
22776 (gen_reference_type_die): Likewise.
22777 * stor-layout.c: Remove Pascal-related comment.
22779 2017-07-13 Martin Liska <mliska@suse.cz>
22781 * opts.c (finish_options): Add quotes to error messages.
22782 (parse_sanitizer_options): Likewise.
22784 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
22786 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
22788 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
22790 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
22792 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
22794 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
22796 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
22798 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
22801 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
22802 provides the hardware capability bits, define the macro
22803 __BUILTIN_CPU_SUPPORTS__.
22804 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
22805 if GLIBC does not provide the hardware capability bits. Add a
22806 gcc_unreachable call if the built-in cpu function is neither
22807 __builtin_cpu_is nor __builtin_cpu_supports.
22808 (rs6000_get_function_versions_dispatcher): Change the warning
22809 that an old GLIBC is used which does not export the capability
22810 bits to be an error.
22811 * doc/extend.texi (target_clones attribute): Document the
22812 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
22813 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
22814 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
22815 the macros defined by GCC if the newer GLIBC is available.
22817 2017-07-12 Jeff Law <law@redhat.com>
22819 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
22820 remaining includes slightly.
22821 * config/riscv/riscv-builtins.c: Include profile-count.h.
22823 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
22826 * config/avr/avr.c (avr_set_current_function): In diagnostic
22827 messages: Quote keywords and (parts of) identifiers.
22828 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
22831 2017-07-12 Carl Love <cel@us.ibm.com>
22833 * config/rs6000/rs6000-c.c: Add support for built-in functions
22834 vector bool char vec_revb (vector bool char);
22835 vector bool short vec_revb (vector short char);
22836 vector bool int vec_revb (vector bool int);
22837 vector bool long long vec_revb (vector bool long long);
22838 * doc/extend.texi: Update the built-in documentation file for the
22839 new built-in functions.
22841 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22843 * config/s390/s390.md: Remove movcc splitter.
22845 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22847 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
22848 load/store on condition.
22850 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
22853 * config/avr/avr.c (avr_encode_section_info)
22854 [progmem && !TREE_READONLY]: Error if progmem object needs
22857 2017-07-11 Michael Collison <michael.collison@arm.com>
22859 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
22862 2017-07-11 Carl Love <cel@us.ibm.com>
22864 * config/rs6000/rs6000-c.c: Add support for builtins
22865 vector unsigned int vec_parity_lsbb (vector signed int);
22866 vector unsigned int vec_parity_lsbb (vector unsigned int);
22867 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
22868 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
22869 vector unsigned long long vec_parity_lsbb (vector signed long long);
22870 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
22871 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
22872 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
22873 * doc/extend.texi: Update the built-in documentation file for the
22874 new built-in functions.
22876 2017-07-11 David Malcolm <dmalcolm@redhat.com>
22878 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
22879 (layout::m_primary_loc): New field.
22880 (layout::layout): Initialize new field. Move location filtering
22881 logic from here to...
22882 (layout::maybe_add_location_range): ...this new method. Add
22883 support for filtering to just the lines already specified by other
22885 (layout::will_show_line_p): New method.
22886 (gcc_rich_location::add_location_if_nearby): New method.
22887 (selftest::test_add_location_if_nearby): New test function.
22888 (selftest::diagnostic_show_locus_c_tests): Call it.
22889 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
22892 2017-07-11 Tom de Vries <tom@codesourcery.com>
22894 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
22895 (bb_first_real_insn): New function.
22896 (nvptx_single): Add extra initialization of broadcasted condition
22899 2017-07-11 Nathan Sidwell <nathan@acm.org>
22901 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
22903 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
22905 * doc/extend.texi (AVR Function Attributes): Remove weblink to
22906 Binutils doc as TEXI will mess them up.
22907 * doc/invoke.texi (AVR Options): Same here.
22909 2017-07-11 Daniel Cederman <cederman@gaisler.com>
22911 * config/sparc/sparc.opt (mfix-ut700): New option.
22912 (mfix-gr712rc): Likewise.
22913 (sparc_fix_b2bst): New variable.
22914 * doc/invoke.texi (SPARC options): Document them.
22915 (ARM options): Fix warnings.
22916 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
22917 instructions to prevent sequences that can trigger the store-store
22918 errata for certain LEON3FT processors.
22919 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
22920 (sparc_option_override): Set sparc_fix_b2bst appropriately.
22921 * config/sparc/sparc.md (fix_b2bst): New attribute.
22922 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
22924 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
22927 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
22929 (*rsqrtsf2_sse): Ditto.
22931 (div<mode>3): Macroize insn from divdf3 and divsf3
22932 using MODEF mode iterator.
22934 2017-07-10 Martin Sebor <msebor@redhat.com>
22936 PR tree-optimization/80397
22937 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
22938 instead of testing for equality to INTEGER_TYPE.
22940 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
22942 * config.gcc: Remove uclibc from arc target spec.
22944 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
22946 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
22948 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
22951 * lto-wrapper.c (remove_option): New function.
22952 (merge_and_complain): Merge PIC/PIE options more realistically.
22954 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
22956 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
22960 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
22961 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
22962 * config.in: Regenerate.
22963 * configure: Regenerate.
22964 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
22965 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
22966 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
22967 (TARGET_GASISR_PROLOGUES): ...target mask.
22968 * common/config/avr/avr-common.c
22969 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
22970 Set -mgas-isr-prologues.
22971 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
22972 INSERT_PASS_BEFORE for it.
22973 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
22974 * config/avr/avr.c (avr_option_override)
22975 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
22976 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
22977 (avr_attribute_table) <no_gccisr>: Add new function attribute.
22978 (avr_set_current_function) <is_no_gccisr>: Init machine field.
22979 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
22981 (make_avr_pass_pre_proep): New function.
22982 (emit_push_sfr) <treg>: Add argument to function and use it
22983 instead of TMP_REG.
22984 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
22985 and set machine->gasisr.yes.
22986 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
22987 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
22988 __gcc_isr.n_pushed to .L__stack_usage.
22989 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
22990 (avr_asm_final_postscan_insn): ...this new static function.
22991 * config/avr/avr.h (machine_function)
22992 <is_no_gccisr, use_L__stack_usage>: New fields.
22993 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
22994 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
22995 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
22996 (gasisr, *gasisr): New expander and insn.
22997 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
22998 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
22999 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
23001 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
23003 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
23006 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
23008 Move jump-tables out of .text again.
23011 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
23012 (ASM_OUTPUT_ADDR_VEC): New function.
23013 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
23014 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
23015 INSN_ADDRESSes as asm comment.
23016 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
23017 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
23018 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
23019 * config/avr/avr.md (*tablejump): Adjust comment.
23020 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
23021 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
23023 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
23024 (avr_output_addr_vec): New proto.
23025 (avr_log_t) <insn_addresses>: New field.
23027 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
23030 * config/i386/i386.c (ix86_function_arg_advance): Set
23031 outgoing_args_on_stack to true if there are outgoing arguments
23033 (ix86_function_arg): Likewise.
23034 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
23035 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
23036 * config/i386/i386.h (machine_function): Add
23037 outgoing_args_on_stack.
23039 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
23041 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
23042 supporting pthreds.
23043 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
23045 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
23047 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
23048 (REAL_H): Remove $(MACHMODE_H).
23049 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
23051 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
23052 $(MACHMODE_H) and double-int.h.
23053 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
23055 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
23058 2017-07-07 Andrew Pinski <apinski@cavium.com>
23060 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
23061 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
23063 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
23065 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
23066 Add warning if GCC was not configured to link against a GLIBC that
23067 exports the hardware capability bits.
23068 (make_resolver_func): Make resolver function private and not a
23069 COMDAT function. Create the name with clone_function_name instead
23070 of make_unique_name.
23073 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
23074 correct operand in doing the split.
23076 2017-07-07 Carl Love <cel@us.ibm.com>
23078 * config/rs6000/rs6000-c: Add support for built-in function
23079 vector unsigned short vec_pack_to_short_fp32 (vector float,
23081 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
23082 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
23083 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
23084 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
23085 (convert_4f32_8i16): Add define_expand.
23086 * doc/extend.texi: Update the built-in documentation file for the
23087 new built-in function.
23089 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23091 * config/sparc/m8.md: New file.
23092 * config/sparc/sparc.md: Include m8.md.
23094 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23096 * config/sparc/sparc.opt: New option -mvis4b.
23097 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
23098 (sparc_option_override): Handle VIS4B.
23099 (enum sparc_builtins): Define
23100 SPARC_BUILTIN_DICTUNPACK{8,16,32},
23101 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
23102 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
23103 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
23104 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
23105 (check_constant_argument): New function.
23106 (sparc_vis_init_builtins): Define builtins
23107 __builtin_vis_dictunpack{8,16,32},
23108 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
23109 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
23110 __builtin_vis_fpcmpde{8,16,32}shl and
23111 __builtin_vis_fpcmpur{8,16,32}shl.
23112 (sparc_expand_builtin): Check that the constant operands to
23113 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
23114 constant and in range.
23115 * config/sparc/sparc-c.c (sparc_target_macros): Handle
23117 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
23118 (SPARC_IMM5_P): Likewise.
23119 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
23120 (enabled): Handle vis4b.
23121 (UNSPEC_DICTUNPACK): New unspec.
23122 (UNSPEC_FPCMPSHL): Likewise.
23123 (UNSPEC_FPUCMPSHL): Likewise.
23124 (UNSPEC_FPCMPDESHL): Likewise.
23125 (UNSPEC_FPCMPURSHL): Likewise.
23126 (cpu_feature): New CPU feature `vis4b'.
23127 (dictunpack{8,16,32}): New insns.
23128 (FPCSMODE): New mode iterator.
23129 (fpcscond): New code iterator.
23130 (fpcsucond): Likewise.
23131 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
23132 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
23133 (fpcmpde{8,16,32}{si,di}shl): Likewise.
23134 (fpcmpur{8,16,32}{si,di}shl): Likewise.
23135 * config/sparc/constraints.md: Define constraints `q' for unsigned
23136 2-bit integer constants and `t' for unsigned 5-bit integer
23138 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
23140 (imm5_operand_dictunpack16): Likewise.
23141 (imm5_operand_dictunpack32): Likewise.
23142 (imm2_operand): Likewise.
23143 * doc/invoke.texi (SPARC Options): Document -mvis4b.
23144 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
23145 ditunpack* and fpcmp*shl builtins.
23147 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23149 * config.gcc: Handle m8 in --with-{cpu,tune} options.
23150 * config.in: Add HAVE_AS_SPARC6 define.
23151 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
23153 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
23155 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
23156 (CPP_CPU_SPEC): Handle m8.
23157 (ASM_CPU_SPEC): Likewise.
23158 * config/sparc/sparc-opts.h (enum processor_type): Add
23160 * config/sparc/sparc.c (m8_costs): New struct.
23161 (sparc_option_override): Handle TARGET_CPU_m8.
23162 (sparc32_initialize_trampoline): Likewise.
23163 (sparc64_initialize_trampoline): Likewise.
23164 (sparc_issue_rate): Likewise.
23165 (sparc_register_move_cost): Likewise.
23166 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
23167 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
23168 (ASM_CPU64_DEFAULT_SPEC): Likewise.
23169 (CPP_CPU_SPEC): Handle M8.
23170 (ASM_CPU_SPEC): Likewise.
23171 (AS_M8_FLAG): Define.
23172 * config/sparc/sparc.md: Add m8 to the cpu attribute.
23173 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
23174 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
23176 * configure: Regenerate.
23177 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
23180 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23182 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
23184 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
23185 ("*movdi_insn_sp32"): Do not set v3pipe.
23186 ("*movsi_insn"): Likewise.
23187 ("*movdi_insn_sp64"): Likewise.
23188 ("*movsf_insn"): Likewise.
23189 ("*movdf_insn_sp32"): Likewise.
23190 ("*movdf_insn_sp64"): Likewise.
23191 ("*zero_extendsidi2_insn_sp64"): Likewise.
23192 ("*sign_extendsidi2_insn"): Likewise.
23193 ("*mov<VM32:mode>_insn"): Likewise.
23194 ("*mov<VM64:mode>_insn_sp64"): Likewise.
23195 ("*mov<VM64:mode>_insn_sp32"): Likewise.
23196 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
23197 ("<vlop:code><VL:mode>3"): Likewise.
23198 ("*not_<vlop:code><VL:mode>3"): Likewise.
23199 ("*nand<VL:mode>_vis"): Likewise.
23200 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
23201 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
23202 ("one_cmpl<VL:mode>2"): Likewise.
23203 ("faligndata<VM64:mode>_vis"): Likewise.
23204 ("alignaddrsi_vis"): Likewise.
23205 ("alignaddrdi_vis"): Likweise.
23206 ("alignaddrlsi_vis"): Likewise.
23207 ("alignaddrldi_vis"): Likewise.
23208 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
23209 ("bmaskdi_vis"): Likewise.
23210 ("bmasksi_vis"): Likewise.
23211 ("bshuffle<VM64:mode>_vis"): Likewise.
23212 ("cmask8<P:mode>_vis"): Likewise.
23213 ("cmask16<P:mode>_vis"): Likewise.
23214 ("cmask32<P:mode>_vis"): Likewise.
23215 ("pdistn<P:mode>_vis"): Likewise.
23216 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
23218 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23220 * config/sparc/sparc.md ("subtype"): New insn attribute.
23221 ("*wrgsr_sp64"): Set insn subtype.
23222 ("*rdgsr_sp64"): Likewise.
23223 ("alignaddrsi_vis"): Likewise.
23224 ("alignaddrdi_vis"): Likewise.
23225 ("alignaddrlsi_vis"): Likewise.
23226 ("alignaddrldi_vis"): Likewise.
23227 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
23228 ("fexpand_vis"): Likewise.
23229 ("fpmerge_vis"): Likewise.
23230 ("faligndata<VM64:mode>_vis"): Likewise.
23231 ("bshuffle<VM64:mode>_vis"): Likewise.
23232 ("cmask8<P:mode>_vis"): Likewise.
23233 ("cmask16<P:mode>_vis"): Likewise.
23234 ("cmask32<P:mode>_vis"): Likewise.
23235 ("fchksm16_vis"): Likewise.
23236 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
23237 ("fmean16_vis"): Likewise.
23238 ("fp<plusminus_insn>64_vis"): Likewise.
23239 ("<plusminus_insn>v8qi3"): Likewise.
23240 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
23241 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
23242 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
23243 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
23244 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
23245 ("*movqi_insn"): Likewise.
23246 ("*movhi_insn"): Likewise.
23247 ("*movsi_insn"): Likewise.
23248 ("movsi_pic_gotdata_op"): Likewise.
23249 ("*movdi_insn_sp32"): Likewise.
23250 ("*movdi_insn_sp64"): Likewise.
23251 ("movdi_pic_gotdata_op"): Likewise.
23252 ("*movsf_insn"): Likewise.
23253 ("*movdf_insn_sp32"): Likewise.
23254 ("*movdf_insn_sp64"): Likewise.
23255 ("*zero_extendhisi2_insn"): Likewise.
23256 ("*zero_extendqihi2_insn"): Likewise.
23257 ("*zero_extendqisi2_insn"): Likewise.
23258 ("*zero_extendqidi2_insn"): Likewise.
23259 ("*zero_extendhidi2_insn"): Likewise.
23260 ("*zero_extendsidi2_insn_sp64"): Likewise.
23261 ("ldfsr"): Likewise.
23262 ("prefetch_64"): Likewise.
23263 ("prefetch_32"): Likewise.
23264 ("tie_ld32"): Likewise.
23265 ("tie_ld64"): Likewise.
23266 ("*tldo_ldub_sp32"): Likewise.
23267 ("*tldo_ldub1_sp32"): Likewise.
23268 ("*tldo_ldub2_sp32"): Likewise.
23269 ("*tldo_ldub_sp64"): Likewise.
23270 ("*tldo_ldub1_sp64"): Likewise.
23271 ("*tldo_ldub2_sp64"): Likewise.
23272 ("*tldo_ldub3_sp64"): Likewise.
23273 ("*tldo_lduh_sp32"): Likewise.
23274 ("*tldo_lduh1_sp32"): Likewise.
23275 ("*tldo_lduh_sp64"): Likewise.
23276 ("*tldo_lduh1_sp64"): Likewise.
23277 ("*tldo_lduh2_sp64"): Likewise.
23278 ("*tldo_lduw_sp32"): Likewise.
23279 ("*tldo_lduw_sp64"): Likewise.
23280 ("*tldo_lduw1_sp64"): Likewise.
23281 ("*tldo_ldx_sp64"): Likewise.
23282 ("*mov<VM32:mode>_insn"): Likewise.
23283 ("*mov<VM64:mode>_insn_sp64"): Likewise.
23284 ("*mov<VM64:mode>_insn_sp32"): Likewise.
23286 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23288 * config/sparc/sparc.md ("type"): New insn type viscmp.
23289 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
23291 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
23292 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
23293 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
23294 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
23296 ("n7_vis_logical_11cycle"): Likewise.
23297 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
23298 * config/sparc/niagara2.md ("niag3_vis": Likewise.
23299 * config/sparc/niagara.md ("niag_vis"): Likewise.
23300 * config/sparc/ultra3.md ("us3_fga"): Likewise.
23301 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
23303 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23305 * config/sparc/sparc.md: New instruction type `bmask'.
23306 (bmaskdi_vis): Use the `bmask' type.
23307 (bmasksi_vis): Likewise.
23308 * config/sparc/ultra3.md (us3_array): Likewise.
23309 * config/sparc/niagara7.md (n7_array): Likewise.
23310 * config/sparc/niagara4.md (n4_array): Likewise.
23311 * config/sparc/niagara2.md (niag2_vis): Likewise.
23312 (niag3_vis): Likewise.
23313 * config/sparc/niagara.md (niag_vis): Likewise.
23315 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
23317 * ipa-comdats.c: Remove optimize check from gate.
23318 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
23319 for functions not optimized.
23320 (ipa_fn_summary_read): Skip optimize check.
23321 (ipa_fn_summary_write): Likewise.
23322 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
23324 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
23326 (can_inline_edge_p): Check flag_pcc_struct_return for match.
23327 (check_callers): Give up on caller which is not optimized.
23328 (inline_small_functions): Likewise.
23329 (ipa_inline): Do not give up when not optimizing.
23330 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
23331 away unoptimizes cdtors.
23332 (whole_program_function_and_variable_visibility): Do
23333 ipa_discover_readonly_nonaddressable_vars in LTO mode.
23334 * ipa.c (process_references): Do not check optimize.
23335 (symbol_table::remove_unreachable_nodes): Update optimize check.
23336 (set_writeonly_bit): Update optimize check.
23337 (pass_ipa_cdtor_merge::gate): Do not check optimize.
23338 (pass_ipa_single_use::gate): Remove.
23340 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23342 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
23343 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
23344 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
23345 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
23346 permute_load, permute_store, adjust_extract, adjust_splat,
23347 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
23348 replace_swap_with_copy, dump_swap_insn_table,
23349 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
23350 recombine_lvx_pattern, recombine_stvx_pattern,
23351 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
23352 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
23353 to file rs6000-p8swap.c.
23354 * config/rs6000/rs6000-p8swap.c: New file.
23355 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
23356 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
23357 and rs6000*-*-* targets.
23359 2017-07-06 David Malcolm <dmalcolm@redhat.com>
23361 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
23363 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
23365 * lto-wrapper.c (merge_and_complain): Do not merge
23366 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
23367 fsigned_zeros, ftrapping_math, fwrapv.
23368 (append_compiler_options): Do not track these options.
23369 (append_linker_options): Likewie
23371 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
23373 * cgraphunit.c (cgraph_node::finalize_function): When
23374 !flag_toplevel_reorde set no_reorder flag.
23375 (varpool_node::finalize_decl): Likewise.
23376 (symbol_table::compile): Drop no toplevel reorder path.
23378 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
23380 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
23381 edges; zero probability is not better than uninitialized.
23383 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
23385 * asan.h (asan_sanitize_allocas_p): Declare.
23386 * asan.c (asan_sanitize_allocas_p): New function.
23387 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
23388 (handle_builtin_alloca): Likewise.
23389 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
23390 if !asan_sanitize_allocas_p.
23391 * params.def (asan-instrument-allocas): Add new option.
23392 * params.h (ASAN_PROTECT_ALLOCAS): Define.
23393 * opts.c (common_handle_option): Disable allocas sanitization for
23396 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
23398 * asan.c: Include gimple-fold.h.
23399 (get_last_alloca_addr): New function.
23400 (handle_builtin_stackrestore): Likewise.
23401 (handle_builtin_alloca): Likewise.
23402 (asan_emit_allocas_unpoison): Likewise.
23403 (get_mem_refs_of_builtin_call): Add new parameter, remove const
23404 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
23405 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
23406 (instrument_builtin_call): Pass gimple iterator to
23407 get_mem_refs_of_builtin_call.
23408 (last_alloca_addr): New global.
23409 * asan.h (asan_emit_allocas_unpoison): Declare.
23410 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
23411 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
23412 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
23413 if function calls alloca.
23414 * gimple-fold.c (replace_call_with_value): Remove static keyword.
23415 * gimple-fold.h (replace_call_with_value): Declare.
23416 * internal-fn.c: Include asan.h.
23417 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
23418 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
23420 2017-07-06 David Malcolm <dmalcolm@redhat.com>
23422 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
23423 (C_SELFTEST_FLAGS): New.
23424 (CPP_SELFTEST_FLAGS): New.
23425 (SELFTEST_DEPS): New, from deps of s-selftest.
23426 (C_SELFTEST_DEPS): New, from deps of s-selftest.
23427 (CPP_SELFTEST_DEPS): New.
23428 (selftest): Add dependency on s-selftest-c++.
23429 (s-selftest): Rename to...
23430 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
23431 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
23432 than SELFTEST_FLAGS.
23433 (selftest-gdb): Rename to...
23434 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
23436 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
23437 (selftest-valgrind): Rename to...
23438 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
23440 (selftest-valgrind): Reintroduce as an alias for
23441 selftest-c-valgrind.
23442 (s-selftest-c++): New.
23443 (selftest-c++-gdb): New.
23444 (selftest-c++-valgrind): New.
23446 2017-07-06 Olivier Hainque <hainque@adacore.com>
23448 * gcc.c (process_command): When deciding if undefined variables
23449 should be ignored when processing specs, accept "gcc -v" as well.
23451 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
23453 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
23454 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
23456 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
23458 * config/arm/arm-cpus.in (armv8-r): Add new entry.
23459 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
23460 * config/arm/arm-tables.opt: Regenerate.
23461 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
23463 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
23465 2017-07-06 Carl Love <cel@us.ibm.com>
23467 * ChangeLog: Clean up from mid air collision
23469 2017-07-06 Carl Love <cel@us.ibm.com>
23471 * config/rs6000/rs6000-c.c: Add support for built-in functions
23472 vector signed int vec_subc (vector signed int, vector signed int);
23473 vector signed __int128 vec_subc (vector signed __int128,
23474 vector signed __int128);
23475 vector unsigned __int128 vec_subc (vector unsigned __int128,
23476 vector unsigned __int128);
23477 vector signed int vec_sube (vector signed int, vector signed int,
23478 vector signed int);
23479 vector unsigned int vec_sube (vector unsigned int,
23480 vector unsigned int,
23481 vector unsigned int);
23482 vector signed __int128 vec_sube (vector signed __int128,
23483 vector signed __int128,
23484 vector signed__int128);
23485 vector unsigned __int128 vec_sube (vector unsigned __int128,
23486 vector unsigned __int128,
23487 vector unsigned __int128);
23488 vector signed int vec_subec (vector signed int, vector signed int,
23489 vector signed int);
23490 vector unsigned int vec_subec (vector unsigned int,
23491 vector unsigned int,
23492 vector unsigned int);
23493 vector signed __int128 vec_subec (vector signed __int128,
23494 vector signed __int128,
23495 vector signed__int128);
23496 vector unsigned __int128 vec_subec (vector unsigned __int128,
23497 vector unsigned __int128,
23498 vector unsigned __int128);
23499 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
23500 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
23501 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
23502 BU_ALTIVEC_OVERLOAD_X definitions.
23503 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
23504 * doc/extend.texi: Update the built-in documentation file for the new
23505 built-in functions.
23507 2017-07-06 David Malcolm <dmalcolm@redhat.com>
23510 * diagnostic-show-locus.c (layout::layout): Use start and finish
23511 spelling location for the start and finish of each range.
23512 * genmatch.c (linemap_client_expand_location_to_spelling_point):
23513 Add unused aspect param.
23514 * input.c (expand_location_1): Add "aspect" param, and use it
23515 to access the correct part of the location.
23516 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
23518 (expand_location_to_spelling_point): Likewise.
23519 (linemap_client_expand_location_to_spelling_point): Add "aspect"
23520 param, and pass it to expand_location_1.
23522 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
23524 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
23525 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
23526 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
23527 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
23528 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
23529 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
23530 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
23531 _mm_maskz_getmant_ss): New intrinsics.
23532 (__builtin_ia32_getexpss128_mask): Changed to ...
23533 __builtin_ia32_getexpss128_round ... this.
23534 (__builtin_ia32_getexpsd128_mask): Changed to ...
23535 __builtin_ia32_getexpsd128_round ... this.
23536 * config/i386/i386-builtin-types.def
23537 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
23538 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
23539 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
23540 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
23541 __builtin_ia32_getmantss_mask_round): New builtins.
23542 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
23543 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
23544 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
23545 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
23546 * config/i386/sse.md
23547 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
23548 avx512f_sgetexp<mode><mask_scalar_name>
23549 <round_saeonly_scalar_name> ... this.
23550 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
23551 %0, %1, %2<round_saeonly_op3>}): Changed to ...
23552 vgetexp<ssescalarmodesuffix>
23553 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
23554 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
23555 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
23556 avx512f_vgetmant<mode><mask_scalar_name>
23557 <round_saeonly_scalar_name> ... this.
23558 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
23559 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
23560 vgetmant<ssescalarmodesuffix>
23561 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
23562 %0<mask_scalar_operand4>, %1, %2
23563 <round_saeonly_scalar_mask_op4>, %3} ... this.
23564 * config/i386/subst.md (mask_scalar_operand4,
23565 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
23566 round_saeonly_scalar_nimm_predicate): New subst attributes.
23568 2017-07-06 Julia Koval <julia.koval@intel.com>
23570 * config/i386/i386.c (ix86_erase_embedded_rounding):
23571 Remove code for old rounding pattern.
23573 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
23575 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
23577 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
23579 * doc/sourcebuild.texi (Test Directives, Variants of
23580 dg-require-support): Add documentation for dg-require-stack-check.
23582 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
23584 * config/i386/subst.md (mask_scalar, round_scalar,
23585 round_saeonly_scalar): New meta-templates.
23586 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
23587 round_scalar_mask_operand3, round_scalar_mask_op3,
23588 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
23589 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
23590 round_saeonly_scalar_constraint,
23591 round_saeonly_scalar_prefix): New subst attribute.
23592 * config/i386/sse.md
23593 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
23594 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
23595 <round_scalar_name> ... this.
23596 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
23597 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
23598 <round_scalar_name> ... this.
23599 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
23600 <sse>_vm<code><mode>3<mask_scalar_name>
23601 <round_saeonly_scalar_name> ... this.
23602 (v<plusminus_mnemonic><ssescalarmodesuffix>
23603 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
23604 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
23605 v<plusminus_mnemonic><ssescalarmodesuffix>
23606 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
23607 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
23608 (v<multdiv_mnemonic><ssescalarmodesuffix>
23609 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
23610 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
23611 v<multdiv_mnemonic><ssescalarmodesuffix>
23612 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
23613 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
23614 (v<maxmin_float><ssescalarmodesuffix>
23615 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
23616 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
23617 v<maxmin_float><ssescalarmodesuffix>
23618 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
23619 %0<mask_scalar_operand3>, %1, %<iptr>2
23620 <round_saeonly_scalar_mask_op3>} ... this.
23622 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
23624 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
23625 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
23627 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
23628 Alan Hayward <alan.hayward@arm.com>
23629 David Sherwood <david.sherwood@arm.com>
23631 * combine.c (simplify_if_then_else): Remove "enum" before
23633 * compare-elim.c (can_eliminate_compare): Likewise.
23634 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
23636 (aarch64_lookup_simd_builtin_type): Likewise.
23637 (aarch64_simd_builtin_type): Likewise.
23638 (aarch64_init_simd_builtin_types): Likewise.
23639 (aarch64_simd_expand_args): Likewise.
23640 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
23642 (aarch64_reverse_mask): Likewise.
23643 (aarch64_simd_emit_reg_reg_move): Likewise.
23644 (aarch64_gen_adjusted_ldpstp): Likewise.
23645 (aarch64_ccmp_mode_to_code): Likewise.
23646 (aarch64_operands_ok_for_ldpstp): Likewise.
23647 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
23648 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
23650 (aarch64_min_divisions_for_recip_mul): Likewise.
23651 (aarch64_reassociation_width): Likewise.
23652 (aarch64_get_condition_code_1): Likewise.
23653 (aarch64_simd_emit_reg_reg_move): Likewise.
23654 (aarch64_simd_attr_length_rglist): Likewise.
23655 (aarch64_reverse_mask): Likewise.
23656 (aarch64_operands_ok_for_ldpstp): Likewise.
23657 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
23658 (aarch64_gen_adjusted_ldpstp): Likewise.
23659 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
23661 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
23662 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
23663 (arm_lookup_simd_builtin_type): Likewise.
23664 (arm_simd_builtin_type): Likewise.
23665 (arm_init_simd_builtin_types): Likewise.
23666 (arm_expand_builtin_args): Likewise.
23667 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
23668 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
23669 (ft32_setup_incoming_varargs): Likewise.
23670 (ft32_function_arg): Likewise.
23671 (ft32_function_arg_advance): Likewise.
23672 (ft32_pass_by_reference): Likewise.
23673 (ft32_arg_partial_bytes): Likewise.
23674 (ft32_valid_pointer_mode): Likewise.
23675 (ft32_addr_space_pointer_mode): Likewise.
23676 (ft32_addr_space_legitimate_address_p): Likewise.
23677 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
23679 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
23680 (ix86_emit_outlined_ms2sysv_restore): Likewise.
23681 (iamcu_alignment): Likewise.
23682 (canonicalize_vector_int_perm): Likewise.
23683 (ix86_noce_conversion_profitable_p): Likewise.
23684 (ix86_mpx_bound_mode): Likewise.
23685 (ix86_operands_ok_for_move_multiple): Likewise.
23686 * config/microblaze/microblaze-protos.h
23687 (microblaze_expand_conditional_branch_reg): Likewise.
23688 * config/microblaze/microblaze.c
23689 (microblaze_expand_conditional_branch_reg): Likewise.
23690 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
23692 (rs6000_reassociation_width): Likewise.
23693 (rs6000_invalid_binary_op): Likewise.
23694 (fusion_p9_p): Likewise.
23695 (emit_fusion_p9_load): Likewise.
23696 (emit_fusion_p9_store): Likewise.
23697 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
23699 (riscv_hard_regno_mode_ok_p): Likewise.
23700 (riscv_address_insns): Likewise.
23701 (riscv_split_symbol): Likewise.
23702 (riscv_legitimize_move): Likewise.
23703 (riscv_function_value): Likewise.
23704 (riscv_hard_regno_nregs): Likewise.
23705 (riscv_expand_builtin): Likewise.
23706 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
23707 (riscv_build_integer): Likewise.
23708 (riscv_split_integer): Likewise.
23709 (riscv_legitimate_constant_p): Likewise.
23710 (riscv_cannot_force_const_mem): Likewise.
23711 (riscv_regno_mode_ok_for_base_p): Likewise.
23712 (riscv_valid_base_register_p): Likewise.
23713 (riscv_valid_offset_p): Likewise.
23714 (riscv_valid_lo_sum_p): Likewise.
23715 (riscv_classify_address): Likewise.
23716 (riscv_legitimate_address_p): Likewise.
23717 (riscv_address_insns): Likewise.
23718 (riscv_load_store_insns): Likewise.
23719 (riscv_force_binary): Likewise.
23720 (riscv_split_symbol): Likewise.
23721 (riscv_force_address): Likewise.
23722 (riscv_legitimize_address): Likewise.
23723 (riscv_move_integer): Likewise.
23724 (riscv_legitimize_const_move): Likewise.
23725 (riscv_legitimize_move): Likewise.
23726 (riscv_address_cost): Likewise.
23727 (riscv_subword): Likewise.
23728 (riscv_output_move): Likewise.
23729 (riscv_canonicalize_int_order_test): Likewise.
23730 (riscv_emit_int_order_test): Likewise.
23731 (riscv_function_arg_boundary): Likewise.
23732 (riscv_pass_mode_in_fpr_p): Likewise.
23733 (riscv_pass_fpr_single): Likewise.
23734 (riscv_pass_fpr_pair): Likewise.
23735 (riscv_get_arg_info): Likewise.
23736 (riscv_function_arg): Likewise.
23737 (riscv_function_arg_advance): Likewise.
23738 (riscv_arg_partial_bytes): Likewise.
23739 (riscv_function_value): Likewise.
23740 (riscv_pass_by_reference): Likewise.
23741 (riscv_setup_incoming_varargs): Likewise.
23742 (riscv_print_operand): Likewise.
23743 (riscv_elf_select_rtx_section): Likewise.
23744 (riscv_save_restore_reg): Likewise.
23745 (riscv_for_each_saved_reg): Likewise.
23746 (riscv_register_move_cost): Likewise.
23747 (riscv_hard_regno_mode_ok_p): Likewise.
23748 (riscv_hard_regno_nregs): Likewise.
23749 (riscv_class_max_nregs): Likewise.
23750 (riscv_memory_move_cost): Likewise.
23751 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
23752 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
23753 (rl78_addr_space_address_mode): Likewise.
23754 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
23756 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
23757 (rs6000_reassociation_width): Likewise.
23758 (rs6000_invalid_binary_op): Likewise.
23759 (fusion_p9_p): Likewise.
23760 (emit_fusion_p9_load): Likewise.
23761 (emit_fusion_p9_store): Likewise.
23762 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
23763 (ok_for_simple_move_operands): Likewise.
23764 (ok_for_simple_move_strict_operands): Likewise.
23765 (ok_for_simple_arith_logic_operands): Likewise.
23766 (visium_legitimize_reload_address): Likewise.
23767 (visium_select_cc_mode): Likewise.
23768 (output_cbranch): Likewise.
23769 (visium_split_double_move): Likewise.
23770 (visium_expand_copysign): Likewise.
23771 (visium_expand_int_cstore): Likewise.
23772 (visium_expand_fp_cstore): Likewise.
23773 * config/visium/visium.c (visium_pass_by_reference): Likewise.
23774 (visium_function_arg): Likewise.
23775 (visium_function_arg_advance): Likewise.
23776 (visium_libcall_value): Likewise.
23777 (visium_setup_incoming_varargs): Likewise.
23778 (visium_legitimate_constant_p): Likewise.
23779 (visium_legitimate_address_p): Likewise.
23780 (visium_legitimize_address): Likewise.
23781 (visium_secondary_reload): Likewise.
23782 (visium_register_move_cost): Likewise.
23783 (visium_memory_move_cost): Likewise.
23784 (prepare_move_operands): Likewise.
23785 (ok_for_simple_move_operands): Likewise.
23786 (ok_for_simple_move_strict_operands): Likewise.
23787 (ok_for_simple_arith_logic_operands): Likewise.
23788 (visium_function_value_1): Likewise.
23789 (rtx_ok_for_offset_p): Likewise.
23790 (visium_legitimize_reload_address): Likewise.
23791 (visium_split_double_move): Likewise.
23792 (visium_expand_copysign): Likewise.
23793 (visium_expand_int_cstore): Likewise.
23794 (visium_expand_fp_cstore): Likewise.
23795 (visium_split_cstore): Likewise.
23796 (visium_select_cc_mode): Likewise.
23797 (visium_split_cbranch): Likewise.
23798 (output_cbranch): Likewise.
23799 (visium_print_operand_address): Likewise.
23800 * expmed.c (flip_storage_order): Likewise.
23801 * expmed.h (emit_cstore): Likewise.
23802 (flip_storage_order): Likewise.
23803 * genrecog.c (validate_pattern): Likewise.
23804 * hsa-gen.c (gen_hsa_addr): Likewise.
23805 * internal-fn.c (expand_arith_overflow): Likewise.
23806 * ira-color.c (allocno_copy_cost_saving): Likewise.
23807 * lra-assigns.c (find_hard_regno_for_1): Likewise.
23808 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
23809 (process_invariant_for_inheritance): Likewise.
23810 * lra-eliminations.c (move_plus_up): Likewise.
23811 * omp-low.c (lower_oacc_reductions): Likewise.
23812 * simplify-rtx.c (simplify_subreg): Likewise.
23813 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
23814 (TARGET_CHKP_BOUND_MODE): Likewise..
23815 * targhooks.c (default_chkp_bound_mode): Likewise.
23816 (default_setup_incoming_vararg_bounds): Likewise.
23817 * targhooks.h (default_chkp_bound_mode): Likewise.
23818 (default_setup_incoming_vararg_bounds): Likewise.
23819 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
23820 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
23821 (have_whole_vector_shift): Likewise.
23822 * tree-vect-stmts.c (vectorizable_load): Likewise.
23823 * doc/tm.texi: Regenerate.
23825 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
23827 Graceful degrade if Binutils PR21472 is not available.
23830 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
23831 .rodata in flash test fails.
23832 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
23833 * confgure: Regenerate.
23834 * config.in: Regenerate.
23835 * config/avr/avr.c (avr_asm_named_section)
23836 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
23837 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
23838 (avr_asm_init_sections): Same.
23840 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
23842 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
23843 (fma<VH:mode>4_intrinsic): Likewise.
23844 (*fmsub<VCVTF:mode>4): Likewise.
23845 (*fmsub<VH:mode>4_intrinsic): Likewise.
23847 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
23850 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
23851 Don't depend on "optimize > 0".
23852 (out_movhi_r_mr, out_movqi_mr_r): Same.
23853 (out_movhi_mr_r, out_movqi_r_mr): Same.
23854 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
23855 io_address_operand on "optimize > 0".
23857 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23859 * tree-loop-distribution.c: Add general explanantion on the pass.
23860 (generate_loops_for_partition): Mark distributed loop.
23861 (pg_add_dependence_edges): New parameter. Handle alias data
23862 dependence specially and record it in the parameter if asked.
23863 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
23864 (init_partition_graph_vertices, add_partition_graph_edge): New.
23865 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
23866 (free_partition_graph_vdata, build_partition_graph): New.
23867 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
23868 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
23869 (data_ref_segment_size, latch_dominated_by_data_ref): New.
23870 (compute_alias_check_pairs, version_loop_by_alias_check): New.
23871 (version_for_distribution_p, finalize_partitions): New.
23872 (distribute_loop): Handle alias data dependence specially. Factor
23873 out loop fusion code as functions and call these functions.
23875 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23877 * tree-loop-distribution.c (classify_partition): New parameter and
23878 better handle reduction statement.
23879 (rdg_build_partitions): Revise comment.
23880 (distribute_loop): Compute statements in all partitions and pass it
23881 to classify_partition.
23883 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23885 * tree-loop-distribution.c (enum partition_type): New.
23886 (struct partition): New field type.
23887 (partition_merge_into): Add parameter. Update partition type.
23888 (data_dep_in_cycle_p, update_type_for_merge): New functions.
23889 (build_rdg_partition_for_vertex): Compute partition type.
23890 (rdg_build_partitions): Dump partition type.
23891 (distribute_loop): Update calls to partition_merge_into.
23893 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23895 * tree-loop-distribution.c (struct ddr_hasher): New.
23896 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
23898 (classify_partition): Call get_data_dependence.
23899 (pg_add_dependence_edges): Ditto.
23900 (distribute_loop): Release data dependence hash table.
23902 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23904 * tree-loop-distribution.c (ref_base_address): Delete.
23905 (similar_memory_accesses): Rename ...
23906 (share_memory_accesses): ... to this. Check if partitions access
23907 the same memory reference.
23908 (distribute_loop): Call share_memory_accesses.
23910 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23912 * tree-loop-distribution.c (struct partition): New field recording
23913 its data reference.
23914 (partition_alloc, partition_free): Init and release data refs.
23915 (partition_merge_into): Merge data refs.
23916 (build_rdg_partition_for_vertex): Collect data refs for partition.
23917 (pg_add_dependence_edges): Change parameters from vector to bitmap.
23919 (distribute_loop): Remve data refs from vertice data of partition
23922 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23924 * tree-loop-distribution.c (params.h): Include header file.
23925 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
23926 (datarefs_vec): New global var.
23927 (create_rdg_vertices): Use datarefs_vec directly.
23928 (free_rdg): Don't free data references.
23929 (build_rdg): Update use. Don't free data references.
23930 (distribute_loop): Compute global variable for data references.
23931 Bail out if there are too many data references.
23933 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23935 * tree-loop-distribution.c (loop_nest): New global var.
23936 (build_rdg): Use loop directly, rather than loop nest.
23937 (pg_add_dependence_edges): Remove loop nest parameter. Use global
23939 (distribute_loop): Compute global variable loop nest. Update use.
23941 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23943 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
23944 (partition_merge_into): New parameter. Dump reason for fusion.
23945 (distribute_loop): Update use of partition_merge_into.
23947 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23949 * tree-loop-distribution.c (bb_top_order_index): New.
23950 (bb_top_order_index_size, bb_top_order_cmp): New.
23951 (stmts_from_loop): Use topological order.
23952 (pass_loop_distribution::execute): Compute and release topological
23953 order for basic blocks.
23955 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23957 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
23960 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23962 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
23963 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
23964 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
23965 * internal-fn.def (LOOP_DIST_ALIAS): New.
23966 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
23967 (fold_loop_internal_call): ... this.
23968 (vect_loop_dist_alias_call): New function.
23969 (set_uid_loop_bbs): Call fold_loop_internal_call.
23970 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
23973 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
23976 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
23977 Require dead FLAGS_REG at the beginning of a peephole.
23979 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
23982 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
23983 arguments in the call to __builtin_ia32_sbb_u32.
23984 (_subborrow_u64): Swap _X and _Y arguments in the call to
23985 __builtin_ia32_sbb_u64.
23987 2017-07-04 Jakub Jelinek <jakub@redhat.com>
23990 * tree-vrp.c (compare_assert_loc): Turn into a function template
23991 with stable template parameter. Only test if a->e is NULL,
23992 !a->e == !b->e has been verified already. Use e == NULL or
23993 e != NULL instead of e or ! e tests. If stable is true, don't use
23994 iterative_hash_expr, on the other side allow a or b or both NULL
23995 and sort the NULLs last.
23996 (process_assert_insertions): Sort using compare_assert_loc<false>
23997 instead of compare_assert_loc, later sort using
23998 compare_assert_loc<true> before calling process_assert_insertions_for
23999 in a loop. Use break instead of continue once seen NULL pointer.
24001 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
24003 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
24004 Cortex-R7 and Cortex-R8 processors.
24006 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
24008 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
24009 uninitialized while src is not.
24011 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
24013 * common/config/arm/arm-common.c: Adjust include path for
24015 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
24016 (arm-cpu.h): Create in build directory. Adjust dependency rules.
24017 (arm-cpu-data.h): Likewise.
24018 (arm-cpu-cdata.h): Likewise.
24019 * config/arm/arm-cpu.h: Delete.
24020 * config/arm/arm-cpu-cdata.h: Delete.
24021 * config/arm/arm-cpu-data.h: Delete.
24023 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
24025 * config/arm/arm-cpus.in (cortex-a55): New.
24026 (cortex-a75): Likewise.
24027 (cortex-a75.cortex-a55): Likewise.
24028 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
24030 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
24031 * config/arm/arm-cpu-cdata.h: Regenerate.
24032 * config/arm/arm-cpu-data.h: Regenerate.
24033 * config/arm/arm-cpu.h: Regenerate.
24034 * config/arm/arm-tables.opt: Regenerate.
24035 * config/arm/arm-tune.md: Regenerate.
24037 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
24039 * haifa-sched.c (sched_create_recovery_edges): Update profile.
24041 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
24043 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
24046 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
24048 PR tree-optimization/81292
24049 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
24050 full_string_p, also call adjust_related_strinfos if the adjustment
24051 is simple, otherwise invalidate related strinfos.
24053 2017-07-04 Martin Liska <mliska@suse.cz>
24056 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
24057 newly created variable as DECL_IGNORED_P.
24059 2017-07-04 Martin Liska <mliska@suse.cz>
24062 * ipa-inline.c (inline_small_functions):
24063 Use xstrdup_for_dump.
24065 2017-07-04 Tom de Vries <tom@codesourcery.com>
24067 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
24069 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
24072 * config/darwin.c (darwin_function_switched_text_sections):
24075 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
24077 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
24079 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
24081 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
24083 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24085 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
24086 min_profitable_iters, and th as inclusive lower bounds.
24087 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
24088 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
24089 for min_profitable_iters and min_profitable_estimate.
24090 (vect_transform_loop): Treat th as an inclusive lower bound.
24091 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
24093 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
24096 * config/darwin.c (darwin_function_switched_text_sections):
24097 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
24098 in two pieces, and suppress the use of buf.
24100 2017-07-03 Nathan Sidwell <nathan@acm.org>
24102 * hash-table.h (hash_table_mod1): Fix indentation.
24104 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24106 PR middle-end/81290
24107 * predict.c (force_edge_cold): Be more careful about propagation
24109 * profile-count.h (profile_probability::guessed,
24110 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
24112 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
24114 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
24116 * doc/invoke.texi (rcpc architecture extension): Document it.
24118 2017-07-03 Richard Biener <rguenther@suse.de>
24120 PR tree-optimization/60510
24121 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
24122 the scalar reduction PHI and use it.
24123 (vectorizable_reduction): Properly guard the single_defuse_cycle
24124 path for non-SLP reduction chains where we cannot use it.
24125 Rework reduc_def/index and vector type deduction. Rework
24126 vector operand gathering during reduction op code-gen.
24127 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
24128 chains dissolve the chain and leave it to non-SLP reduction
24131 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24133 * tree-data-ref.h (dr_alignment): Declare.
24134 * tree-data-ref.c (dr_alignment): New function.
24135 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
24136 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
24138 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
24140 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24142 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
24143 and base_misalignment fields.
24144 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
24145 * tree-data-ref.c: Include builtins.h.
24146 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
24147 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
24148 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
24149 * tree-vect-data-refs.c: Include tree-cfg.h.
24150 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
24151 fields instead of calculating an alignment here.
24152 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
24153 innermost_loop_behavior fields.
24155 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24157 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
24159 (DR_STEP_ALIGNMENT): New macro.
24160 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
24161 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
24162 (create_data_ref): Print it.
24163 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
24164 to tell whether the step preserves vector (mis)alignment.
24165 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
24166 Move the check for an integer step and generalise to all INTEGER_CST.
24167 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
24168 Print the outer step alignment.
24170 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24172 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
24173 with offset_alignment.
24174 (DR_ALIGNED_TO): Delete.
24175 (DR_OFFSET_ALIGNMENT): New macro.
24176 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
24177 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
24178 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
24179 (create_data_ref): Likewise.
24180 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
24181 (vect_analyze_data_refs): Likewise.
24182 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
24183 creating dummy innermost behavior.
24185 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24187 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
24188 with a "innermost_loop_behavior *" and refeence tree.
24189 * tree-data-ref.c (dr_analyze_innermost): Likewise.
24190 (create_data_ref): Update call accordingly.
24191 * tree-predcom.c (find_looparound_phi): Likewise.
24193 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24195 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
24196 fields with dr_wrt_vec_loop.
24197 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
24198 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
24199 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
24200 (vect_dr_behavior): New function.
24201 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
24202 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
24203 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
24204 track whether the step preserves the misalignment.
24205 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
24206 Use vect_dr_behavior.
24207 (vect_setup_realignment): Update call accordingly.
24208 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
24209 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
24210 call to vect_create_addr_base_for_vector_ref.
24211 (vect_create_cond_for_align_checks): Likewise.
24212 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
24213 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
24214 (vect_recog_mask_conversion_pattern): Likewise.
24215 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
24216 (new_stmt_vec_info): Remove redundant zeroing.
24218 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
24220 * common/config/arm/arm-common.c (arm_be8_option): New function.
24221 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
24222 (ISA_ARMv6): Add isa_bit_be8.
24223 * config/arm/arm.h (arm_be8_option): Add prototype.
24224 (BE8_SPEC_FUNCTION): New define.
24225 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
24226 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
24227 (mlittle-endian): Similarly.
24228 (mbe8, mbe32): New options.
24229 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
24230 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
24232 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24234 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
24236 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24238 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
24239 (cleanup_tree_cfg_bb): Use it.
24240 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
24242 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
24244 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24247 * loop-doloop.c (add_test): Update profile.
24249 2017-07-03 Martin Liska <mliska@suse.cz>
24252 * sanopt.c (rewrite_usage_of_param): New function.
24253 (sanitize_rewrite_addressable_params): Likewise.
24254 (pass_sanopt::execute): Call rewrite_usage_of_param.
24256 2017-07-03 Richard Biener <rguenther@suse.de>
24258 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
24259 back to using VIEW_CONVERT_EXPR.
24261 2017-07-03 Martin Liska <mliska@suse.cz>
24264 * doc/extend.texi: Document when a resolver function is
24265 generated for target_clones.
24267 2017-07-03 Martin Liska <mliska@suse.cz>
24269 * asan.c (asan_emit_stack_protection): Unpoison just red zones
24270 and shadow memory of auto variables which are subject of
24271 use-after-scope sanitization.
24272 (asan_expand_mark_ifn): Add do set only when is_poison.
24274 2016-07-03 Richard Biener <rguenther@suse.de>
24276 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
24278 (vect_force_simple_reduction): Record reduction def -> phi mapping.
24279 (vectorizable_reduction): Perform reduction PHI creation when
24280 visiting a reduction PHI and adjust and simplify code generation
24281 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
24282 (vect_transform_loop): Visit reduction PHIs.
24283 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
24284 defs into the SLP tree.
24285 (vect_build_slp_tree): Reduction defs terminate the recursion.
24286 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
24288 (vect_get_vec_defs_for_stmt_copy): Export.
24289 (vect_get_vec_defs): Likewise.
24290 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
24292 (vect_get_vec_defs_for_stmt_copy): Declare.
24293 (vect_get_vec_defs): Likewise.
24295 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24297 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
24298 parameter with a "loop" parameter and use it instead of the
24299 loop containing DR_STMT. Don't check simple_iv when doing
24300 BB analysis. Describe the two analysis modes in the comment.
24302 2017-07-03 Tom de Vries <tom@codesourcery.com>
24304 PR tree-optimization/69468
24305 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
24306 (find_same_succ_bb): Handle ignore_edge_flags.
24308 2017-07-03 Tom de Vries <tom@codesourcery.com>
24310 PR tree-optimization/81192
24311 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
24313 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
24315 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
24317 2017-07-03 Tom de Vries <tom@codesourcery.com>
24319 PR tree-optimization/81192
24320 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
24321 BB_SAME_SUCC (bb) == NULL.
24323 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24325 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
24328 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24330 * dumpfile.c: Include profile-count.h
24331 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
24333 (insert_cond_bb): Update profile.
24334 * tree-cfg.h (insert_cond_bb): Update prototype.
24335 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
24336 * tree-dump.c: Do not include tree-cfg.
24338 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24340 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
24342 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24344 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
24347 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24349 * tree-complex.c (expand_complex_div_wide): update profile.
24351 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
24352 Alan Hayward <alan.hayward@arm.com>
24353 David Sherwood <david.sherwood@arm.com>
24355 * Makefile.in (MACHMODE_H): Remove insn-modes.h
24356 (CORETYPES_H): New define.
24357 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
24358 (insn-modes-inline.h, s-modes-inline-h): New rules.
24359 (generated_files): Add insn-modes-inline.h.
24360 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
24361 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
24362 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
24363 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
24364 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
24365 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
24366 (build/gencodes.o, build/genconditions.o): Likewise.
24367 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
24368 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
24369 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
24370 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
24371 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
24372 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
24373 * coretypes.h: Include everything up to real.h for generators.
24374 Include insn-modes.h first. Include wide-int-print.h after
24375 wide-int.h. Include insn-modes-inline.h and then machmode.h.
24376 * machmode.h: Don't include insn-modes.h here.
24377 * function-tests.c: Remove includes of signop.h, machmode.h,
24378 double-int.h and wide-int.h.
24380 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
24382 * optc-save-gen.awk: Likewise.
24383 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
24384 * godump.c: Remove include of wide-int-print.h.
24385 * pretty-print.h: Likewise.
24386 * wide-int-print.cc: Likewise.
24387 * wide-int.cc: Likewise.
24388 * hash-map-tests.c: Remove include of signop.h.
24389 * hash-set-tests.c: Likewise.
24390 * rtl-tests.c: Likewise.
24391 * mkconfig.sh: Remove include of machmode.h.
24392 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
24394 (emit_insn_modes_inline_h): ...this new function. Emit the code
24395 into an insn-modes-inline.h header file, adding appropriate
24396 include guards and end comments.
24397 (emit_insn_modes_c_header): Remove include of machmode.h.
24398 (emit_min_insn_modes_c_header): Include coretypes.h rather than
24400 (main): Handle -i flag and call emit_insn_modes_inline_h when
24403 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
24405 * tree-ssa-strlen.c (strinfo): Rename the length field to
24406 nonzero_chars. Add a full_string_p field.
24407 (compare_nonzero_chars, zero_length_string_p): New functions.
24408 (get_addr_stridx): Add an offset_out parameter.
24409 Use compare_nonzero_chars.
24410 (get_stridx): Update accordingly. Use compare_nonzero_chars.
24411 (new_strinfo): Update after above changes to strinfo.
24412 (set_endptr_and_length): Set full_string_p.
24413 (get_string_length): Update after above changes to strinfo.
24414 (unshare_strinfo): Update call to new_strinfo.
24415 (maybe_invalidate): Likewise.
24416 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
24417 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
24418 as a uhwi instead of an shwi. Update after above changes to
24419 strinfo and new_strinfo.
24420 (zero_length_string): Assert that chainsi contains full strings.
24421 Use zero_length_string_p. Update call to new_strinfo.
24422 (adjust_related_strinfos): Update after above changes to strinfo.
24423 Copy full_string_p from origsi.
24424 (adjust_last_stmt): Use zero_length_string_p.
24425 (handle_builtin_strlen): Update after above changes to strinfo and
24426 new_strinfo. Install the lhs as the string length if the previous
24427 entry didn't describe a full string.
24428 (handle_builtin_strchr): Update after above changes to strinfo
24430 (handle_builtin_strcpy): Likewise.
24431 (handle_builtin_strcat): Likewise.
24432 (handle_builtin_malloc): Likewise.
24433 (handle_pointer_plus): Likewise.
24434 (handle_builtin_memcpy): Likewise. Track nonzero characters
24435 that aren't necessarily followed by a nul terminator.
24436 (handle_char_store): Likewise.
24438 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
24440 PR tree-optimization/80769
24441 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
24442 for malloc and calloc. Document the new invariant that all related
24443 strinfos have delayed lengths or none do.
24444 (verify_related_strinfos): Move earlier in file.
24445 (set_endptr_and_length): New function, split out from...
24446 (get_string_length): ...here. Also set the lengths of related
24448 (zero_length_string): Assert that chainsi has known (rather than
24450 (adjust_related_strinfos): Likewise.
24452 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
24454 PR tree-optimization/81136
24455 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
24456 assert that two references with the same misalignment have the same
24457 compile-time misalignment if those compile-time misalignments
24460 2017-07-01 Andi Kleen <ak@linux.intel.com>
24462 * print-tree.c (print_node): Print all attributes.
24464 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
24466 * cfg.c (scale_bbs_frequencies): New function.
24467 * cfg.h (scale_bbs_frequencies): Declare it.
24468 * cfgloopanal.c (single_likely_exit): Cleanup.
24469 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
24471 (scale_loop_profile): Likewise.
24472 (loop_version): Likewise.
24473 (create_empty_loop_on_edge): Update.
24474 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
24475 scale_loop_frequencies, scale_loop_profile, loopify,
24476 loop_version): Update prototypes.
24477 * modulo-sched.c (sms_schedule): Update.
24478 * predict.c (unlikely_executed_edge_p): Also check probability.
24479 (probably_never_executed_edge_p): Fix typo.
24480 * tree-if-conv.c (version_loop_for_if_conversion): Update.
24481 * tree-parloops.c (gen_parallel_loop): Update.
24482 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
24483 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
24484 * tree-ssa-loop-split.c (split_loop): Update.
24485 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
24486 * tree-vect-loop-manip.c (vect_do_peeling): Update.
24487 (vect_loop_versioning): Update.
24488 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
24490 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
24492 * trans-mem.c (split_bb_make_tm_edge): Update profile.
24494 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
24496 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
24497 to keep profile consistent.
24499 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
24501 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
24502 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
24503 * profile-count.h (max_safe_multiplier): Make unsigned.
24504 (profile_count::guessed_zero): New.
24506 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
24508 * bb-reorder.c (fix_up_crossing_landing_pad,
24509 fix_crossing_conditional_branches): Use make_single_succ_edge
24510 to keep profile consistent.
24512 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
24514 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
24517 2017-07-01 Jakub Jelinek <jakub@redhat.com>
24520 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
24521 the right scopes, make sure cond_jump isn't preserved between multiple
24522 iterations. Search for fallthru edge whenever there are 3+ edges and
24523 use find_fallthru_edge for it.
24525 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
24527 Patch by Alexander Monakov <amonakov@ispras.ru>
24528 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
24529 probabilities consistently.
24531 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
24533 * pa.c (pa_expand_compare_and_swap_loop): Update call of
24534 emit_cmp_and_jump_insns.
24536 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
24539 * tree-inline.c (expand_call_inline): Combine profile statuses.
24541 2017-06-30 Andrew Pinski <apinski@cavium.com>
24543 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
24544 fold_stmt returned true.
24546 2017-06-30 Nathan Sidwell <nathan@acm.org>
24548 * ggc.h (empty_string): Delete.
24549 * cfgexpand.c (expand_asm_stmt): Use plain "".
24550 * optabs.c (expand_asm_memory_barrier): Likewise.
24551 * stringpool.c (empty_string): Delete.
24552 (digit_vector, digit_string): Delete.
24553 (ggc_alloc_string): Use plain "", don't optimize single digit
24554 strings. Use ggc_alloc_atomic.
24556 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
24558 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
24559 comparison set and one other set, use the cost of the non-comparison
24562 2017-06-30 Nathan Sidwell <nathan@acm.org>
24564 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
24567 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
24569 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
24570 loops. Remove now unneeded calls to gimple_switch_set_label() that
24571 just set removed labels to NULL_TREE.
24573 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
24575 * tree-ssanames.c (set_range_info_raw): Abstract from ...
24576 (set_range_info): ...here. Only call set_range_info_raw if domain
24578 (set_nonzero_bits): Call set_range_info_raw.
24579 * tree-ssanames.h (set_range_info_raw): New.
24581 2017-06-30 Jakub Jelinek <jakub@redhat.com>
24584 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
24585 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
24586 of nonimmediate_operand and <store_mask_constraint> instead of m
24587 for the input operand. For V8FI iterator, always split if input
24588 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
24589 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
24590 <store_mask_predicate> instead of register_operand and
24591 <store_mask_constraint> instead of v for the input operand. Make
24592 sure both operands aren't MEMs for if not <mask_applied>.
24594 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
24596 * lto-wrapper.c (copy_file) Close both file descriptors before
24599 2017-06-30 Martin Liska <mliska@suse.cz>
24602 * multiple_target.c (create_dispatcher_calls): Make ifunc
24603 also for function that don't have calls or are not referenced.
24605 2017-06-30 Richard Biener <rguenther@suse.de>
24607 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
24608 analyze the first scalar stmt. Move vector type computation
24609 for the BB case here from ...
24610 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
24611 live operation processing in the SLP case properly.
24613 2017-06-30 Richard Biener <rguenther@suse.de>
24615 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
24617 2017-06-30 Martin Liska <mliska@suse.cz>
24620 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
24621 before BUILT_IN_UNWIND_RESUME when ASAN is used.
24623 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
24625 * doc/invoke.texi (AArch64): Add missing options and remove redundant
24628 2017-06-30 Richard Biener <rguenther@suse.de>
24630 PR tree-optimization/81249
24631 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
24632 condition reduction result to original scalar type.
24634 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24636 * profile-count.h (enum profile_quality): Fix typos and whitespace
24639 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24641 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
24642 type for branch probabilities.
24644 2017-06-29 Julian Brown <julian@codesourcery.com>
24645 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24647 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
24648 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
24649 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
24650 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
24652 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24654 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
24655 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
24656 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
24657 CC usage from generic code to here.
24658 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
24659 CC usage into the target macros.
24661 2017-06-29 Maya Rashish <coypu@sdf.org>
24663 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
24666 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
24668 * arm/arm-builtins.c: Include profile-count.h
24669 * except.c (sjlj_emit_function_enter): Use
24670 profile_probability::unlikely.
24672 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24674 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
24675 and tocrel_offset be pointer args rather than implicitly using
24677 (legitimate_constant_pool_address_p, rs6000_emit_move,
24678 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
24679 tocrel_offset and use in toc_relative_expr_p call.
24680 (print_operand, print_operand_address): Use static tocrel_base_oac
24681 and tocrel_offset_oac.
24682 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
24685 2017-06-29 Maya Rashish <coypu@sdf.org>
24687 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
24689 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
24691 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
24692 objects, take into account only the alignment of 'op0' and 'mode1' if
24695 2017-06-29 Steve Ellcey <sellcey@cavium.com>
24697 * ccmp.c (ccmp_tree_comparison_p): New function.
24698 (ccmp_candidate_p): Update to use above function.
24699 (get_compare_parts): New function.
24700 (expand_ccmp_next): Update to use new functions.
24701 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
24703 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
24704 take mode as argument.
24705 * ccmp.h (expand_ccmp_expr): Add mode as argument.
24706 * expr.c (expand_expr_real_1): Pass mode as argument.
24708 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
24710 * combine.c (combine_instructions): Print insns to dump_file, together
24713 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
24715 * asan.c (asan_emit_stack_protection): Update.
24716 (create_cond_insert_point): Update.
24717 * auto-profile.c (afdo_propagate_circuit): Update.
24718 * basic-block.h (struct edge_def): Turn probability to
24719 profile_probability.
24720 (EDGE_FREQUENCY): Update.
24721 * bb-reorder.c (find_traces_1_round): Update.
24722 (better_edge_p): Update.
24723 (sanitize_hot_paths): Update.
24724 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
24725 (make_single_succ_edge): Update.
24726 (check_bb_profile): Update.
24727 (dump_edge_info): Update.
24728 (update_bb_profile_for_threading): Update.
24729 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
24730 probabilitycount to 0.
24731 * cfgbuild.c (compute_outgoing_frequencies): Update.
24732 * cfgcleanup.c (try_forward_edges): Update.
24733 (outgoing_edges_match): Update.
24734 (try_crossjump_to_edge): Update.
24735 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
24736 (expand_gimple_tailcall): Update.
24737 (construct_init_block): Use make_single_succ_edge.
24738 (construct_exit_block): Use make_single_succ_edge.
24739 * cfghooks.c (verify_flow_info): Update.
24740 (redirect_edge_succ_nodup): Update.
24741 (split_edge): Update.
24742 (account_profile_record): Update.
24743 * cfgloopanal.c (single_likely_exit): Update.
24744 * cfgloopmanip.c (scale_loop_profile): Update.
24745 (set_zero_probability): Remove.
24746 (duplicate_loop_to_header_edge): Update.
24747 * cfgloopmanip.h (loop_version): Update prototype.
24748 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
24749 (force_nonfallthru_and_redirect): Update.
24750 (update_br_prob_note): Update.
24751 (rtl_verify_edges): Update.
24752 (purge_dead_edges): Update.
24753 (rtl_lv_add_condition_to_bb): Update.
24754 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
24755 * cgraphunit.c (init_lowered_empty_function): Update.
24756 (cgraph_node::expand_thunk): Update.
24757 * cilk-common.c: Include profile-count.h
24758 * dojump.c (inv): Remove.
24759 (jumpifnot): Update.
24760 (jumpifnot_1): Update.
24761 (do_jump_1): Update.
24763 (do_jump_by_parts_greater_rtx): Update.
24764 (do_compare_rtx_and_jump): Update.
24765 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
24766 do_jump_1. do_compare_rtx_and_jump): Update prototype.
24767 * dwarf2cfi.c: Include profile-count.h
24768 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
24769 (sjlj_emit_dispatch_table): Likewise.
24770 * explow.c: Include profile-count.h
24771 * expmed.c (emit_store_flag_force): Update.
24772 (do_cmp_and_jump): Update.
24773 * expr.c (compare_by_pieces_d::generate): Update.
24774 (compare_by_pieces_d::finish_mode): Update.
24775 (emit_block_move_via_loop): Update.
24776 (store_expr_with_bounds): Update.
24777 (store_constructor): Update.
24778 (expand_expr_real_2): Update.
24779 (expand_expr_real_1): Update.
24780 * expr.h (try_casesi, try_tablejump): Update prototypes.
24781 * gimple-pretty-print.c (dump_probability): Update.
24782 (dump_profile): New.
24783 (dump_gimple_label): Update.
24784 (dump_gimple_bb_header): Update.
24785 * graph.c (draw_cfg_node_succ_edges): Update.
24786 * hsa-gen.c (convert_switch_statements): Update.
24787 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
24788 (find_if_case_1): Update.
24789 (find_if_case_2): Update.
24790 * internal-fn.c (expand_arith_overflow_result_store): Update.
24791 (expand_addsub_overflow): Update.
24792 (expand_neg_overflow): Update.
24793 (expand_mul_overflow): Update.
24794 (expand_vector_ubsan_overflow): Update.
24795 * ipa-cp.c (good_cloning_opportunity_p): Update.
24796 * ipa-split.c (split_function): Use make_single_succ_edge.
24797 * ipa-utils.c (ipa_merge_profiles): Update.
24798 * loop-doloop.c (add_test): Update.
24799 (doloop_modify): Update.
24800 * loop-unroll.c (compare_and_jump_seq): Update.
24801 (unroll_loop_runtime_iterations): Update.
24802 * lra-constraints.c (lra_inheritance): Update.
24803 * lto-streamer-in.c (input_cfg): Update.
24804 * lto-streamer-out.c (output_cfg): Update.
24805 * mcf.c (adjust_cfg_counts): Update.
24806 * modulo-sched.c (sms_schedule): Update.
24807 * omp-expand.c (expand_omp_for_init_counts): Update.
24808 (extract_omp_for_update_vars): Update.
24809 (expand_omp_ordered_sink): Update.
24810 (expand_omp_for_ordered_loops): Update.
24811 (expand_omp_for_generic): Update.
24812 (expand_omp_for_static_nochunk): Update.
24813 (expand_omp_for_static_chunk): Update.
24814 (expand_cilk_for): Update.
24815 (expand_omp_simd): Update.
24816 (expand_omp_taskloop_for_outer): Update.
24817 (expand_omp_taskloop_for_inner): Update.
24818 * omp-simd-clone.c (simd_clone_adjust): Update.
24819 * optabs.c (expand_doubleword_shift): Update.
24820 (expand_abs): Update.
24821 (emit_cmp_and_jump_insn_1): Update.
24822 (expand_compare_and_swap_loop): Update.
24823 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
24824 * predict.c (predictable_edge_p): Update.
24825 (edge_probability_reliable_p): Update.
24826 (set_even_probabilities): Update.
24827 (combine_predictions_for_insn): Update.
24828 (combine_predictions_for_bb): Update.
24829 (propagate_freq): Update.
24830 (estimate_bb_frequencies): Update.
24831 (force_edge_cold): Update.
24832 * profile-count.c (profile_count::dump): Add missing space into dump.
24833 (profile_count::debug): Add newline.
24834 (profile_count::differs_from_p): Explicitly convert to unsigned.
24835 (profile_count::stream_in): Update.
24836 (profile_probability::dump): New member function.
24837 (profile_probability::debug): New member function.
24838 (profile_probability::differs_from_p): New member function.
24839 (profile_probability::differs_lot_from_p): New member function.
24840 (profile_probability::stream_in): New member function.
24841 (profile_probability::stream_out): New member function.
24842 * profile-count.h (profile_count_quality): Rename to ...
24843 (profile_quality): ... this one.
24844 (profile_probability): New.
24845 (profile_count): Update.
24846 * profile.c (compute_branch_probabilities): Update.
24847 * recog.c (peep2_attempt): Update.
24848 * sched-ebb.c (schedule_ebbs): Update.
24849 * sched-rgn.c (find_single_block_region): Update.
24850 (compute_dom_prob_ps): Update.
24851 (schedule_region): Update.
24852 * sel-sched-ir.c (compute_succs_info): Update.
24853 * stmt.c (struct case_node): Update.
24854 (do_jump_if_equal): Update.
24855 (get_outgoing_edge_probs): Update.
24856 (conditional_probability): Update.
24857 (emit_case_dispatch_table): Update.
24858 (expand_case): Update.
24859 (expand_sjlj_dispatch_table): Update.
24860 (emit_case_nodes): Update.
24861 * targhooks.c: Update.
24862 * tracer.c (better_p): Update.
24863 (find_best_successor): Update.
24864 * trans-mem.c (expand_transaction): Update.
24865 * tree-call-cdce.c: Update.
24866 * tree-cfg.c (gimple_split_edge): Upate.
24867 (move_sese_region_to_fn): Upate.
24868 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
24869 * tree-eh.c (lower_resx): Upate.
24870 (cleanup_empty_eh_move_lp): Upate.
24871 * tree-if-conv.c (version_loop_for_if_conversion): Update.
24872 * tree-inline.c (copy_edges_for_bb): Update.
24873 (copy_cfg_body): Update.
24874 * tree-parloops.c (gen_parallel_loop): Update.
24875 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
24876 (gimple_gen_time_profiler): Update.
24877 * tree-ssa-dce.c (remove_dead_stmt): Update.
24878 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
24879 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
24880 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
24881 (unloop_loops): Update.
24882 (try_peel_loop): Update.
24883 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
24884 * tree-ssa-loop-split.c (connect_loops): Update.
24885 (split_loop): Update.
24886 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
24887 (hoist_guard): Update.
24888 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
24889 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
24890 (value_replacement): Update.
24891 * tree-ssa-reassoc.c (branch_fixup): Update.
24892 * tree-ssa-tail-merge.c (replace_block_by): Update.
24893 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
24894 (create_edge_and_update_destination_phis): Update.
24895 (compute_path_counts): Update.
24896 (recompute_probabilities): Update.
24897 (update_joiner_offpath_counts): Update.
24898 (freqs_to_counts_path): Update.
24899 (duplicate_thread_path): Update.
24900 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
24901 (struct switch_conv_info): Update.
24902 (gen_inbound_check): Update.
24903 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
24904 (vect_do_peeling): Update.
24905 (vect_loop_versioning): Update.
24906 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
24907 (optimize_mask_stores): Update.
24908 * ubsan.c (ubsan_expand_null_ifn): Update.
24909 * value-prof.c (gimple_divmod_fixed_value): Update.
24910 (gimple_divmod_fixed_value_transform): Update.
24911 (gimple_mod_pow2): Update.
24912 (gimple_mod_pow2_value_transform): Update.
24913 (gimple_mod_subtract): Update.
24914 (gimple_mod_subtract_transform): Update.
24915 (gimple_ic): Update.
24916 (gimple_stringop_fixed_value): Update.
24917 (gimple_stringops_transform): Update.
24918 * value-prof.h: Update.
24920 2017-06-29 Carl Love <cel@us.ibm.com>
24922 * config/rs6000/rs6000-c.c: Add support for built-in functions
24923 vector signed int vec_signed (vector float);
24924 vector signed long long vec_signed (vector double);
24925 vector signed int vec_signed2 (vector double, vector double);
24926 vector signed int vec_signede (vector double);
24927 vector signed int vec_signedo (vector double);
24928 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
24929 instruction generator.
24930 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
24931 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
24932 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
24934 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
24935 vunsignede_v2df): Add define_expands.
24936 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
24937 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
24938 VEC_UNSIGNEDO): Add definitions.
24939 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
24940 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
24941 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
24942 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
24943 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
24944 * config/rs6000/altivec.h (vec_signed, vec_signed2,
24945 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
24946 vec_unsignede, vec_unsignedo): Add builtin defines.
24947 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
24949 * doc/extend.texi: Update the built-in documentation file for the
24950 new built-in functions.
24952 2017-06-29 Richard Biener <rguenther@suse.de>
24954 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
24955 reduction chains to LOOP_VINFO_REDUCTIONS.
24956 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
24957 SLP reductions after processing reduction chains.
24959 2017-06-29 Nathan Sidwell <nathan@acm.org>
24961 * builtins.c (fold_builtin_FUNCTION): Use
24962 lang_hooks.decl_printable_name.
24964 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
24966 PR middle-end/81194
24967 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
24968 with only one label.
24969 * stmt.c (expand_case): Assert NCASES is greater than one.
24971 2017-06-29 Richard Biener <rguenther@suse.de>
24973 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
24975 (group_case_labels): Likewise.
24976 (find_taken_edge): Push sanity checking on val to workers...
24977 (find_taken_edge_cond_expr): ... here
24978 (find_taken_edge_switch_expr): ... and here, handle cases
24979 with just a default label.
24980 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
24981 (group_case_labels): Likewise.
24982 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
24983 group_case_labels does anything cleanup the CFG again.
24985 2017-06-29 Bin Cheng <bin.cheng@arm.com>
24987 PR tree-optimization/81196
24988 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
24989 exit condition comparing two IVs.
24991 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
24993 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
24994 profile to the dummy entry at the end of the list of architectures.
24995 * config/arm/arm-cpu-cdata.h: Regenerated.
24997 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24998 Michael Collison <michael.collison@arm.com>
25001 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
25003 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
25004 (*aarch64_reg_<mode>3_minus_mask): New pattern.
25005 (*aarch64_<optab>_reg_di3_mask2): New pattern.
25006 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
25007 of shift when the shift amount is masked with constant equal to
25008 the size of the mode.
25009 * config/aarch64/predicates.md (subreg_lowpart_operator): New
25012 2017-06-29 Martin Liska <mliska@suse.cz>
25014 * config/i386/i386.opt: Change range from [1,5] to [0,5].
25016 2017-06-29 Yury Gribov <tetra2005@gmail.com>
25019 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
25022 (edge_growth_cache_entry::edge_growth_cache_entry): New
25024 (reset_edge_growth_cache): Update to use constructor.
25026 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25028 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
25029 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
25030 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
25032 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
25034 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
25035 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
25037 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
25039 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
25040 (*-linux-uclibc*): Add t-uclibc tmake_file.
25041 * config/t-musl: New.
25042 * config/t-uclibc: New.
25044 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
25046 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
25048 (gen_comm_data): Emit architectural setting of arch_prof.
25049 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
25051 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
25052 (armv8-m.base, armv8-m.main): Likewise.
25053 * arm-protos.h (arm_build_target): Add profile field.
25054 (arch_option): Likewise.
25055 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
25057 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
25058 arm_active_target.profile.
25060 2017-06-28 Richard Biener <rguenther@suse.de>
25062 PR middle-end/81227
25063 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
25064 TYPE_OVERFLOW_WRAPS.
25065 * match.pd (negate_expr_p): Likewise.
25066 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
25067 fold_build2, not fold_binary.
25069 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
25071 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
25072 Convert memory address to Pmode.
25073 (aarch64_print_operand): Assert MEM operands are always Pmode.
25075 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
25078 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
25079 Remove redundant if.
25080 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
25081 * config/arm/aarch-common-protos.h
25082 (aarch_forward_to_shift_is_not_shifted_re): Remove.
25083 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
25085 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
25088 * multiple_target.c (create_dispatcher_calls): Set the default
25089 clone to be static, not public.
25091 2017-06-28 Richard Biener <rguenther@suse.de>
25093 * tree-vect-loop.c (vectorizable_reduction): Move special
25094 cond reduction IV var creation ...
25095 (vect_create_epilog_for_reduction): ... here. Remove induction_index
25096 parameter. Use STMT_VINFO_VECTYPE.
25097 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
25100 2017-06-28 Martin Liska <mliska@suse.cz>
25103 * ipa-visibility.c (non_local_p): Handle visibility.
25105 2017-06-28 Martin Liska <mliska@suse.cz>
25108 * common.opt: Add IntegerRange to various options.
25109 * opt-functions.awk (integer_range_info): New function.
25110 * optc-gen.awk: Add integer_range_info to cl_options struct.
25111 * opts-common.c (decode_cmdline_option): Handle
25112 CL_ERR_INT_RANGE_ARG.
25113 (cmdline_handle_error): Likewise.
25114 * opts.c (print_filtered_help): Show valid interval in
25115 when --help is provided.
25116 * opts.h (struct cl_option): Add range_min and range_max fields.
25117 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
25119 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
25121 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
25122 (x * C EQ/NE y * C): New transformation.
25124 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
25126 * genmultilib (combination_space): Accept '+' in option names.
25128 2017-06-28 Martin Liska <mliska@suse.cz>
25131 * asan.c (instrument_derefs): Bail out inner references
25132 that are hard register variables.
25134 2017-06-28 Jakub Jelinek <jakub@redhat.com>
25137 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
25138 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
25140 2017-06-28 Richard Biener <rguenther@suse.de>
25142 * tree-vectorizer.h (vect_get_vec_defs): Remove.
25143 (vect_get_slp_defs): Adjust.
25144 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
25146 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
25148 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
25149 get_initial_defs_for_reduction instead of vect_get_vec_defs.
25150 (vectorizable_reduction): Adjust.
25151 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
25153 (vect_get_slp_defs): Likewise.
25154 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
25155 (vectorizable_bswap): Adjust.
25156 (vectorizable_call): Likewise.
25157 (vectorizable_conversion): Likewise.
25158 (vectorizable_assignment): Likewise.
25159 (vectorizable_shift): Likewise.
25160 (vectorizable_operation): Likewise.
25161 (vectorizable_store): Likewise.
25162 (vectorizable_condition): Likewise.
25163 (vectorizable_comparison): Likewise.
25165 2017-06-28 Michael Collison <michael.collison@arm.com>
25168 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
25170 (arm_gen_movmemqi): Removed unused variable 'i'.
25171 Convert 'for' loop into 'while' loop.
25172 (arm_expand_prologue): Remove last unnecessary set of insn.
25173 (thumb_pop): Remove unused variable 'pushed_words'.
25174 (thumb_exit): Remove last unnecessary set of regs_to_pop.
25176 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25178 * config/s390/predicates.md: Use s390_rel_address_ok_p.
25179 * config/s390/s390-protos.h: Add prototype of
25180 s390_rel_address_ok_p.
25181 * config/s390/s390.c (s390_got_symbol): New function.
25182 (s390_rel_address_ok_p): New function.
25183 (legitimize_pic_address): Use s390_rel_address_ok_p.
25184 (s390_load_got): Use s390_got_symbol.
25185 (s390_option_override): Issue error if
25186 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
25187 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
25189 * config/s390/s390.opt: New option mpic-data-is-text-relative.
25191 2017-06-27 Andrew Pinski <apinski@cavium.com>
25193 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
25194 (X * copysign (1.0, X)): New pattern.
25195 (X * copysign (1.0, -X)): New pattern.
25196 (copysign (-1.0, CST)): New pattern.
25198 2017-06-27 Joseph Myers <joseph@codesourcery.com>
25200 * genmultilib (combination_space): Remove variable.
25201 Validate reuse rules against regular expression for any sequence
25202 of multilib options in any order.
25204 2017-06-27 Michael Collison <michael.collison@arm.com>
25206 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
25207 call aarch64_split_simd_combine.
25208 * (aarch64_combine_internal<mode>): Delete pattern.
25209 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
25210 Allow register and subreg operands.
25212 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25214 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
25215 specific need, just fallback on defaults.
25216 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
25218 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25219 Olivier Hainque <hainque@adacore.com>
25221 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
25223 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
25224 targets. Pick a default if no particular attempt applied.
25225 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
25228 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25230 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
25231 (x86_64-wrs-vxworks7): Likewise.
25233 2017-06-27 Marek Polacek <polacek@redhat.com>
25236 * ubsan.c (instrument_null): Check get_base_address's result for null.
25238 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
25240 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
25242 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
25244 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
25245 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
25246 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
25247 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
25248 New function types.
25249 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
25250 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
25251 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
25252 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
25253 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
25254 BUILT_IN_FEUPDATEENV): New builtins.
25255 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
25256 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
25257 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
25258 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
25260 (builtin_structptr_types): Adjust size.
25261 * tree.c (builtin_structptr_types): Add four entries.
25263 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25264 Olivier Hainque <hainque@adacore.com>
25266 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
25267 (TLS_SYM): New local macro, forcing reference to __tls__ on
25268 link command lines for VxWorks 7 RTPs, triggering initialization
25270 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
25271 OS features TLS support, true for RTPs on VxWorks 7.
25272 * config/vxworks.c (vxworks_override_options): Setup emutls
25275 2017-06-27 Jakub Jelinek <jakub@redhat.com>
25277 * predict.c (test_prediction_value_range): Use -1U instead of -1
25278 to avoid narrowing conversion warning.
25279 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
25280 to avoid narrowing conversion warning.
25281 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
25283 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
25285 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25287 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
25288 64bit configurations.
25289 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
25290 (SIZE_TYPE): Likewise.
25291 * config/vxworks.c (vxworks_emutls_var_fields): Use
25292 long_unsigned_type_node instead of unsigned_type_node as the offset
25293 field type, which is "pointer" mode in emutls.c.
25295 2017-06-27 Jakub Jelinek <jakub@redhat.com>
25298 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
25300 PR middle-end/81207
25301 * gimple-fold.c (replace_call_with_call_and_fold): Handle
25302 gimple_vuse copying separately from gimple_vdef copying.
25304 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25306 * value-prof.c (free_hist): Remove call to memset and the enclosing if
25309 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
25310 Olivier Hainque <hainque@adacore.com>
25312 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
25313 for all vxworks7 targets.
25314 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
25315 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
25316 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
25317 variations for VX6/VX7 and 32/64bits later on in ...
25318 (VXWORKS_LIB_SPEC): Leverage new macros.
25319 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
25320 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
25322 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
25324 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
25325 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
25327 2017-06-26 Carl Love <cel@us.ibm.com>
25329 * config/rs6000/rs6000-c.c: Add support for built-in functions
25330 vector bool char vec_reve (vector bool char);
25331 vector signed char vec_reve (vector signed char);
25332 vector unsigned char vec_reve (vector unsigned char);
25333 vector bool int vec_reve (vector bool int);
25334 vector signed int vec_reve (vector signed int);
25335 vector unsigned int vec_reve (vector unsigned int);
25336 vector bool long long vec_reve (vector bool long long);
25337 vector signed long long vec_reve (vector signed long long);
25338 vector unsigned long long vec_reve (vector unsigned long long);
25339 vector bool short vec_reve (vector bool short);
25340 vector signed short vec_reve (vector signed short);
25341 vector double vec_reve (vector double);
25342 vector float vec_reve (vector float);
25343 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
25344 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
25345 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
25346 (altivec_vreve): New pattern.
25347 * config/rs6000/altivec.h (vec_reve): New define.
25348 * doc/extend.texi (vec_rev): Update the built-in documentation file
25349 for the new built-in functions.
25351 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25353 PR tree-optimization/71815
25354 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
25356 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
25358 (slsr_process_phi): Likewise.
25359 (replace_uncond_cands_and_profitable_phis): Don't replace a
25360 multiply candidate with a stride of 1 (copy or cast).
25361 (phi_incr_cost): Call uses_consumed_by_stmt rather than
25363 (lowest_cost_path): Likewise.
25364 (total_savings): Likewise.
25366 2017-06-26 Richard Biener <rguenther@suse.de>
25369 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
25370 Use def_builtin_pure for all gather builtins.
25372 2017-06-26 Richard Biener <rguenther@suse.de>
25374 PR tree-optimization/81203
25375 * tree-tailcall.c (find_tail_calls): Do not move stmts into
25376 non-dominating BBs.
25378 2017-06-26 Marek Polacek <polacek@redhat.com>
25381 * doc/invoke.texi: Document -Wmultistatement-macros.
25383 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
25385 * doc/sourcebuild.texi (ARM-specific attributes): Document new
25386 arm_neon_ok_no_float_abi effective target.
25388 2017-06-26 Richard Biener <rguenther@suse.de>
25390 PR tree-optimization/80928
25391 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
25392 (copy_bbs): Set BB_DUPLICATED flag early.
25393 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
25395 (execute_on_shrinking_pred): Likewise.
25396 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
25397 BB_DUPLICATED blocks.
25398 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
25399 iterate over all PHIs considering removal of *gsi.
25401 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
25403 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
25406 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25408 * config/rs6000/rs6000-string.c: (expand_block_clear,
25409 do_load_for_compare, select_block_compare_mode,
25410 compute_current_alignment, expand_block_compare,
25411 expand_strncmp_align_check, expand_strn_compare,
25412 expand_block_move, rs6000_output_load_multiple)
25413 Move functions related to string/block move/compare
25414 to a separate file.
25415 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
25416 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
25417 for this function which is now used in two files.
25418 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
25419 * config.gcc: Add rs6000-string.o to extra_objs for
25420 targets powerpc*-*-* and rs6000*-*-*.
25422 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
25425 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
25426 32-bit, since indexed is not valid for DImode.
25427 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
25428 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
25429 (define_peephole2 for Altivec d-form load): Add 32-bit support.
25430 (define_peephole2 for Altivec d-form store): Likewise.
25433 * multiple_target.c (create_dispatcher_calls): Only create the
25434 dispatcher call if the function is the default clone of a
25435 versioned function.
25437 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
25439 PR middle-end/80902
25440 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
25441 a call, force the call to not be a tail call.
25443 2017-06-23 Jeff Law <law@redhat.com>
25445 * doc/contrib.texi: Add entry for Steven Pemberton's work on
25448 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
25450 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
25451 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
25452 handling for early expansion of vector shifts (sl,sr,sra,rl).
25453 (builtin_function_type): Add vector shift right instructions
25454 to the unsigned argument list.
25456 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
25458 rtl-optimizatoin/79286
25459 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
25460 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
25461 trap. PIC register plus a const unspec without offset can never trap.
25463 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
25465 * tree.h (builtin_structptr_type): New type.
25466 (builtin_structptr_types): Declare new array.
25467 * tree.c (builtin_structptr_types): New array.
25468 (free_lang_data, build_common_tree_nodes): Use it.
25470 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
25473 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
25476 2017-06-22 Matt Turner <mattst88@gmail.com>
25478 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
25479 Lake models to skylake case. Assume skylake for unknown
25480 models with clflushopt.
25482 2017-06-22 Jeff Law <law@redhat.com>
25484 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
25485 frame sizes that do not satisfy aarch64_uimm12_shift.
25487 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
25489 * profile-count.h (apply_probability,
25490 apply_scale, probability_in): Fix checks for zero.
25492 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
25494 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
25495 * doc/cppdiropts.texi (-I @var{dir}): Document it.
25497 2016-06-22 Richard Biener <rguenther@suse.de>
25499 * tree-vect-loop.c (vect_model_reduction_cost): Handle
25500 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
25501 REDUC_MAX_EXPR support.
25502 (vectorizable_reduction): Likewise.
25503 (vect_create_epilog_for_reduction): Likewise.
25505 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
25507 * match.pd (A / (1 << B) -> A >> B): New.
25508 * generic-match-head.c: Include optabs-tree.h.
25509 * gimple-match-head.c: Likewise.
25510 * optabs-tree.h (target_supports_op_p): New.
25511 * optabs-tree.c (target_supports_op_p): New.
25513 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
25515 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
25516 $gcc_cv_ld --help output.
25517 (gcc_cv_ld_demangle): Likewise.
25518 (gcc_cv_ld_eh_frame_hdr): Likewise.
25519 (gcc_cv_ld_pie): Likewise.
25520 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
25521 (gcc_cv_ld_buildid): Likewise.
25522 (gcc_cv_ld_sysroot): Likewise.
25523 (ld_bndplt_support): Likewise.
25524 (ld_pushpopstate_support): Likewise.
25525 * configure: Regenerate.
25526 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
25528 2017-06-21 Jakub Jelinek <jakub@redhat.com>
25531 * config/i386/sse.md (round<mode>2): Renumber match_dup and
25532 operands indexes to avoid gap between operands and match_dups.
25534 2017-06-21 Andrew Pinski <apinski@cavium.com>
25536 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
25537 Increment Arith_shift and Arith_shift_reg by 1.
25538 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
25540 * config/aarch64/aarch64.c (thunderx_tunings): Enable
25541 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
25542 (aarch64_strip_extend): Add new argument and test for it.
25543 (aarch64_cheap_mult_shift_p): New function.
25544 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
25545 add a cost if it is true.
25546 Update calls to aarch64_strip_extend.
25547 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
25549 2017-06-21 Andrew Pinski <apinski@cavium.com>
25551 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
25553 (thunderxt88): Likewise.
25554 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
25555 (thunderx_prefetch_tune): New variable.
25556 (thunderx2t99_prefetch_tune): Update for the correct values.
25557 (thunderxt88_tunings): New variable.
25558 (thunderx_tunings): Use thunderx_prefetch_tune instead of
25559 generic_prefetch_tune.
25560 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
25562 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25564 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
25565 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
25566 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
25567 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
25568 (aarch64_atomic_cas<mode>, GPI): Likewise.
25570 2017-06-21 Martin Liska <mliska@suse.cz>
25572 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
25573 statements on cold and hot labels.
25574 * predict.c (tree_estimate_probability_bb): Remove the
25575 prediction from this place.
25577 2017-06-21 Martin Liska <mliska@suse.cz>
25579 PR tree-optimization/79489
25580 * gimplify.c (maybe_add_early_return_predict_stmt): New
25582 (gimplify_return_expr): Call the function.
25583 * predict.c (tree_estimate_probability_bb): Remove handling
25585 * predict.def: Update comment about early return predictor.
25586 * gimple-predict.h (is_gimple_predict): New function.
25587 * predict.def: Change default value of early return to 66.
25588 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
25590 * passes.def: Put pass_strip_predict_hints to the beginning of
25593 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
25595 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
25596 FUNCTION_DECL declarations.
25597 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
25599 (dwaf2out_decl): Likewise.
25600 * godump.c (go_early_global_decl): Skip call to the real debug hook
25601 for FUNCTION_DECL declarations.
25602 * passes.c (rest_of_decl_compilation): Skip call to the
25603 early_global_decl debug hook for FUNCTION_DECL declarations, unless
25604 -fdump-go-spec is passed.
25606 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
25608 * config/i386/i386.c (struct builtin_isa): New field pure_p.
25609 Reorder for compactness.
25610 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
25611 (def_builtin_pure, def_builtin_pure2): New functions.
25612 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
25614 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
25616 * match.pd (nop_convert): New predicate.
25617 ((A +- CST1) +- CST2): Allow some NOP conversions.
25619 2017-06-21 Jakub Jelinek <jakub@redhat.com>
25622 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
25623 with ctors/dtors if GOVD_SHARED is set.
25625 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
25627 * config/aarch64/aarch64.md (movti_aarch64):
25628 Emit mov rather than orr.
25629 (movtf_aarch64): Likewise.
25630 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
25631 Emit mov rather than orr.
25633 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
25635 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
25636 Swap alternatives, make integer dup more expensive.
25638 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
25640 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
25641 Return true for non-tls symbols.
25643 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
25645 * config/aarch64/aarch64-cores.def (cortex-a55): New.
25646 (cortex-a75): Likewise.
25647 (cortex-a75.cortex-a55): Likewise.
25648 * config/aarch64/aarch64-tune.md: Regenerate.
25649 * doc/invoke.texi (-mtune): Document new values for -mtune.
25651 2017-06-21 Tom de Vries <tom@codesourcery.com>
25653 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
25654 stack_size feature.
25655 (Effective-Target Keywords, Other attributes): Suggest using
25656 dg-add-options stack_size feature to get stack limit in stack_size
25657 effective target documentation.
25659 2017-06-21 Julian Brown <julian@codesourcery.com>
25660 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25662 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
25663 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
25664 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
25666 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
25667 attribute type list for neon_multiply.
25668 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
25669 attribute type list for neon_multiply.
25670 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
25671 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
25672 attribute type list for neon_multiply.
25673 * config/arm/types.md (crypto_pmull): Add.
25674 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
25675 attribute type list.
25677 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
25679 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
25682 2017-06-20 Jakub Jelinek <jakub@redhat.com>
25684 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
25685 to make sure not to dereference a NULL cost_classes_ptr pointer.
25687 2017-06-20 Carl Love <cel@us.ibm.com>
25689 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
25690 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
25691 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
25692 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
25693 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
25694 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
25695 VMULOSW): New enum "unspec" values.
25696 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
25697 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
25698 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
25699 altivec_vmulosw): New patterns.
25700 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
25701 VMULOSW): Add definitions.
25703 2017-06-20 Julia Koval <julia.koval@intel.com>
25705 * config/i386/i386.c: Fix rounding expand for new pattern.
25706 * config/i386/subst.md: Fix pattern (parallel -> unspec).
25708 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
25710 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
25711 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
25713 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
25715 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
25718 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
25720 * config/aarch64/aarch64-cores.def: Rearrange to sort by
25721 architecture, then by implementer ID.
25722 * config/aarch64/aarch64-tune.md: Regenerate.
25724 2017-06-20 Richard Biener <rguenther@suse.de>
25726 PR middle-end/81097
25727 * fold-const.c (split_tree): Fold to type before negating.
25729 2017-06-20 David Malcolm <dmalcolm@redhat.com>
25731 * diagnostic-show-locus.c
25732 (selftest::test_fixit_deletion_affecting_newline): New function.
25733 (selftest::diagnostic_show_locus_c_tests): Call it.
25735 2017-06-20 Andreas Schwab <schwab@suse.de>
25738 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
25741 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25743 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
25744 __ARM_FEATURE_COPROC according to support.
25746 2017-06-20 Jakub Jelinek <jakub@redhat.com>
25748 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
25749 Rewritten to avoid overflow for > 32-bit pointers.
25752 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
25753 by removing enum keyword.
25754 (ubsan_type_descriptor): Likewise. Formatting fix.
25757 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
25758 splitter): Require TARGET_SSE2 in the condition.
25760 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
25763 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
25764 for doing vector set of SFmode on ISA 3.0.
25765 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
25766 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
25768 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
25769 SFmode value into a V4SF variable that was extracted from another
25770 V4SF variable without converting the element to double precision
25771 and back to single precision vector format.
25772 (vsx_insert_extract_v4sf_p9_2): Likewise.
25774 2017-06-19 Jakub Jelinek <jakub@redhat.com>
25776 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
25777 in UWHI to avoid undefined overflow.
25780 * ubsan.h (enum ubsan_encode_value_phase): New.
25781 (ubsan_encode_value): Change second argument to
25782 enum ubsan_encode_value_phase with default value of
25783 UBSAN_ENCODE_VALUE_GENERIC.
25784 * ubsan.c (ubsan_encode_value): Change second argument to
25785 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
25786 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
25787 create_tmp_var_raw instead of create_tmp_var and use a
25789 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
25790 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
25791 ubsan_encode_value callers.
25794 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
25795 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
25796 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
25798 2017-06-19 Richard Biener <rguenther@suse.de>
25800 PR middle-end/81118
25801 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
25802 estimates if we changed anything.
25804 2017-06-19 Richard Biener <rguenther@suse.de>
25806 PR tree-optimization/80887
25807 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
25808 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
25809 simplified lookups, then reset mprts_hook.
25810 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
25812 (try_to_simplify): Likewise.
25814 2017-06-19 Martin Liska <mliska@suse.cz>
25817 * gimplify.c (gimplify_switch_expr):
25818 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
25820 2017-06-19 Martin Liska <mliska@suse.cz>
25822 * doc/install.texi: Document that PGO runs in 4 stages.
25824 2017-06-19 Martin Liska <mliska@suse.cz>
25827 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
25828 to dispatcher function name.
25829 * multiple_target.c (replace_function_decl): New function.
25830 (create_dispatcher_calls): Redirect both edges and references.
25832 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
25834 * profile-count.c (profile_count::dump): Dump quality.
25835 (profile_count::differs_from_p): Update for unsigned val.
25836 * profile-count.h (profile_count_quality): New enum.
25837 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
25839 2017-06-19 Richard Biener <rguenther@suse.de>
25841 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
25842 struct function as arg.
25843 (estimate_numbers_of_iterations): Export overload with loop arg.
25844 (free_numbers_of_iterations_estimates_loop): Use an overload of
25845 free_numbers_of_iterations_estimates instead.
25846 * tree-cfg.c (remove_bb): Adjust.
25847 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
25848 * tree-parloops.c (gen_parallel_loop): Likewise.
25849 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
25851 (tree_unroll_loops_completely): Likewise.
25852 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
25853 Use an overload instead and export.
25854 (estimated_loop_iterations): Adjust.
25855 (max_loop_iterations): Likewise.
25856 (likely_max_loop_iterations): Likewise.
25857 (estimate_numbers_of_iterations): Take struct function as arg
25859 (loop_exits_before_overflow): Adjust.
25860 (free_numbers_of_iterations_estimates_loop): Use an overload.
25861 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
25862 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
25864 2017-06-19 Richard Biener <rguenther@suse.de>
25867 * ipa-prop.c (find_constructor_constant_at_offset): Handle
25868 RANGE_EXPR conservatively.
25870 2017-06-16 Carl Love <cel@us.ibm.com>
25872 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
25873 definitions for vec_float, vec_float2, vec_floato,
25874 vec_floate built-ins.
25875 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
25876 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
25878 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
25879 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
25880 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
25881 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
25882 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
25883 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
25884 vec_floato): Add builtin defines.
25885 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
25886 Update the built-in documentation file for the new built-in
25889 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25891 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
25892 (mthumb): Mark as the negative of -marm.
25894 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25896 * doc/invoke.texi (ARM Options, -mcpu): Document supported
25898 (ARM Options, -mtune): Document that this accepts the same
25899 extension options as -mcpu.
25900 (ARM Options, -mfpu): Document addition of -mfpu=auto.
25902 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25904 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
25905 permitted extensions.
25907 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25909 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
25910 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
25911 (armv8-m.main): Add option +nodsp.
25912 * config/arm/arm-cpu-cdata.h: Regenerated.
25914 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25916 * config/arm/t-fuchsia: New file.
25917 * config.gcc (arm*-*-fuchsia*): Use it.
25919 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25921 * config/arm/t-symbian: Rewrite for new option infrastructure.
25923 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25925 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
25926 (MULTILIB_REQUIRED): Likewise.
25928 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25930 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
25931 (MULTILIB_RESUE): Likewise.
25932 (MULTILIB_MATCHES): Likewise.
25933 (MULTLIB_REQUIRED): Likewise.
25935 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25937 * config/arm/t-rtems: Rewrite for new option framework.
25939 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25941 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
25942 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
25943 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
25944 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
25945 * config/arm/t-multilib: ... here.
25946 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
25947 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
25948 armv7-a and armv8*-a when A-profile libraries have not been built.
25949 * config/arm/t-rmprofile: Rewrite.
25951 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25953 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
25954 with a backslash. Remove the backslash after substituting unescaped
25956 * doc/fragments.texi (MULTILIB_REUSE): Document it.
25958 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25960 * config.gcc: (arm*-*-*): When building a-profile libraries, force
25961 the driver to pass through the default setting of -mfloat-abi.
25962 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
25964 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
25965 (all_feat_combs): New rule.
25966 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
25968 * config/arm/t-aprofile: Rewrite.
25970 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25972 * config/arm/arm.h (FPUTYPE_AUTO): Define.
25973 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
25974 fpu is not specified by the user/command-line.
25975 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
25976 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
25977 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
25978 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
25979 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
25980 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
25982 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25984 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
25985 * config/arm/t-arm-elf: Rewritten.
25987 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25989 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
25990 have some floating-point instructions.
25991 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
25992 (TARGET_MAYBE_HARD_FLOAT): New macro.
25993 * config/arm/arm-builtins.c (arm_init_builtins): Use
25994 TARGET_MAYBE_HARD_FLOAT.
25995 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
25997 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25999 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
26000 (configargs.h): Include it.
26001 (arm_print_hint_for_fpu_option): New function.
26002 (arm_parse_fpu_option): New function.
26003 (candidate_extension): New class.
26004 (arm_canon_for_multilib): New function.
26005 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
26006 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
26007 (ARCH_CANONICAL_SPECS): New macro.
26008 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
26010 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26012 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
26013 are set after handling multilib fragments. Set target_cpu_default2
26016 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26018 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
26020 (arm*-*-*): Set target_cpu_default2 to a quoted string.
26021 * config/arm/parsecpu.awk (check_cpu): Validate any extension
26023 (check_arch): Likewise.
26024 * config/arm/arm.c (arm_configure_build_target): Handle
26025 TARGET_CPU_DEFAULT being a string constant. Scan any feature
26026 options in the default.
26028 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26030 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
26031 when an option is an alias of another.
26032 * config/arm/parsecpu.awk (optalias): New parser token.
26033 (gen_comm_data): Mark non-alias options as such. Emit entries
26034 for extension aliases.
26035 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
26036 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
26037 (armv6kz, armv6zk, armv6t2): Likewise.
26038 (armv7): Make vfpv3-d16 an alias.
26039 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
26041 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
26042 Sort in canonical order.
26043 (armv8-a): Sort in canonical order.
26044 (armv8.1-a, armv8.2-a): Likewise.
26045 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
26047 (cortex-a9): Sort in canonical order.
26048 * config/arm/arm.c (selftests.h): Include it.
26049 (arm_test_cpu_arch_data): New function.
26050 (arm_run_self_tests): New function.
26051 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
26052 (targetm): Move declaration to the end of the file.
26053 * arm-cpu-cdata.h: Regenerated.
26055 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26057 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
26058 call to target_mode_check describing the type of option passed.
26059 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
26060 (arm_target_thumb_only): Use arm_parse_arch_option_name or
26061 arm_parse_cpu_option_name to match parameters against list of
26063 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
26064 arm_arch_core_flags data structure.
26065 * config/arm/arm-cpu_cdata.h: Regenerated.
26067 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26069 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
26071 (arm_print_hint_for_cpu_option): Likewise.
26072 (arm_print_hint_for_arch_option): Likewise.
26073 (arm_parse_cpu_option_name): Likewise.
26074 (arm_parse_arch_option_name): Likewise.
26075 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
26076 of entries in the all_fpus list.
26077 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
26078 (arm_parse_cpu_option_name): Declare.
26079 (arm_parse_arch_option_name): Declare.
26080 (arm_parse_option_features): Declare.
26081 (arm_intialize_isa): Declare.
26082 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
26084 (gen_comm_data): ... here. Make definitions non-static.
26085 * config/arm/arm-cpu-data.h: Regenerated.
26086 * config/arm/arm-cpu-cdata.h: Regenerated.
26088 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26090 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
26091 (cpu_arch_extension): New structure.
26092 (cpu_arch_option, arch_option, cpu_option): New structures.
26093 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
26094 architecture types.
26095 (gen_data): Generate new format data tables.
26096 * config/arm/arm.c (cpu_tune): New structure.
26097 (cpu_option, processors): Delete.
26098 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
26099 (arm_print_hint_for_cpu_option): ... this and ...
26100 (arm_print_hint_for_arch_option): ... this.
26101 (arm_parse_arch_cpu_name): Delete. Replace with ...
26102 (arm_parse_cpu_option_name): ... this and ...
26103 (arm_parse_arch_option_name): ... this.
26104 (arm_unrecognized_feature): Change type of target parameter to
26106 (arm_parse_arch_cpu_features): Delete. Replace with ...
26107 (arm_parse_option_features): ... this.
26108 (arm_configure_build_target): Rework to use new configuration data
26110 (arm_print_tune_info): Rework for new configuration data tables.
26111 * config/arm/arm-cpu-data.h: Regenerated.
26112 * config/arm/arm-cpu.h: Regenerated.
26114 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26116 * Makefile.in (OBJS): Move sbitmap.o from here ...
26117 (OBJS-libcommon): ... to here.
26119 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26121 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
26122 (ISA_ALL_CRYPTO): New macro.
26123 (ISA_ALL_SIMD): New macro
26124 (ISA_ALL_FP): New macro.
26125 * config/arm/arm.c (fpu_bitlist): Update initializer.
26126 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
26128 (arm9e): Add fpu. Add option for nofp
26129 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
26130 (arm926ej-s, arm1026ej-s): Likewise.
26131 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
26132 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
26133 neon-fp16, neon-vfpv4, nofp and nosimd.
26134 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
26135 (cortex-a8): Add fpu. Add option for nofp.
26136 (cortex-a9): Add fpu. Add options for nosimd and nofp.
26137 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
26138 (cortex-r4f): Add fpu.
26139 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
26140 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
26142 (cortex-r8): Likewise.
26143 (cortex-m4): Add fpu. Add option for nofp.
26144 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
26145 (cortex-a17.cortex-a7): Likewise.
26146 (cortex-a32): Add fpu. Add options for crypto and nofp.
26147 (cortex-a35, cortex-a53): Likewise.
26148 (cortex-a57): Add fpu. Add option for crypto.
26149 (cortex-a72, cortex-a73): Likewise.
26150 (exynos-m1): Likewise.
26151 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
26152 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
26153 (cortex-m33): Add fpu. Add option for nofp.
26154 * config/arm/arm-cpu-cdata.h: Regenerated
26155 * config/arm/arm-cpu-data.h: Regenerated.
26157 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26159 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
26160 (armv5te, armv5tej): Likewise.
26161 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
26162 (armv7): Add options fp and vfpv3-d16.
26163 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
26164 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
26166 (armv7ve): Likewise.
26167 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
26168 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
26169 (armv8-a): Add nocrypto option.
26170 (armv8.1-a, armv8.2-a): Likewise.
26171 (armv8-m.main): add options fp, fp.dp and nofp.
26173 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26175 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
26177 (armv8-a+crc): Delete.
26178 (armv8.1-a): Add options simd, crypto and nofp.
26179 (armv8.2-a): Add options fp16, simd, crypto and nofp.
26180 (armv8.2-a+fp16): Delete.
26181 (armv8-m.main): Add option dsp.
26182 (armv8-m.main+dsp): Delete.
26183 (cortex-a8): Add fpu. Add nofp option.
26184 (cortex-a9): Add fpu. Add nofp and nosimd options.
26185 * config/arm/parsecpu.awk (gen_data): Generate option tables and
26186 link to main cpu and architecture data structures.
26187 (gen_comm_data): Only put isa attributes from the main architecture
26189 (option): New statement for architecture and CPU entries.
26190 * arm.c (struct cpu_option): New structure.
26191 (struct processors): Add entry for options.
26192 (arm_unrecognized_feature): New function.
26193 (arm_parse_arch_cpu_name): Ignore any characters after the first
26195 (arm_parse_arch_cpu_feature): New function.
26196 (arm_configure_build_target): Separate out any CPU and architecture
26197 features and parse separately. Don't error out if -mfpu=auto is
26198 used with only an architecture string.
26199 (arm_print_asm_arch_directives): New function.
26200 (arm_file_start): Call it.
26201 * config/arm/arm-cpu-cdata.h: Regenerated.
26202 * config/arm/arm-cpu-data.h: Likewise.
26203 * config/arm/arm-tables.opt: Likewise.
26205 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26207 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
26208 assembler when it is not -mfpu=auto.
26210 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26212 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
26213 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
26214 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
26215 (ASM_CPU_SPEC): Rewrite.
26216 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
26217 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
26218 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
26219 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
26220 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
26221 copied string is NUL-terminated. Also strip any characters prefixed
26223 (arm_rewrite_selected_arch): New function.
26224 (arm_rewrite_march): New function.
26226 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26228 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
26229 (x_arm_cpu_string, x_arm_tune_string): Likewise.
26230 (march, mcpu, mtune): Convert to string-based options.
26231 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
26232 (arm_parse_arch_cpu_name): New function.
26233 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
26234 identify selected architecture or CPU.
26235 (arm_option_save): New function.
26236 (TARGET_OPTION_SAVE): Redefine.
26237 (arm_option_restore): Restore string options.
26238 (arm_option_print): Print string options.
26240 2017-06-16 Martin Sebor <msebor@redhat.com>
26242 PR tree-optimization/80933
26243 PR tree-optimization/80934
26244 * builtins.c (fold_builtin_3): Do not handle bcmp here.
26245 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
26246 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
26247 (gimple_fold_builtin): Call them.
26249 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26251 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
26252 as unlikely; update profile.
26254 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26256 * predict.c (force_edge_cold): Handle declaring edges impossible
26259 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26261 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
26263 (try_unroll_loop_completely): Fix reporting.
26265 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26267 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
26269 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
26272 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
26273 if given a non-constant argument for an intrinsic which requires a
26276 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26278 * profile.c (compare_freqs): New function.
26279 (branch_prob): Sort edge list.
26280 (find_spanning_tree): Assume that the list is priority sorted.
26282 2017-06-16 Richard Biener <rguenther@suse.de>
26284 PR tree-optimization/81090
26285 * passes.def (pass_record_bounds): Remove.
26286 * tree-pass.h (make_pass_record_bounds): Likewise.
26287 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
26288 make_pass_record_bounds): Likewise.
26289 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
26290 not free niter estimates at the beginning but at the end.
26291 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
26293 2017-06-16 Richard Biener <rguenther@suse.de>
26295 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
26296 initializer to workaround ICE in host GCC 4.8.
26298 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26300 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
26302 (clone_inlined_nodes): Update.
26304 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
26306 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
26307 prefetch settings, and enable prefetching by default at -O3.
26309 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
26311 * config/aarch64/aarch64.c (aarch64_override_options_internal):
26312 Set flag_prefetch_loop_arrays according to tuning data.
26314 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
26316 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
26317 New tune structure.
26318 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
26319 [Unrelated to main purpose of the patch] Place the pointer field last
26320 to enable type checking errors when tune structure are wrongly merged.
26321 * config/aarch64/aarch64.c (generic_prefetch_tune,)
26322 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
26323 (thunderx2t99_prefetch_tune): New tune constants.
26324 (tune_params *_tunings): Update all tunings (no functional change).
26325 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
26326 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
26327 from tunings structures.
26329 2017-06-16 Jakub Jelinek <jakub@redhat.com>
26332 * ubsan.c (instrument_null): Add T argument, use it instead
26333 of computing it based on IS_LHS.
26334 (instrument_object_size): Likewise.
26335 (pass_ubsan::execute): Adjust instrument_null and
26336 instrument_object_size callers to pass gimple_get_lhs or
26337 gimple_assign_rhs1 result to it. Use instrument_null instead of
26338 calling get_base_address and instrument_mem_ref. Handle
26339 aggregate call arguments for object-size sanitization.
26341 2017-06-16 Yury Gribov <tetra2005@gmail.com>
26343 PR tree-optimization/81089
26344 * tree-vrp.c (is_masked_range_test): Validate operands of
26347 2017-06-15 Martin Sebor <msebor@redhat.com>
26350 * dumpfile.c (dump_register): Avoid calling memset to initialize
26351 a class with a default ctor.
26352 * gcc.c (struct compiler): Remove const qualification.
26353 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
26354 * hash-table.h: Ditto.
26355 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
26357 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
26358 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
26360 * params.h (struct param_info): Make struct members non-const.
26361 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
26362 with default initialization.
26363 * vec.h (vec_copy_construct, vec_default_construct): New helper
26365 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
26366 with vec_copy_construct.
26367 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
26368 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
26369 * doc/invoke.texi (-Wclass-memaccess): Document.
26371 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26373 * emit-rtl.h (is_leaf): Update comment about local
26374 register allocator.
26376 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
26379 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
26380 for a variable to have a section before checking if the section has a
26382 Set section to.persistent if persistent attribute is set.
26383 Warn if .persistent attribute is used on an automatic variable.
26385 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
26387 PR rtl-optimization/80474
26388 * reorg.c (update_block): Do not ignore instructions in a delay slot.
26390 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
26392 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
26395 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
26397 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
26398 (casesi): Emit bounds checking as RTL.
26399 (casesi_internal_mips16_<mode>): Remove bounds checking.
26401 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
26403 * config/xtensa/xtensa.c (xtensa_option_override): Append
26404 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
26405 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
26406 xtensa_doloop_hooks): Define unconditionally.
26407 (xtensa_reorg_loops): Only call reorg_loops in the presence of
26409 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
26410 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
26411 for it in xtensa_option_override.
26412 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
26413 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
26415 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
26417 * doc/cppopts.texi: Document '-' special value to -MF.
26419 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
26421 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
26422 (cortex_a53_fconst): Likewise.
26423 (cortex_a53_fpmul): Likewise.
26424 (cortex_a53_f_load_64): Likewise.
26425 (cortex_a53_f_load_many): Likewise.
26426 (cortex_a53_advsimd_alu): Likewise.
26427 (cortex_a53_advsimd_alu_q): Likewise.
26428 (cortex_a53_advsimd_mul): Likewise.
26429 (cortex_a53_advsimd_mul_q): Likewise.
26430 (fpmac bypass): Add new bypass for fpmac-fpmac case.
26431 Add missing fmul, r2f_cvt and fconst cases.
26433 2017-06-14 Richard Biener <rguenther@suse.de>
26435 PR middle-end/81088
26436 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
26438 (fold_binary_loc): When associating do not treat pre-existing
26439 TREE_OVERFLOW on literal constants as a reason to allow
26440 TREE_OVERFLOW on associated literal constants.
26442 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
26444 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
26445 (MASK_FEATURES): New macro.
26446 * config/sparc/sparc.c (sparc_option_override): Remove the special
26447 handling of -mfpu and generalize it to all MASK_FEATURES switches.
26449 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
26451 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
26452 a division of 0 if non-call exceptions are enabled.
26454 2017-06-14 Andrew Pinski <apinski@cavium.com>
26455 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26458 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
26459 Improve vector initialization code gen for only variable case.
26461 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
26463 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
26465 2017-06-14 Richard Biener <rguenther@suse.de>
26467 PR tree-optimization/81083
26468 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
26471 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26473 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
26474 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
26475 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
26476 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
26477 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
26478 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
26480 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26482 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
26483 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
26485 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26487 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
26489 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26491 * config/rs6000/t-rtems: Don't handle SPE.
26493 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26495 * config/rs6000/t-linux: Don't handle SPE.
26497 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26499 * config/rs6000/eabispe.h: Delete file.
26501 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26503 * config/rs6000/t-spe: Delete file.
26505 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26507 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
26508 (rs6000_legitimate_offset_address_p): Return false for anything in
26509 V2SImode or V2SFmode.
26511 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26513 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
26514 except V2SF and V2SI. Rearrange the vector modes, and add comments.
26515 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
26517 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
26518 (rs6000_legitimate_offset_address_p): Ditto.
26519 (rs6000_emit_move): Ditto.
26520 (rs6000_init_builtins): Remove V4HI_type_node.
26522 2017-06-13 Martin Liska <mliska@suse.cz>
26525 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
26526 (gate_asan): Likewise.
26527 * asan.h (asan_no_sanitize_address_p): Remove the function.
26528 (sanitize_flags_p): New function.
26529 * builtins.def: Fix coding style.
26530 * common.opt: Use renamed enum value.
26531 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
26532 * doc/extend.texi: Document no_sanitize attribute.
26533 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
26534 to SANITIZE_UNDEFINED_NONDEFAULT.
26535 * gcc.c (sanitize_spec_function): Use the renamed enum value.
26536 * gimple-fold.c (optimize_atomic_compare_exchange_p):
26537 Use sanitize_flags_p.
26538 * gimplify.c (gimplify_function_tree): Likewise.
26539 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
26540 * opts.c (parse_no_sanitize_attribute): New function.
26541 (common_handle_option): Use renamed enum value.
26542 * opts.h (parse_no_sanitize_attribute): Declare.
26543 * tree.c (sanitize_flags_p): New function.
26544 * tree.h: Declared here.
26545 * tsan.c: Use sanitize_flags_p.
26546 * ubsan.c (ubsan_expand_null_ifn): Likewise.
26547 (instrument_mem_ref): Likewise.
26548 (instrument_bool_enum_load): Likewise.
26549 (do_ubsan_in_current_function): Remove the function.
26550 (pass_ubsan::execute): Use sanitize_flags_p.
26551 * ubsan.h: Remove do_ubsan_in_current_function
26552 * tree-cfg.c (print_no_sanitize_attr_value): New function.
26553 (dump_function_to_file): Use it here.
26555 2017-06-13 Martin Jambor <mjambor@suse.cz>
26557 PR tree-optimization/80803
26558 PR tree-optimization/81063
26559 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
26560 (propagate_subaccesses_across_link): Enqueue subtree whenever
26561 necessary instead of relying on the caller.
26563 2017-06-13 Martin Jambor <mjambor@suse.cz>
26565 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
26566 that have a first_link.
26567 (sort_and_splice_var_accesses): Do not check first_link before
26569 (subtree_mark_written_and_enqueue): Likewise.
26570 (propagate_all_subaccesses): Likewise and do not stop at first
26571 parent with a first_link.
26573 2017-06-13 Martin Jambor <mjambor@suse.cz>
26575 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
26578 2017-06-13 Yury Gribov <tetra2005@gmail.com>
26580 * match.pd: New pattern.
26582 2017-06-13 Yury Gribov <tetra2005@gmail.com>
26584 * tree-vrp.c (is_masked_range_test): New function.
26585 (register_edge_assert_for): Determine ranges for
26588 2017-06-13 Yury Gribov <tetra2005@gmail.com>
26590 PR tree-optimization/67328
26591 * fold-const.c (maskable_range_p): New function.
26592 (build_range_check): Generate bittests if possible.
26594 2017-06-13 Martin Liska <mliska@suse.cz>
26596 * gimple-pretty-print.c (dump_probability): Add new argument.
26597 (dump_edge_probability): Dump both probability and count.
26598 (dump_gimple_label): Likewise.
26599 (dump_gimple_bb_header): Likewise.
26601 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
26604 * config/avr/avr-devices.c: Fix indentation.
26605 * config/avr/gen-avr-mmcu-specs.c: Dito.
26607 2017-06-13 Richard Biener <rguenther@suse.de>
26609 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
26610 instead get vector type from stmt_info.
26611 (vectorizable_reduction): Adjust. Remove dead code.
26613 2017-06-13 Richard Biener <rguenther@suse.de>
26615 PR middle-end/81065
26616 * fold-const.c (extract_muldiv_1): Remove bogus distribution
26617 case of C * (x * C2 + C3).
26618 (fold_addr_of_array_ref_difference): Properly fold index difference.
26620 2017-06-12 David S. Miller <davem@davemloft.net>
26623 * config/sparc/sparc.md (return expander): Emit frame blockage if
26624 function uses alloca.
26626 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
26628 * combine.c (make_field_assignment): Check len rather than the mode
26629 precision when calling force_to_mode.
26631 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
26633 Support multilibs and devices that see flash in RAM address range.
26636 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
26637 (avr_mcu_t) <flash_pm_offset>: New field.
26638 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
26639 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
26640 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
26641 (AVR_TINY_PM_OFFSET): Remove macro.
26642 * config/avr/avr.opt (-mshort-calls): New option.
26643 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
26644 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
26645 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
26646 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
26647 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
26648 instead of avr_arch->have_jmp_call.
26649 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
26650 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
26651 avr_arch->flash_pm_offset to define.
26652 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
26653 new field flash_pm_offset. Add entry for avrxmega3.
26654 (avr_texinfo): Add entry for avrxmega3.
26655 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
26656 attiny212, attiny214,
26657 attiny412, attiny414, attiny416, attiny417,
26658 attiny814, attiny816, attiny817,
26659 attiny1614, attiny1616, attiny1617,
26660 attiny3214, attiny3216, attiny3217.
26661 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
26662 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
26663 (avr_print_operand_address) [AVR_TINY]: Same.
26664 (avr_asm_init_sections) <readonly_data_section>: Only patch
26665 callback if avr_arch->flash_pm_offset = 0.
26666 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
26667 for rodata if avr_arch->flash_pm_offset != 0.
26668 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
26669 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
26670 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
26671 (m_options): Append opt_rcall.
26672 (m_dirnames): Append dir_rcall.
26673 * config/avr/t-multilib: Regenerate.
26675 * configure.ac [target=avr]: Check whether avrxmega3 default
26676 linker description file works as needed.
26677 * configure: Regenerate.
26678 * doc/avr-mmcu.texi: Regenerate.
26679 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
26680 <__AVR_ARCH__>: Document avrxmega3 and 103.
26681 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
26682 <__AVR_SHORT_CALLS__>: Document it.
26683 <__AVR_PM_BASE_ADDRESS__>: Document it.
26684 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
26685 (AVR Variable Attributes) <progmem>: Document this is
26686 not needed for avrxmega3.
26687 (AVR Named Address Spaces) <__flash>: Dito.
26689 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
26691 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
26693 2017-06-12 Doug Rupp <rupp@adacore.com>
26695 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
26696 Append vxworks-stdint.h to the tm_file list.
26697 * config/vxworks-stdint.h: New file.
26699 2017-06-12 Martin Liska <mliska@suse.cz>
26701 PR tree-optimization/81041
26702 * tree-profile.c (gimple_gen_ic_func_profiler):
26703 Create an extra BB in profile-generate
26704 (gimple_gen_time_profiler): Likewise.
26706 2017-06-12 Jakub Jelinek <jakub@redhat.com>
26708 PR tree-optimization/81003
26709 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
26710 (update_range_test): Use it instead of force_gimple_operand_gsi.
26712 2017-06-12 Richard Biener <rguenther@suse.de>
26714 PR tree-optimization/81053
26715 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
26716 with backedge value not defined in loop. Simplify def stmt
26719 2017-06-11 Tom de Vries <tom@codesourcery.com>
26722 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
26724 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
26725 nvptx_cannot_force_const_mem.
26727 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
26729 * opts.c (finish_options): Move test for flag_split_stack after
26730 it has been initialized.
26732 2017-06-11 Jason Merrill <jason@redhat.com>
26734 * tree.h (id_equal): New.
26735 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
26736 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
26737 instead of strcmp of IDENTIFIER_POINTER.
26739 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
26741 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
26742 (mark_all_inlined_calls_cdtor): Fix formating.
26743 (inline_transform): Rescale profile before inlining.
26745 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
26747 * cgraph.h (cgraph_edge::clone): Update prototype.
26748 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
26749 (cgraph_node::create_clone): Update.
26750 (cgraph_node::create_version_clone): Update.
26751 * tree-inline.c (copy_bb): Update.
26752 (expand_call_inline): Update.
26754 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
26756 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
26757 factored out from ...
26758 (rs6000_emit_prologue): ... here.
26760 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
26762 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
26763 factored out from ...
26764 (rs6000_emit_prologue): ... here.
26766 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
26768 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
26770 (handle_missing_profiles): Fix computation of tp_first_run.
26771 (counts_to_freqs): Do not touch freqs when count is 0.
26773 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
26775 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
26778 2017-06-10 Tom de Vries <tom@codesourcery.com>
26780 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
26781 attributes): Document signal effective target.
26783 2017-06-10 Tom de Vries <tom@codesourcery.com>
26785 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
26786 Document effective target stack_size.
26788 2017-06-09 David Malcolm <dmalcolm@redhat.com>
26790 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
26791 to the edit_context if they can be auto-applied.
26793 2017-06-9 Ian Lance Taylor <iant@golang.org>
26795 * opts.c (finish_options): If -fsplit-stack, disable implicit
26796 -forder-blocks-and-partition.
26797 * doc/invoke.texi (Optimize Options): Document that when using
26798 -fsplit-stack -forder-blocks-and-partition is not implicitly
26801 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
26803 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
26804 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
26805 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
26806 * builtins.def (abort, trap, unreachable): Declare cold.
26807 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
26808 * tree-core.h (ECF_COLD): New.
26809 * tree.c (set_call_expr_flags): Handle ECF_COLD.
26810 (build_common_builtin_nodes): Mark unreachable and abort as cold.
26812 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
26814 * predict.c (unlikely_executed_stmt_p): Cleanup.
26816 2017-06-09 Richard Biener <rguenther@suse.de>
26818 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
26819 model if the ref is always written to.
26821 2017-06-09 Tamar Christina <tamar.christina@arm.com>
26823 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
26825 2017-06-09 Tamar Christina <tamar.christina@arm.com>
26827 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
26830 2017-06-09 Tom de Vries <tom@codesourcery.com>
26833 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
26834 "target cannot support label values" when encountering LABEL_REF.
26836 2017-06-09 Martin Liska <mliska@suse.cz>
26838 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
26839 (gimple_gen_ic_func_profiler): Emit direct comparison
26840 of __gcov_indirect_call_callee with NULL.
26841 (gimple_gen_time_profiler): Change probability from
26842 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
26844 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
26846 * profile.c (edge_gcov_counts): Turn to pointer.
26847 (compute_branch_probabilities, compute_branch_probabilities): Update.
26848 (branch_prob): Do not clear edge_gcov_count.
26849 * profile.h (edge_gcov_counts): Turn to pointer.
26850 (edge_gcov_count): Update.
26852 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
26854 * gimple.h (gimple_check_failed): Mark cold.
26856 2017-06-09 Richard Biener <rguenther@suse.de>
26858 PR tree-optimization/66623
26859 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
26860 refactor check_reduction into two parts, properly computing
26861 whether we have to check reduction validity for outer loop
26864 2017-06-09 Richard Biener <rguenther@suse.de>
26866 PR tree-optimization/79483
26867 * graphite-scop-detection.c (order): New global.
26868 (get_order): Compute bb to order mapping that satisfies code
26869 generation constraints.
26870 (cmp_pbbs): New helper.
26871 (build_scops): Start domwalk at entry block, sort generated
26874 2017-06-09 Richard Biener <rguenther@suse.de>
26876 PR middle-end/81007
26877 * ipa-polymorphic-call.c
26878 (ipa_polymorphic_call_context::restrict_to_inner_class):
26879 Skip FIELD_DECLs with error_mark_node type.
26880 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
26883 2017-06-09 Martin Liska <mliska@suse.cz>
26885 * predict.c (struct branch_predictor): New struct.
26886 (test_prediction_value_range): New test.
26887 (predict_c_tests): New function.
26888 * selftest-run-tests.c (selftest::run_tests): Run the function.
26889 * selftest.h: Declare new tests.
26891 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
26894 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
26895 gen_add3_insn did not fail.
26896 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
26897 r0, construct that number in a temporary reg and add that reg to r0.
26898 If asked to put the result in r0 as well, fail.
26900 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
26902 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
26903 for early expansion of vec_eqv.
26905 2017-06-08 Jakub Jelinek <jakub@redhat.com>
26907 PR middle-end/81005
26908 * ubsan.c (instrument_null): Avoid pointless code temporary.
26909 (pass_ubsan::execute): Instrument aggregate arguments of calls.
26911 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
26915 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
26918 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
26921 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
26923 * predict.c (unlikely_executed_edge_p): Move ahead.
26924 (probably_never_executed_edge_p): Use it.
26926 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
26928 PR middle-end/79988
26929 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
26930 gimple_call_builtin_p call.
26932 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
26934 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
26935 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
26936 rtl_check_failed_type2, rtl_check_failed_code1,
26937 rtl_check_failed_code2, rtl_check_failed_code_mode,
26938 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
26939 rtvec_check_failed_bounds, rtl_check_failed_flag,
26940 _fatal_insn_not_found, _fatal_insn): Likewise.
26941 * tree.h (tree_contains_struct_check_failed,
26942 tree_check_failed, tree_not_check_failed,
26943 tree_class_check_failed, tree_range_check_failed,
26944 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
26945 tree_vec_elt_check_failed, phi_node_elt_check_failed,
26946 tree_operand_check_failed, omp_clause_check_failed,
26947 omp_clause_operand_check_failed, omp_clause_range_check_failed):
26950 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
26952 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
26953 flag_branch_probabilities.
26954 * ipa-inline.c (edge_badness): Likewise.
26955 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
26956 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
26957 * predict.c (maybe_hot_frequency_p): Likewise.
26958 (probably_never_executed): Likewise.
26959 * sched-ebb.c (schedule_ebbs): Likewise.
26960 * sched-rgn.c (find_single_block_region): Likewise.
26961 * tracer.c (tail_duplicate): Likewise.
26963 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
26965 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
26966 longer requires x_flag_profile_use.
26968 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
26970 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
26971 instead of flag_reorder_blocks_and_partition.
26972 * dbxout.c (dbxout_function_end): Likewise.
26973 * dwarf2out.c (gen_subprogram_die): Likewise.
26974 * haifa-sched.c (sched_create_recovery_edges): Likewise.
26975 * hw-doloop.c (reorg_loops): Likewise.
26976 * varasm.c (assemble_start_function,
26977 assemble_end_function): Likewise.
26978 (decide_function_section): Do not check for
26979 flag_reorder_blocks_and_partition.
26981 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
26983 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
26985 (chkp_get_hard_register_fake_addr_expr): Ditto.
26986 (chkp_build_addr_expr): Add check for hard reg case.
26987 (chkp_parse_array_and_component_ref): Ditto.
26988 (chkp_find_bounds_1): Ditto.
26989 (chkp_process_stmt): Don't generate bounds store for
26992 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
26994 * predict.c (maybe_hot_bb_p): Do not check profile status.
26995 (maybe_hot_edge_p): Likewise.
26996 (probably_never_executed): Check for zero counts even if profile
26998 (unlikely_executed_edge_p): New function.
26999 (unlikely_executed_stmt_p): New function.
27000 (unlikely_executed_bb_p): New function.
27001 (set_even_probabilities): Use unlikely predicates.
27002 (combine_predictions_for_bb): Likewise.
27003 (predict_paths_for_bb): Likewise.
27004 (predict_paths_leading_to_edge): Likewise.
27005 (determine_unlikely_bbs): New function.
27006 (estimate_bb_frequencies): Use it.
27007 (compute_function_frequency): Use zero counts even if profile is
27009 * profile-count.h: Fix typo.
27011 2017-08-08 Julia Koval <julia.koval@intel.com>
27013 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
27014 _mm512_mask_cvtsepi16_storeu_epi8,
27015 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
27016 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
27017 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
27018 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
27019 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
27020 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
27021 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
27022 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
27023 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
27024 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
27025 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
27026 __builtin_ia32_pmovuswb256mem_mask,
27027 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
27028 __builtin_ia32_pmovwb512mem_mask): New builtins.
27030 2017-08-08 Julia Koval <julia.koval@intel.com>
27032 PR target/73350,80862
27033 * config/i386/subst.md (round): Fix round pattern.
27034 * config/i386/i386.c (ix86_erase_embedded_rounding):
27035 Fix erasing rounding for the fixed pattern.
27037 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27039 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
27041 2017-06-08 Martin Liska <mliska@suse.cz>
27043 PR gcov-profile/80911
27044 * gcov.c (block_info::block_info): New constructor.
27046 2017-06-07 Carl Love <cel@us.ibm.com>
27048 * config/rs6000/rs6000-c: The return type of the following
27049 built-in functions was implemented as int not long long. Fix sign
27050 of return value for the unsigned version of vec_mulo and vec_mule.
27051 vector unsigned long long vec_bperm (vector unsigned long long,
27052 vector unsigned char)
27053 vector signed long long vec_mule (vector signed int,
27055 vector unsigned long long vec_mule (vector unsigned int,
27056 vector unsigned int)
27057 vector signed long long vec_mulo (vector signed int,
27059 vector unsigned long long vec_mulo (vector unsigned int,
27060 vector unsigned int)
27061 * doc/extend.texi: Fix the documentation for the built-in
27064 2017-06-07 Carl Love <cel@us.ibm.com>
27067 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
27070 2017-06-07 Carl Love <cel@us.ibm.com>
27072 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
27073 support, Generate doublehv for signed int/float for BE case only.
27075 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
27077 * doc/invoke.texi (mcx16): Rewrite.
27079 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27081 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
27082 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
27083 *mov<mode>_softfloat, and an anonymous splitter): Use
27084 nonimmediate_operand instead of rs6000_nonimmediate_operand.
27086 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27088 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
27090 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
27091 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
27092 (rs6000_debug_reg_global): Adjust.
27093 (rs6000_init_hard_regno_mode_ok): Adjust.
27094 (rs6000_dbx_register_number): Adjust.
27095 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
27096 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
27097 Remove SPE_ACC and SPEFSCR.
27098 (REG_ALLOC_ORDER): Ditto.
27099 (FRAME_POINTER_REGNUM): Change to 111.
27100 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
27101 (REG_CLASS_NAMES): Ditto.
27102 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
27103 (REGISTER_NAMES): Ditto.
27104 (ADDITIONAL_REG_NAMES): Ditto.
27105 (rs6000_reg_names): Ditto.
27106 * config/rs6000/rs6000.md: Renumber some register number
27109 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27111 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
27113 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
27114 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
27116 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
27117 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
27118 Delete the SPE high registers.
27119 (REG_ALLOC_ORDER): Ditto.
27120 (enum reg_class): Remove SPE_HIGH_REGS.
27121 (REG_CLASS_NAMES): Ditto.
27122 (REG_CLASS_CONTENTS): Delete the SPE high registers.
27123 (REGISTER_NAMES): Ditto.
27124 (rs6000_reg_names): Ditto.
27125 * doc/tm.texi.in: Remove SPE as example.
27126 * doc/tm.texi: Regenerate.
27128 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27130 * config/rs6000/8540.md (ppc8540_brinc): Delete.
27131 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
27132 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
27133 * config/rs6000/rs6000.md (type): Remove "brinc".
27135 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27137 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
27138 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
27139 * config/rs6000/linuxspe.h: Delete file.
27140 * config/rs6000/rs6000.md: Don't include spe.md.
27141 * config/rs6000/spe.h: Delete file.
27142 * config/rs6000/spe.md: Delete file.
27143 * config/rs6000/t-rs6000: Remove spe.md.
27145 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27147 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
27148 (reg_or_none500mem_operand): Delete.
27149 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
27150 instead of reg_or_none500mem_operand.
27152 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27154 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
27155 handling of SPE flags.
27156 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
27158 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27160 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
27162 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
27163 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
27164 paired_divv2sf3): Similar.
27165 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
27166 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
27167 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
27169 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
27170 Rename the paired_* instruction patterns.
27171 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
27173 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
27174 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
27175 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
27176 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
27177 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
27178 PAIRED_VECTOR_MODE.
27179 (struct machine_function): Delete field spe_insn_chain_scanned_p.
27180 (spe_func_has_64bit_regs_p): Delete.
27181 (spe_expand_predicate_builtin): Delete.
27182 (spe_expand_evsel_builtin): Delete.
27183 (TARGET_DWARF_REGISTER_SPAN): Do not define.
27184 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
27185 (invalid_e500_subreg): Delete.
27186 (rs6000_legitimize_address): Always force_reg op2 as well, for
27187 paired single memory accesses.
27188 (rs6000_member_type_forces_blk): Delete.
27189 (rs6000_spe_function_arg): Delete.
27190 (rs6000_expand_unop_builtin): Delete SPE handling.
27191 (rs6000_expand_binop_builtin): Ditto.
27192 (spe_expand_stv_builtin): Delete.
27193 (bdesc_2arg_spe): Delete.
27194 (spe_expand_builtin): Delete.
27195 (spe_expand_predicate_builtin): Delete.
27196 (spe_expand_evsel_builtin): Delete.
27197 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
27198 (spe_init_builtins): Delete.
27199 (spe_func_has_64bit_regs_p): Delete.
27200 (savres_routine_name): Delete "info" parameter. Adjust callers.
27201 (rs6000_emit_stack_reset): Ditto.
27202 (rs6000_dwarf_register_span): Delete.
27203 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
27204 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
27205 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
27207 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
27209 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
27210 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
27211 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
27212 mulv2sf3, divv2sf3): Delete expanders.
27214 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27216 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
27218 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27220 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
27221 * config/rs6000/rs6000.c: Ditto.
27223 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27225 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
27226 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
27227 comparison_operator.
27229 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27231 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
27232 * config/rs6000/rs6000.opt: Ditto.
27233 * config/rs6000/t-rtems: Ditto.
27235 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27237 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
27238 TARGET_E500_SINGLE by 0, simplify.
27239 * config/rs6000/rs6000.c: Ditto.
27240 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
27241 (spe_build_register_parallel): Delete.
27242 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
27243 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
27244 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
27245 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
27246 (E500_CONVERT): Delete.
27247 * config/rs6000/spe.md: Remove many patterns and all define_constants.
27249 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27251 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
27252 * config/rs6000/dfp.md: Ditto.
27253 (negdd2, *negdd2_fpr): Merge.
27254 (absdd2, *absdd2_fpr): Merge.
27255 (negtd2, *negtd2_fpr): Merge.
27256 (abstd2, *abstd2_fpr): Merge.
27257 * config/rs6000/e500.h: Delete file.
27258 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
27259 TARGET_FPRS by 1 and simplify.
27260 * config/rs6000/rs6000-c.c: Ditto.
27261 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
27262 TARGET_DF_SPE by 0.
27263 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
27265 * config/rs6000/rs6000.md: Ditto.
27266 (floatdidf2, *floatdidf2_fpr): Merge.
27267 (move_from_CR_gt_bit): Delete.
27268 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
27269 (E500_CR_IOR_COMPARE): Delete.
27270 (All patterns that require !TARGET_FPRS): Delete.
27271 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
27273 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27275 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
27277 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27279 * graphds.c (add_edge): Intitialize edge's attached data.
27280 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
27281 pointer parameter. Call pointed function on each edge during
27282 graph traversing. Skip traversing the edge when the function
27284 (graphds_dfs, graphds_scc): Ditto.
27285 (for_each_edge): New parameter. Pass the new parameter to callback
27287 * graphds.h (skip_edge_callback): New function pointer type.
27288 (graphds_dfs, graphds_scc): New function pointer parameter.
27289 (graphds_edge_callback, for_each_edge): New parameter.
27291 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27293 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
27294 out code checking if runtime alias check is possible to below ...
27295 Call the new function.
27296 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
27297 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
27299 2017-06-07 Marek Polacek <polacek@redhat.com>
27302 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
27303 TYPE_OVERFLOW_WRAPS check.
27305 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27307 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
27308 if versioning is required.
27309 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
27310 peeling with the check for versioning.
27312 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27314 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
27315 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
27316 Set true to new parameter if new ssa variable is defined.
27317 (vect_gen_vector_loop_niters): Refactor. Set range information
27318 for the new vector loop bound variable.
27319 (vect_do_peeling): Ditto.
27321 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27323 * tree-affine.c (ssa.h): Include header file.
27324 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
27325 has wrapping overflow behavior.
27327 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27329 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
27331 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27333 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
27334 (tree_to_aff_combination): ... here.
27336 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27338 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
27339 reg_pressure model function.
27340 (ivopts_global_cost_for_size): Delete.
27341 (determine_set_costs, iv_ca_recount_cost): Call new model function
27342 ivopts_estimate_reg_pressure.
27344 2017-06-07 Tamar Christina <tamar.christina@arm.com>
27346 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
27347 expensive than udiv. Remove floating point cases from mod.
27349 2017-06-07 Tamar Christina <tamar.christina@arm.com>
27351 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
27352 Increase idiv cost.
27354 2017-06-07 Tamar Christina <tamar.christina@arm.com>
27356 * config/aarch64/aarch64.md
27357 (copysignsf3): Fix mask generation.
27359 2017-06-07 Jakub Jelinek <jakub@redhat.com>
27361 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
27363 (class dump_manager): Add register_dumps method.
27364 * dumpfile.c: Include langhooks.h.
27365 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
27366 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
27367 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
27368 (dump_manager::dump_register): Start with 512 entries instead of 32.
27369 (dump_manager::register_dumps): New method.
27370 * toplev.c (general_init): Instead of invoking register_dumps
27371 langhook, invoke register_dumps method on the dump manager.
27372 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
27375 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
27377 * doc/md.texi: Clarify the restrictions on a define_insn condition.
27378 Say that # requires an associated define_split to exist, and that
27379 the define_split must be suitable for use after register allocation.
27381 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27383 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
27384 (compute_outgoing_frequencies): Also initialize zero counts.
27385 (find_many_sub_basic_blocks): Do not produce uninitialized profile
27386 around loops; preserve more of profile when nothing changes.
27388 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
27390 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
27392 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
27393 * config/arm/arm-cpu-cdata.h: Regenerate.
27394 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
27395 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
27396 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
27397 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
27398 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
27400 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
27402 * config/arm/t-rmprofile: Likewise.
27403 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
27405 2017-06-06 David S. Miller <davem@davemloft.net>
27408 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
27409 blockage if function uses alloca.
27411 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27413 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
27414 New "uid" fields to hold pretty-print IDs of group and ref.
27415 Memory references are now identified as <group_id>:<ref_id>
27416 instead of using [random] addresses.
27417 (dump_mem_details): Simplify, no functional change.
27418 (dump_mem_ref): Simplify and make output more concise.
27419 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
27420 (find_or_create_group): Initialize group uid.
27421 (record_ref): Initialize ref uid. Improve debug output.
27422 (prune_group_by_reuse, should_issue_prefetch_p,)
27423 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
27424 (mark_nontemporal_store, determine_loop_nest_reuse):
27425 Improve debug output.
27427 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27429 * dbgcnt.def (prefetch): New debug counter.
27430 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
27431 (schedule_prefetches): Stop issueing prefetches if debug counter
27434 2017-06-06 Tom de Vries <tom@codesourcery.com>
27436 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
27437 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
27439 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27441 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
27442 Use aarch64_reg_or_zero predicate for operand 4.
27443 (aarch64_compare_and_swap<mode> define_insn_and_split):
27444 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
27445 (aarch64_store_exclusive<mode>): Likewise for operand 2.
27447 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
27449 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
27450 (arm_compute_save_core_reg_mask): This.
27451 (thumb1_compute_save_reg_mask): Rename into ...
27452 (thumb1_compute_save_core_reg_mask): This.
27453 (arm_compute_save_reg0_reg12_mask): Adapt comment.
27454 (arm_compute_frame_layout): Likewise.
27456 2017-06-06 Richard Biener <rguenther@suse.de>
27458 PR tree-optimization/80974
27459 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
27460 keep or clear leaders SSA info.
27462 2017-06-06 Tom de Vries <tom@codesourcery.com>
27464 * config/nvptx/nvptx.c (split_mode_p): New function.
27465 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
27467 2017-06-06 Tom de Vries <tom@codesourcery.com>
27469 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
27471 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27474 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
27477 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27479 * shrink-wrap.c (handle_simple_exit): Update profile.
27480 (try_shrink_wrapping): Upate profile.
27482 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27484 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
27485 (tree_guess_outgoing_edge_probabilities): New.
27486 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
27487 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
27489 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27491 * ipa-split.c (split_function): Initialize return bb profile.
27493 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27495 * profile.c (compute_branch_probabilities): Also initialize
27496 EXIT_BLOCK profile.
27498 2017-06-06 Richard Biener <rguenther@suse.de>
27500 PR tree-optimization/80928
27501 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
27502 (vect_analyze_loop_operations): Properly guard analysis for
27504 (vect_transform_loop): Likewise.
27505 (vect_analyze_loop_2): Also reset SLP type on PHIs.
27506 (vect_model_induction_cost): Do not cost for pure SLP.
27507 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
27508 of induction in inner loop vectorization.
27509 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
27510 (vect_get_and_check_slp_defs): Handle vect_induction_def.
27511 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
27513 (vect_analyze_slp_cost_1): Cost induction.
27514 (vect_detect_hybrid_slp_stmts): Handle PHIs.
27515 (vect_get_slp_vect_defs): Likewise.
27516 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
27517 (vect_transform_stmt): Handle SLP reductions.
27518 * tree-vectorizer.h (vectorizable_induction): Adjust.
27520 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
27522 * config/rs6000/rs6000.c (make_resolver_func): Update
27523 init_lowered_empty_function call.
27525 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
27527 * doc/invoke.texi: Document the -fprofile-abs-path option.
27528 * common.opt (fprofile-abs-path): New option.
27529 * gcov-io.h (gcov_write_filename): Declare.
27530 * gcov-io.c (gcov_write_filename): New function.
27531 * coverage.c (coverage_begin_function): Use gcov_write_filename.
27532 * profile.c (output_location): Likewise.
27534 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
27536 * shring-wrap.c: Revert accidental commit.
27538 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
27540 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
27542 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
27544 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
27546 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
27547 profile in callgraph edge.
27548 * profile-count.h (apply_probability): If THIS is 0, then result is 0
27549 (apply_scale): Likewise.
27550 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
27551 Also scale profile when inlining function with zero profile.
27552 (initialize_cfun): Update exit block profile even when it is zero.
27553 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
27554 when profile is read.
27556 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
27558 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
27559 (CLONE_*): New constants to define the processors we can generate
27560 code for with the target_clone attribute.
27561 (rs6000_clone_map): New array to identify which clone processors
27562 the current program is running on.
27563 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
27564 target_clone attribute.
27565 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
27566 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
27567 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
27568 (cpu_expand_builtin): Add support for target_clone attribute.
27569 (rs6000_valid_attribute_p): Allow "default" attribute.
27570 (get_decl_name): New debug function to simplify printing the
27571 current function name in debugging statements.
27572 (rs6000_clone_priority): New functions to support the target_clone
27573 attribute, and be able to generate code to switch between ISA 2.05
27574 through ISA 3.0 (power6 through power9).
27575 (rs6000_compare_version_priority): Likewise.
27576 (rs6000_get_function_versions_dispatcher): Likewise.
27577 (make_resolver_func): Likewise.
27578 (add_condition_to_bb): Likewise.
27579 (dispatch_function_versions): Likewise.
27580 (rs6000_generate_version_dispatcher_body): Likewise.
27581 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
27582 (fusion_gpr_load_p): Fix a spacing issue.
27583 * doc/extend.texi (Common Function Attributes): Document that the
27584 PowerPC supports the target_clone attribute.
27586 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
27588 * config/arm/arm.h: explain F symbol found in description of ARM
27589 register allocation in its legend.
27591 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
27593 * config/mips/frame-header-opt.c: Include profile-count.h.
27594 * config/riscv/riscv.c: Include profile-count.h
27596 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
27598 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
27600 (sm_set_flag_if_changed): Add bbs field.
27601 (execute_sm_if_changed_flag_set): Pass BBS.
27602 (execute_sm): Update.
27604 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27606 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
27609 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27611 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
27612 (peephole2): New peephole2 to emit the above.
27613 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
27615 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27617 * config/aarch64/aarch64.c (define_peephole2 above
27618 *sub_<shift>_<mode>): New peephole.
27620 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
27622 * config/i386/i386.c (make_resolver_func): Update.
27623 * Makefile.in: Add profile-count.h and profile-count.o
27624 * auto-profile.c (afdo_indirect_call): Update to new API.
27625 (afdo_set_bb_count): Update.
27626 (afdo_propagate_edge): Update.
27627 (afdo_propagate_circuit): Update.
27628 (afdo_calculate_branch_prob): Update.
27629 (afdo_annotate_cfg): Update.
27630 * basic-block.h: Include profile-count.h
27631 (struct edge_def): Turn count to profile_count.
27632 (struct basic_block_def): Likewie.
27633 (REG_BR_PROB_BASE): Move to profile-count.h
27634 (RDIV): Move to profile-count.h
27635 * bb-reorder.c (max_entry_count): Turn to profile_count.
27636 (find_traces): Update.
27637 (rotate_loop):Update.
27638 (connect_traces):Update.
27639 (sanitize_hot_paths):Update.
27640 * bt-load.c (migrate_btr_defs): Update.
27641 * cfg.c (RDIV): Remove.
27642 (init_flow): Use alloc_block.
27643 (alloc_block): Uninitialize count.
27644 (unchecked_make_edge): Uninitialize count.
27645 (check_bb_profile): Update.
27646 (dump_edge_info): Update.
27647 (dump_bb_info): Update.
27648 (update_bb_profile_for_threading): Update.
27649 (scale_bbs_frequencies_int): Update.
27650 (scale_bbs_frequencies_gcov_type): Update.
27651 (scale_bbs_frequencies_profile_count): New.
27652 * cfg.h (update_bb_profile_for_threading): Update.
27653 (scale_bbs_frequencies_profile_count): Declare.
27654 * cfgbuild.c (compute_outgoing_frequencies): Update.
27655 (find_many_sub_basic_blocks): Update.
27656 * cfgcleanup.c (try_forward_edges): Update.
27657 (try_crossjump_to_edge): Update.
27658 * cfgexpand.c (expand_gimple_tailcall): Update.
27659 (construct_exit_block): Update.
27660 * cfghooks.c (verify_flow_info): Update.
27661 (dump_bb_for_graph): Update.
27662 (split_edge): Update.
27663 (make_forwarder_block): Update.
27664 (duplicate_block): Update.
27665 (account_profile_record): Update.
27666 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
27667 (get_estimated_loop_iterations): Update.
27668 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
27669 (single_likely_exit): Update.
27670 * cfgloopmanip.c (scale_loop_profile): Update.
27672 (set_zero_probability): Update.
27673 (lv_adjust_loop_entry_edge): Update.
27674 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
27675 (purge_dead_edges): Update.
27676 (rtl_account_profile_record): Update.
27677 * cgraph.c (cgraph_node::create): Uninitialize count.
27678 (symbol_table::create_edge): Uninitialize count.
27679 (cgraph_update_edges_for_call_stmt_node): Update.
27680 (cgraph_edge::dump_edge_flags): Update.
27681 (cgraph_node::dump): Update.
27682 (cgraph_edge::maybe_hot_p): Update.
27683 * cgraph.h: Include profile-count.h
27684 (create_clone), create_edge, create_indirect_edge): Update.
27685 (cgraph_node): Turn count to profile_count.
27686 (cgraph_edge0: Likewise.
27687 (make_speculative, clone): Update.
27688 (create_edge): Update.
27689 (init_lowered_empty_function): Update.
27690 * cgraphclones.c (cgraph_edge::clone): Update.
27691 (duplicate_thunk_for_node): Update.
27692 (cgraph_node::create_clone): Update.
27693 * cgraphunit.c (cgraph_node::analyze): Update.
27694 (cgraph_node::expand_thunk): Update.
27695 * final.c (dump_basic_block_info): Update.
27696 * gimple-streamer-in.c (input_bb): Update.
27697 * gimple-streamer-out.c (output_bb): Update.
27698 * graphite.c (print_global_statistics): Update.
27699 (print_graphite_scop_statistics): Update.
27700 * hsa-brig.c: Include basic-block.h.
27701 * hsa-dump.c: Include basic-block.h.
27702 * hsa-gen.c (T sum_slice): Update.
27703 (convert_switch_statements):Update.
27704 * hsa-regalloc.c: Include basic-block.h.
27705 * ipa-chkp.c (chkp_produce_thunks): Update.
27706 * ipa-cp.c (struct caller_statistics): Update.
27707 (init_caller_stats): Update.
27708 (gather_caller_stats): Update.
27709 (ipcp_cloning_candidate_p): Update.
27710 (good_cloning_opportunity_p): Update.
27711 (get_info_about_necessary_edges): Update.
27712 (dump_profile_updates): Update.
27713 (update_profiling_info): Update.
27714 (update_specialized_profile): Update.
27715 (perhaps_add_new_callers): Update.
27716 (decide_about_value): Update.
27717 (ipa_cp_c_finalize): Update.
27718 * ipa-devirt.c (struct odr_type_warn_count): Update.
27719 (struct decl_warn_count): Update.
27720 (struct final_warning_record): Update.
27721 (possible_polymorphic_call_targets): Update.
27722 (ipa_devirt): Update.
27723 * ipa-fnsummary.c (redirect_to_unreachable): Update.
27724 * ipa-icf.c (sem_function::merge): Update.
27725 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
27726 * ipa-inline.c (compute_uninlined_call_time): Update.
27727 (compute_inlined_call_time): Update.
27728 (want_inline_small_function_p): Update.
27729 (want_inline_self_recursive_call_p): Update.
27730 (edge_badness): Update.
27731 (lookup_recursive_calls): Update.
27732 (recursive_inlining): Update.
27733 (inline_small_functions): Update.
27734 (dump_overall_stats): Update.
27735 (dump_inline_stats): Update.
27736 * ipa-profile.c (ipa_profile_generate_summary): Update.
27737 (ipa_propagate_frequency): Update.
27738 (ipa_profile): Update.
27739 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
27740 * ipa-utils.c (ipa_merge_profiles): Update.
27741 * loop-doloop.c (doloop_modify): Update.
27742 * loop-unroll.c (report_unroll): Update.
27743 (unroll_loop_runtime_iterations): Update.
27744 * lto-cgraph.c (lto_output_edge): Update.
27745 (lto_output_node): Update.
27746 (input_node): Update.
27747 (input_edge): Update.
27748 (merge_profile_summaries): Update.
27749 * lto-streamer-in.c (input_cfg): Update.
27750 * lto-streamer-out.c (output_cfg): Update.
27751 * mcf.c (create_fixup_graph): Update.
27752 (adjust_cfg_counts): Update.
27753 (sum_edge_counts): Update.
27754 * modulo-sched.c (sms_schedule): Update.
27755 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
27756 * predict.c (maybe_hot_count_p): Update.
27757 (probably_never_executed): Update.
27758 (dump_prediction): Update.
27759 (combine_predictions_for_bb): Update.
27760 (propagate_freq): Update.
27761 (handle_missing_profiles): Update.
27762 (counts_to_freqs): Update.
27763 (rebuild_frequencies): Update.
27764 (force_edge_cold): Update.
27765 * predict.h: Include profile-count.h
27766 (maybe_hot_count_p, counts_to_freqs): UPdate.
27767 * print-rtl-function.c: Do not include cfg.h
27768 * print-rtl.c: Include basic-block.h
27769 * profile-count.c: New file.
27770 * profile-count.h: New file.
27771 * profile.c (is_edge_inconsistent): Update.
27772 (correct_negative_edge_counts): Update.
27773 (is_inconsistent): Update.
27774 (set_bb_counts): Update.
27775 (read_profile_edge_counts): Update.
27776 (compute_frequency_overlap): Update.
27777 (compute_branch_probabilities): Update; Initialize and deinitialize
27779 (branch_prob): Update.
27780 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
27781 (edge_gcov_count): New.
27782 (bb_gcov_count): New.
27783 * shrink-wrap.c (try_shrink_wrapping): Update.
27784 * tracer.c (better_p): Update.
27785 * trans-mem.c (expand_transaction): Update.
27786 (ipa_tm_insert_irr_call): Update.
27787 (ipa_tm_insert_gettmclone_call): Update.
27788 * tree-call-cdce.c: Update.
27789 * tree-cfg.c (gimple_duplicate_sese_region): Update.
27790 (gimple_duplicate_sese_tail): Update.
27791 (gimple_account_profile_record): Update.
27792 (execute_fixup_cfg): Update.
27793 * tree-inline.c (copy_bb): Update.
27794 (copy_edges_for_bb): Update.
27795 (initialize_cfun): Update.
27796 (freqs_to_counts): Update.
27797 (copy_cfg_body): Update.
27798 (expand_call_inline): Update.
27799 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
27800 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
27801 (try_unroll_loop_completely): Update.
27802 (try_peel_loop): Update.
27803 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
27804 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
27805 * tree-ssa-loop-split.c (connect_loops): Update.
27806 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
27807 * tree-ssa-reassoc.c (branch_fixup): Update.
27808 * tree-ssa-tail-merge.c (replace_block_by): Update.
27809 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
27810 (compute_path_counts): Update.
27811 (update_profile): Update.
27812 (recompute_probabilities): Update.
27813 (update_joiner_offpath_counts): Update.
27814 (estimated_freqs_path): Update.
27815 (freqs_to_counts_path): Update.
27816 (clear_counts_path): Update.
27817 (ssa_fix_duplicate_block_edges): Update.
27818 (duplicate_thread_path): Update.
27819 * tree-switch-conversion.c (case_bit_test_cmp): Update.
27820 (struct switch_conv_info): Update.
27821 * tree-tailcall.c (decrease_profile): Update.
27822 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
27823 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
27824 * value-prof.c (check_counter): Update.
27825 (gimple_divmod_fixed_value): Update.
27826 (gimple_mod_pow2): Update.
27827 (gimple_mod_subtract): Update.
27828 (gimple_ic_transform): Update.
27829 (gimple_stringop_fixed_value): Update.
27830 * value-prof.h (gimple_ic): Update.
27832 2017-06-02 Carl Love <cel@us.ibm.com>
27834 * config/rs6000/rs6000-c: Add support for built-in functions
27835 vector double vec_doublee (vector signed int);
27836 vector double vec_doublee (vector unsigned int);
27837 vector double vec_doublee (vector float);
27838 vector double vec_doubleh (vector signed int);
27839 vector double vec_doubleh (vector unsigned int);
27840 vector double vec_doubleh (vector float);
27841 vector double vec_doublel (vector signed int);
27842 vector double vec_doublel (vector unsigned int);
27843 vector double vec_doublel (vector float);
27844 vector double vec_doubleo (vector signed int);
27845 vector double vec_doubleo (vector unsigned int);
27846 vector double vec_doubleo (vector float);.
27847 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
27848 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
27850 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
27851 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
27852 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
27854 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
27855 vec_doublel, vec_doubleh.
27856 * doc/extend.texi: Update the built-in documentation file for the
27857 new built-in functions.
27859 2017-06-02 David Malcolm <dmalcolm@redhat.com>
27862 * ipa-inline-analysis.c (free_growth_caches): Set
27863 edge_removal_hook_holder to NULL after removing it.
27865 2017-06-02 Sudakshina Das <sudi.das@arm.com>
27867 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
27868 comparision with zero.
27870 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
27871 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
27872 for early expansion of vec_min and vec_max builtins.
27873 (builtin_function_type): Add min/max unsigned variants to those
27874 identified as having unsigned arguments.
27876 2017-06-02 Olivier Hainque <hainque@adacore.com>
27878 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
27880 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27882 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
27883 Use VALL_F16 iterator rather than VALL.
27885 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27887 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
27888 Emit CBNZ inside loop when doing a strong exchange and comparing
27889 against zero. Generate the CC flags after the loop.
27891 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
27893 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
27894 (dl_section_ref): New.
27895 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
27896 On AIX, append an expression to subtract the size of the
27897 section length to dl_section_ref.
27899 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
27901 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
27902 for early expansion of vector absolute builtins.
27904 2017-06-02 Richard Biener <rguenther@suse.de>
27906 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
27907 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
27909 2017-06-02 Richard Biener <rguenther@suse.de>
27911 PR tree-optimization/80948
27912 * tree-tailcall.c (find_tail_calls): Track stmts to move in
27913 stmt order as well.
27915 2017-06-02 Richard Biener <rguenther@suse.de>
27917 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
27919 * tree-vect-stmts.c (process_use): Do not mark backedge defs
27920 for inductions as relevant.
27922 2017-06-02 Richard Biener <rguenther@suse.de>
27924 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
27925 (vectorizable_induction): ... this. Remove dead code.
27927 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
27929 * builtins. (expand_builtin_alloca): Remove second parameter and
27930 infer its value from the first parameter instead.
27931 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
27933 2017-06-02 Jakub Jelinek <jakub@redhat.com>
27935 PR rtl-optimization/80903
27936 * loop-doloop.c (add_test): Unshare sequence.
27938 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
27940 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
27942 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
27944 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
27946 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
27947 xlogue_layout::get_instance, logue_layout::xlogue_layout,
27948 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
27949 (xlogue_layout::get_stub_rtx): Make static.
27950 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
27951 (xlogue_layout::compute_stub_managed_regs): Rename to...
27952 (xlogue_layout::count_stub_managed_regs): ...this.
27953 (xlogue_layout::is_stub_managed_reg): New function.
27954 (xlogue_layout::m_stub_names): Rename to...
27955 (xlogue_layout::s_stub_names): ...this, make static.
27956 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
27957 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
27958 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
27959 xlogue_layout::s_stub_names): Instantiate statics.
27960 (stub_managed_regs): Remove.
27961 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
27962 (disable_call_ms2sysv_xlogues): Rename to...
27963 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
27964 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
27966 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
27967 change after reload_completed.
27968 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
27970 (ix86_expand_prologue): Likewise.
27971 (ix86_expand_epilogue): Likewise.
27972 (ix86_expand_split_stack_prologue): Likewise.
27973 (ix86_compute_frame_layout): Remove frame parameter ...
27974 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
27975 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
27977 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
27978 (ix86_frame): Move from here ...
27979 * config/i386/i386.h (ix86_frame): ... to here.
27980 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
27981 complete ix86_frame data structure instead. Remove some_ld_name.
27983 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
27985 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
27986 symbols that hold a DECL_VALUE_EXPR.
27988 2017-06-01 Martin Jambor <mjambor@suse.cz>
27990 PR tree-optimization/80898
27991 * tree-sra.c (process_subtree_disqualification): Removed.
27992 (disqualify_candidate): Do not acll
27993 process_subtree_disqualification.
27994 (subtree_mark_written_and_enqueue): New function.
27995 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
27996 RHS has been disqualified and re-queue LHS if necessary. Apart
27997 from that, ignore disqualified RHS.
27999 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28001 * config/s390/s390.c (s390_emit_epilogue): Disable early return
28002 address fetch for z10 or later.
28004 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28006 * config/arc/arc.md (tst_movb): Add guard when splitting.
28008 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28010 * config/arc/arc.c (arc_can_eliminate): Test against
28011 arc_frame_pointer_needed.
28013 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28015 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
28016 to prevent store reordering.
28017 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
28018 (type): Add block type.
28019 (stack_tie): Define special instruction to be used in
28022 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28024 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
28025 constraint. It is not valid for the pattern.
28026 (noncommutative_binary_comparison): Likewise.
28028 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28030 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
28033 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28035 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
28036 be used by the reg-alloc.
28038 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28040 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
28041 reg-alloc when having mul64 or mul32x16 instructions.
28042 (mulsidi3): Likewise.
28043 (umulsidi3): Likewise.
28044 (mulsi32x16): New pattern.
28045 (mulsi64): Likewise.
28046 (mulsidi64): Likewise.
28047 (umulsidi64): Likewise.
28048 (MUL32x16_REG): Define.
28049 (mul64_600): Use MUL32x16_REG.
28050 (mac64_600): Likewise.
28051 (umul64_600): Likewise.
28052 (umac64_600): Likewise.
28054 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28056 * config/arc/arc.md (mulsi3_700): Make it commutative.
28058 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
28060 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
28062 (*sign_extendsidi2_insn): Likewise for movstosw.
28064 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
28066 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
28067 the type of the input discriminant value. Convert the
28068 discriminant value of signedness vary.
28070 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
28072 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
28073 Add to -Wall section.
28075 2017-06-01 Richard Biener <rguenther@suse.de>
28077 PR middle-end/66313
28078 * fold-const.c (fold_plusminus_mult_expr): If the factored
28079 factor may be zero use a wrapping type for the inner operation.
28080 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
28081 and handle moved defs.
28082 (process_assignment): Properly guard the unary op case. Return a
28083 tri-state indicating that moving the stmt before the call may allow
28084 to continue. Pass through to_move.
28085 (find_tail_calls): Handle moving unrelated defs before
28088 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
28091 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
28092 splitter result in the canonical way.
28094 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
28096 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
28097 also for 32bit target. Update insn attributes.
28098 (zero-extendsidi2 splitter): Allow all registers for operand 1.
28100 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
28102 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
28103 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
28104 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
28105 (_mm_maskz_min_ss): New intrinsics.
28107 2017-05-31 Martin Liska <mliska@suse.cz>
28109 * tree-vect-loop.c (vect_create_epilog_for_reduction):
28110 Change comment style to one we normally use.
28111 (vectorizable_reduction): Likewise.
28112 (vectorizable_induction): Likewise.
28113 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
28114 (vectorizable_call): Likewise.
28115 (vectorizable_simd_clone_call): Likewise.
28116 (vectorizable_conversion): Likewise.
28117 (vectorizable_assignment): Likewise.
28118 (vectorizable_shift): Likewise.
28119 (vectorizable_operation): Likewise.
28120 (vectorizable_store): Likewise.
28121 (vectorizable_load): Likewise.
28122 * tree-vectorizer.h: Likewise.
28124 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
28126 * passes.c (emergency_dump_function): New.
28127 * tree-pass.h (emergency_dump_function): Declare.
28128 * plugin.c (plugins_internal_error_function): Remove.
28129 * plugin.h (plugins_internal_error_function): Remove declaration.
28130 * toplev.c (internal_error_function): New static function. Use it...
28131 (general_init): ...here.
28133 2017-05-31 Graham Markall <graham.markall@embecosm.com>
28135 * config/arc/arc.c (arc_print_operand): Handle constant operands.
28136 (arc_rtx_costs): Add costs for new patterns.
28137 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
28138 * config/arc/predicates.md: Add _1_2_3_operand predicate.
28140 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
28142 * tree-ssa-strlen.c (get_next_strinfo): New function.
28143 (get_stridx_plus_constant): Use it.
28144 (zero_length_string): Likewise.
28145 (adjust_related_strinfos): Likewise.
28146 (adjust_last_stmt): Likewise.
28148 2017-05-31 Richard Biener <rguenther@suse.de>
28151 * config/i386/i386.c (ix86_expand_builtin): Remove assert
28152 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
28154 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
28156 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
28157 loop_vinfo argument and use of dependence distance vectors.
28158 Check instead whether the two references differ only in their
28159 initial value and assume that they have the same alignment if the
28160 difference is a multiple of the vector alignment.
28161 (vect_analyze_data_refs_alignment): Update call accordingly.
28163 2017-05-31 Martin Liska <mliska@suse.cz>
28166 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
28168 2017-05-31 Bin Cheng <bin.cheng@arm.com>
28170 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
28171 (create_intersect_range_checks): Move from ...
28172 * tree-data-ref.c (create_intersect_range_checks_index)
28173 (create_intersect_range_checks): ... to here.
28174 (create_runtime_alias_checks): New function factored from ...
28175 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
28176 here. Call above function.
28177 * tree-data-ref.h (create_runtime_alias_checks): New function.
28179 2017-05-31 Bin Cheng <bin.cheng@arm.com>
28181 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
28182 segment length for dr_b and compute it in wide_int.
28184 2017-05-31 Richard Biener <rguenther@suse.de>
28186 PR tree-optimization/80906
28187 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
28188 and pass through iv_map.
28189 (copy_bb_and_scalar_dependences): Adjust.
28190 (translate_pending_phi_nodes): Likewise.
28191 (copy_loop_close_phi_args): Handle code-generating IVs instead
28194 2017-05-30 David Malcolm <dmalcolm@redhat.com>
28196 * diagnostic-color.c (color_dict): Add "type-diff".
28197 (parse_gcc_colors): Update comment.
28198 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
28199 -fdiagnostics-show-template-tree and -fno-elide-type.
28200 (GCC_COLORS): Add type-diff to example.
28202 (-fdiagnostics-show-template-tree): New.
28203 (-fno-elide-type): New.
28204 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
28205 the pp_format_decoder callback. Call any m_format_postprocessor's
28207 (pretty_printer::pretty_printer): Initialize
28208 m_format_postprocessor.
28209 (pretty_printer::~pretty_printer): Delete any
28210 m_format_postprocessor.
28211 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
28212 (class format_postprocessor): New class.
28213 (struct pretty_printer::format_decoder): Document the new parameters.
28214 (struct pretty_printer::m_format_postprocessor): New field.
28215 * tree-diagnostic.c (default_tree_printer): Update for new
28216 bool and const char ** params.
28217 * tree-diagnostic.h (default_tree_printer): Likewise.
28219 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
28221 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
28222 (lwa_operand): Delete rs6000_gen_cell_microcode test.
28223 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
28224 rs6000_gen_cell_microcode code.
28225 (rs6000_final_prescan_insn): Delete.
28226 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
28227 "warn-cell-microcode" entries.
28228 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
28229 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
28230 throughout. Change cc_reg_not_micro_cr0_operand to
28231 cc_reg_not_cr0_operand throughout.
28232 (*extendhi<mode>2_noload): Delete.
28233 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
28234 (mwarn-cell-microcode): Delete.
28235 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
28236 -mgen-cell-microcode and -mwarn-cell-microcode.
28238 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
28241 * config/i386/constraints.md (Yd): New constraint.
28243 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
28244 and (?Yd, r) alternatives. Update insn attributes.
28245 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
28246 and (?*Yd, r) alternatives. Update insn attributes.
28247 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
28249 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
28251 * gimplify.c (gimplify_modify_expr): Don't create a
28252 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
28255 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
28257 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
28259 2017-05-30 Richard Biener <rguenther@suse.de>
28261 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
28262 and reduc_def fields.
28263 (STMT_VINFO_REDUC_TYPE): New define.
28264 (STMT_VINFO_REDUC_DEF): Likewise.
28265 (vect_force_simple_reduction): Adjust prototype.
28266 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
28267 (vect_is_simple_reduction): Remove check_reduction argument.
28268 (vect_force_simple_reduction): Adjust and set
28269 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
28270 (vectorizable_reduction): Do not re-do reduction analysis
28271 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
28272 * tree-parloops.c (gather_scalar_reductions): Adjust.
28274 2017-05-30 Richard Biener <rguenther@suse.de>
28276 PR middle-end/80901
28277 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
28280 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
28282 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
28283 Introduce unknown_misalignment parameter and remove vf.
28284 (vect_peeling_hash_get_lowest_cost):
28285 Pass unknown_misalignment parameter.
28286 (vect_enhance_data_refs_alignment):
28287 Fix unsupportable data ref treatment.
28289 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
28291 * tree-vect-data-refs.c (vect_get_data_access_cost):
28292 Workaround for SLP handling.
28293 (vect_enhance_data_refs_alignment):
28294 Compute costs for doing no peeling at all, compare to the best
28295 peeling costs so far and avoid peeling if cheaper.
28297 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
28299 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
28300 Return peeling info and set costs to zero for unlimited cost
28302 (vect_enhance_data_refs_alignment): Also inspect all datarefs
28303 with unknown misalignment. Compute and costs for unknown
28304 misalignment, compare them to the costs for known misalignment
28305 and choose the cheapest for peeling.
28307 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
28309 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
28310 (vect_get_peeling_costs_all_drs): Create function.
28311 (vect_peeling_hash_get_lowest_cost):
28312 Use vect_get_peeling_costs_all_drs.
28313 (vect_peeling_supportable): Create function.
28314 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
28316 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
28318 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
28319 DR_HAS_NEGATIVE_STEP.
28320 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
28321 (vect_enhance_data_refs_alignment): Use.
28322 (vect_duplicate_ssa_name_ptr_info): Use.
28323 * tree-vectorizer.h (dr_misalignment): Use.
28324 (known_alignment_for_access_p): Use.
28326 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
28329 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
28331 (has_section_name): New function.
28333 2017-05-30 Martin Liska <mliska@suse.cz>
28336 * auto-profile.c (get_function_decl_from_block): Fix
28339 2017-05-30 Richard Biener <rguenther@suse.de>
28341 PR middle-end/80876
28342 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
28344 2017-05-30 Martin Liska <mliska@suse.cz>
28346 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
28347 * dumpfile.h (struct dump_file_info): Remove ctors.
28349 2017-05-30 Martin Liska <mliska@suse.cz>
28351 * predict.def: Fix GNU coding style.
28353 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
28355 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
28356 Mark 'to' argument with ATTRIBUTE_UNUSED.
28358 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
28360 * config/xtensa/xtensa.c (xtensa_emit_call): Use
28361 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
28362 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
28365 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
28367 * doc/install.texi (Options specification): Restore entry of
28368 --enable-sjlj-exceptions.
28370 2017-05-27 Michael Eager <eager@eagercon.com>
28373 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
28375 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
28377 * config/microblaze/microblaze.h
28378 (FIXED_REGISTERS): Update in macro.
28379 (CALL_USED_REGISTERS): Update in macro.
28381 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
28383 * doc/install.texi: Add links to macOS binary distributions.
28385 2017-05-27 Jakub Jelinek <jakub@redhat.com>
28389 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
28391 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
28393 2017-05-26 Martin Liska <mliska@suse.cz>
28395 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
28397 2017-05-26 Martin Liska <mliska@suse.cz>
28399 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
28400 always leading ';; '.
28401 (dump_bb_info): Likewise.
28402 (brief_dump_cfg): Likewise.
28403 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
28404 * dumpfile.c: Remove usage of TDF_VERBOSE.
28405 * dumpfile.h (enum dump_kind): Likewise.
28406 (dump_gimple_bb_header): Do not use TDF_COMMENT.
28407 * print-tree.c (debug_verbose): Remove.
28408 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
28409 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
28410 * tree-diagnostic.c (default_tree_printer): Replace
28411 TDF_DIAGNOSTIC with TDF_SLIM.
28413 2017-05-26 Bin Cheng <bin.cheng@arm.com>
28415 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
28416 in parameter loop, rather than loop_vinfo.
28417 (create_intersect_range_checks): Ditto.
28418 (vect_create_cond_for_alias_checks): Update call to above functions.
28420 2017-05-26 Bin Cheng <bin.cheng@arm.com>
28422 PR tree-optimization/80815
28423 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
28424 for merging runtime alias checks. Handle negative DR_STEPs.
28426 2017-05-26 Bin Cheng <bin.cheng@arm.com>
28428 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
28430 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
28431 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
28432 out code pruning runtime alias checks.
28433 * tree-data-ref.c (prune_runtime_alias_test_list): New function
28434 factored out from above.
28435 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
28437 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
28439 (prune_runtime_alias_test_list): New decalaration.
28441 2017-05-26 Bin Cheng <bin.cheng@arm.com>
28443 * tree-vect-data-refs.c (compare_tree): Rename and move ...
28444 * tree-data-ref.c (data_ref_compare_tree): ... to here.
28445 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
28446 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
28447 (operator==, comp_dr_with_seg_len_pair): Ditto.
28448 (vect_prune_runtime_alias_test_list): Ditto.
28450 2017-05-26 Martin Liska <mliska@suse.cz>
28453 * params.def: Bound partial-inlining-entry-probability param.
28455 2017-05-26 Marek Polacek <polacek@redhat.com>
28458 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
28461 2017-05-26 Richard Biener <rguenther@suse.de>
28463 PR tree-optimization/80842
28464 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
28467 2017-05-26 Richard Biener <rguenther@suse.de>
28469 PR tree-optimization/80844
28470 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
28472 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
28474 * doc/md.texi (Machine Constraints): Update x86 family
28475 machine constraints section to match 'config/i386/constraints.md'.
28477 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
28479 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
28481 2017-05-25 Nathan Sidwell <nathan@acm.org>
28483 * doc/invoke.texi (--enable-languages): Update documentation.
28485 2017-05-25 Martin Liska <mliska@suse.cz>
28487 * dumpfile.c: Add TDF_FOLDING.
28488 * dumpfile.h (enum dump_kind): Likewise.
28489 * genmatch.c (dt_simplify::gen_1): Use it.
28491 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
28493 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
28495 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
28497 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
28498 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
28500 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
28502 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
28503 * match.pd (X == C): Rewrite it here.
28504 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
28505 with_certain_nonzero_bits2): New predicates.
28506 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
28508 2017-05-24 Nathan Sidwell <nathan@acm.org>
28510 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
28513 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
28516 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
28518 * config/powerpcspe: New port. Files are copied from the rs6000
28519 port, with "rs6000" in filenames replaced by "powerpcspe".
28521 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
28523 PR rtl-optimization/80754
28524 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
28526 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
28528 * config/sparc/sparc.md (length): Return the correct value for -mflat
28529 sibcalls to match output_sibcall.
28531 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
28535 * config/rs6000/rs6000.c (struct machine_function): Add new field
28537 (rs6000_get_separate_components): Init that field, use it.
28538 (rs6000_components_for_bb): Use the field.
28540 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
28542 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
28544 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
28546 PR middle-end/80823
28547 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
28549 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28552 * config/s390/s390.c (s390_check_qrst_address): Check incoming
28553 address against address_operand predicate.
28554 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
28556 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
28558 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
28559 parameters passed indirectly.
28561 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
28563 * config/i386/i386.md (*movdi_internal): Remove SSE4
28564 alternative 18 (?r, *v). Update insn attributes.
28565 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
28566 Update insn attributes.
28567 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
28568 Update insn attributes.
28569 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
28570 alternative 1 (r, v). Remove isa attribute.
28571 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
28572 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
28573 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
28575 2017-05-23 Tom de Vries <tom@codesourcery.com>
28577 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
28580 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
28582 * cgraphunit.c (symbol_table::process_new_functions): Update.
28583 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
28584 (inline_generate_summary): Rename to ...
28585 (ipa_fn_summary_generate): ... this one.
28586 (inline_read_summary): Rename to ...
28587 (ipa_fn_summary_read): ... this one.
28588 (inline_write_summary): Rename to ...
28589 (ipa_fn_summary_write): ... this one.
28590 (inline_free_summary): Rename to ...
28591 (ipa_free_fn_summary): ... this one.
28592 (pass_data_local_fn_summary, pass_local_fn_summary,
28593 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
28594 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
28595 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
28596 make_pass_ipa_fn_summary): New.
28597 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
28598 inline_write_summary, inline_free_summary): Remove.
28599 (ipa_free_fn_summary) : New.
28600 * ipa-inline.c (ipa_inline): Update.
28601 (pass_ipa_inline): Do not generate summaries.
28602 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
28604 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
28605 and add pass_ipa_fn_summary.
28606 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
28608 (make_pass_inline_parameters): Remove.
28610 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
28612 * omp-low.c (struct omp_context): Remove "default_kind" member.
28615 * omp-offload.c (execute_oacc_device_lower): Remove the
28616 parallelism dimensions function attributes for unparallelized
28617 OpenACC kernels constructs.
28619 2017-05-23 Martin Liska <mliska@suse.cz>
28621 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
28623 (cgraph_edge::make_speculative): Likewise.
28624 (cgraph_edge::resolve_speculation): Likewise.
28625 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
28626 (cgraph_node::dump): Likewise.
28627 * cgraph.h: Likewise.
28628 * cgraphunit.c (analyze_functions): Likewise.
28629 (symbol_table::compile): Likewise.
28630 * ipa-cp.c (print_all_lattices): Likewise.
28631 (determine_versionability): Likewise.
28632 (initialize_node_lattices): Likewise.
28633 (ipcp_verify_propagated_values): Likewise.
28634 (estimate_local_effects): Likewise.
28635 (update_profiling_info): Likewise.
28636 (create_specialized_node): Likewise.
28637 (perhaps_add_new_callers): Likewise.
28638 (decide_about_value): Likewise.
28639 (decide_whether_version_node): Likewise.
28640 (identify_dead_nodes): Likewise.
28641 (ipcp_store_bits_results): Likewise.
28642 * ipa-devirt.c (dump_targets): Likewise.
28643 (ipa_devirt): Likewise.
28644 * ipa-icf.c (sem_item::dump): Likewise.
28645 (sem_function::equals): Likewise.
28646 (sem_variable::equals): Likewise.
28647 (sem_item_optimizer::read_section): Likewise.
28648 (sem_item_optimizer::execute): Likewise.
28649 (congruence_class::dump): Likewise.
28650 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
28651 (dump_inline_summary): Likewise.
28652 (estimate_node_size_and_time): Likewise.
28653 (inline_analyze_function): Likewise.
28654 * ipa-inline-transform.c (inline_call): Likewise.
28655 * ipa-inline.c (report_inline_failed_reason): Likewise.
28656 (want_early_inline_function_p): Likewise.
28657 (edge_badness): Likewise.
28658 (update_edge_key): Likewise.
28659 (inline_small_functions): Likewise.
28660 * ipa-profile.c (ipa_profile): Likewise.
28661 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
28662 (ipa_make_edge_direct_to_target): Likewise.
28663 (remove_described_reference): Likewise.
28664 (ipa_impossible_devirt_target): Likewise.
28665 (propagate_controlled_uses): Likewise.
28666 (ipa_print_node_params): Likewise.
28667 (ipcp_transform_function): Likewise.
28668 * ipa-pure-const.c (pure_const_read_summary): Likewise.
28669 (propagate_pure_const): Likewise.
28670 * ipa-reference.c (generate_summary): Likewise.
28671 (read_write_all_from_decl): Likewise.
28672 (propagate): Likewise.
28673 (ipa_reference_read_optimization_summary): Likewise.
28674 * ipa-utils.c (ipa_merge_profiles): Likewise.
28675 * ipa.c (walk_polymorphic_call_targets): Likewise.
28676 (symbol_table::remove_unreachable_nodes): Likewise.
28677 (ipa_single_use): Likewise.
28678 * passes.c (execute_todo): Likewise.
28679 * predict.c (drop_profile): Likewise.
28680 * symtab.c (symtab_node::get_dump_name): New function.
28681 (symtab_node::dump_name): Likewise.
28682 (symtab_node::dump_asm_name): Likewise.
28683 (symtab_node::dump_references): Likewise.
28684 (symtab_node::dump_referring): Likewise.
28685 (symtab_node::dump_base): Likewise.
28686 (symtab_node::debug_symtab): Likewise.
28687 * tree-sra.c (convert_callers_for_node): Likewise.
28688 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
28689 * value-prof.c (init_node_map): Likewise.
28691 2017-05-23 Martin Liska <mliska@suse.cz>
28693 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
28694 and symtab_node::debug_symtab to symbol_table::debug.
28695 * cgraphunit.c (analyze_functions): Use the renamed function.
28696 (symbol_table::compile): Likewise.
28697 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
28698 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
28699 * passes.c (execute_todo): Likewise.
28700 * symtab.c (symbol_table::dump): New function.
28701 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
28703 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
28705 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
28706 that nonconst implies exec.
28708 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
28710 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
28711 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
28712 (inline_edge_summary_vec): Turn into ...
28713 (ipa_call_summaries): ... this one.
28714 (redirect_to_unreachable, edge_set_predicate,
28715 evaluate_properties_for_edge, inline_summary_alloc,
28716 reset_ipa_call_summary, reset_inline_summary,
28717 inline_summary_t::duplicate): Update.
28718 (inline_edge_duplication_hook): Turn to ...
28719 (ipa_call_summary_t::duplicate): ... this one.
28720 (inline_edge_removal_hook): Turn to ...
28721 (ipa_call_summary_t::remove): ... this one.
28722 (dump_inline_edge_summary): Turn to ...
28723 (dump_ipa_call_summary): ... this one.
28724 (estimate_function_body_sizes): Update.
28725 (inline_update_callee_summaries): Update.
28726 (remap_edge_change_prob): Update.
28727 (remap_edge_summaries): Update.
28728 (inline_merge_summary): Update.
28729 (do_estimate_edge_time): Update.
28730 (inline_generate_summary): Update.
28731 (inline_read_section): Update.
28732 (inline_read_summary): Update.
28733 (inline_free_summary): Update.
28734 * ipa-inline.c (can_inline_edge_p): Update.
28735 (compute_inlined_call_time): Update.
28736 (want_inline_small_function_p): Update.
28737 (edge_badness): Update.
28738 (early_inliner): Update.
28739 * ipa-inline.h (inline_edge_summary): Turn to ...
28740 (ipa_call_summary): ... this one.
28741 (ipa_call_summary_t): New class.
28742 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
28743 (ipa_call_summaries): New.
28744 (inline_edge_summary): Remove.
28745 (estimate_edge_growth): Update.
28746 * ipa-profile.c (ipa_propagate_frequency_1): Update.
28747 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
28748 * ipa-split.c (execute_split_functions): Update.
28749 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
28751 2017-05-23 Tom de Vries <tom@codesourcery.com>
28753 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
28754 attributes): Document rdrand effective target.
28756 2017-05-23 Tom de Vries <tom@codesourcery.com>
28758 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
28759 attributes): Sort alphabetically.
28761 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
28763 * config/avr/genmultilib.awk: Use gsub instead of gensub.
28765 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
28768 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
28769 V2DF/V2DI splat into two separate patterns, one that handles
28770 registers, and the other that only handles memory. Drop support
28771 for splatting from a GPR on ISA 2.07 and then splitting the
28772 splat into direct move and splat.
28773 (vsx_splat_<mode>_reg): Likewise.
28774 (vsx_splat_<mode>_mem): Likewise.
28776 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
28778 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
28780 2017-05-22 Jakub Jelinek <jakub@redhat.com>
28782 PR middle-end/80809
28783 * omp-low.c (finish_taskreg_remap): New function.
28784 (finish_taskreg_scan): If unit size of ctx->record_type
28785 is non-constant, unshare the size expression and replace
28786 decls in it with possible outer var refs.
28788 PR middle-end/80809
28789 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
28790 GOVD_SHARED rather than GOVD_PRIVATE with it.
28791 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
28792 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
28794 PR middle-end/80853
28795 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
28796 as last argument to build_outer_var_ref for pointer bases of array
28797 section reductions.
28799 2017-05-19 Martin Sebor <msebor@redhat.com>
28801 * print-tree.c (print_node): Print DECL_READ_P flag.
28803 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
28805 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
28806 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
28807 * cgraph.c: Likewise.
28808 * cgraphunit.c: Likewise.
28809 * gengtype.c: Likewise.
28810 * ipa-cp.c: Likewise.
28811 * ipa-devirt.c: Likewise.
28812 * ipa-icf.c: Likewise.
28813 * ipa-predicate.c: Likewise.
28814 * ipa-profile.c: Likewise.
28815 * ipa-prop.c: Likewise.
28816 * ipa-split.c: Likewise.
28818 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
28819 edge_predicate_pool, dump_inline_hints,
28820 inline_summary::account_size_time, redirect_to_unreachable,
28821 edge_set_predicate, set_hint_predicate,
28822 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
28823 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
28824 inline_summary_t::remove, remap_hint_predicate_after_duplication,
28825 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
28826 ipa_call_summary_t::remove, initialize_growth_caches,
28827 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
28828 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
28829 mark_modified, unmodified_parm_1, unmodified_parm,
28830 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
28831 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
28832 compute_bb_predicates, will_be_nonconstant_expr_predicate,
28833 will_be_nonconstant_predicate, record_modified_bb_info,
28834 get_minimal_bb, record_modified, param_change_prob,
28835 phi_result_unknown_predicate, predicate_for_phi_result,
28836 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
28837 estimate_function_body_sizes, compute_inline_parameters,
28838 compute_inline_parameters_for_curren, pass_data_inline_parameters,
28839 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
28840 inline_update_callee_summaries, remap_edge_change_prob,
28841 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
28842 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
28843 inline_analyze_function, inline_summary_t::insert,
28844 inline_generate_summary, read_ipa_call_summary, inline_read_section,
28845 inline_read_summary, write_ipa_call_summary, inline_write_summary,
28846 inline_free_summary): Move to ipa-fnsummary.h
28847 (predicate_t): Remove.
28848 * ipa-fnsummary.c: New file.
28849 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
28850 (enum inline_hints_vals, inline_hints, agg_position_info,
28851 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
28852 inline_summaries, ipa_call_summary, ipa_call_summary_t,
28853 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
28854 dump_inline_summary, dump_inline_hints, inline_generate_summary,
28855 inline_read_summary, inline_write_summary, inline_free_summary,
28856 inline_analyze_function, initialize_inline_failed,
28857 inline_merge_summary, inline_update_overall_summary,
28858 compute_inline_parameters): Move to ipa-fnsummary.h
28859 * ipa-fnsummary.h: New file.
28860 * ipa-inline-transform.h: Include ipa-inline.h.
28861 * ipa-inline.c: LIkewise.
28863 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
28865 * ipa-inline.c (edge_badness): Use inlined_time instead of
28866 inline_summaries->get.
28868 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
28870 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
28872 2017-05-22 Nathan Sidwell <nathan@acm.org>
28874 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
28875 (fdump-lang): Document 'raw' option.
28876 * dumpfile.h (TDI_tu): Delete.
28877 * dumpfile.c (dump_files): Remove translation-unit.
28878 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
28880 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
28882 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
28883 command option from $(AWK) call.
28884 * config/avr/genmultilib.awk: Simplify and rewrite so that it
28885 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
28886 [FORMAT]: Remove handling of variable.
28887 * config/avr/t-multilib: Regenerate.
28889 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
28891 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
28893 (dump_inline_summary): Do not print self_time.
28894 (estimate_function_body_sizes): Do not set self_time.
28895 (compute_inline_parameters): Likewise.
28896 (inline_read_section, inline_write_summary): Do not stream self_time.
28897 * ipa-inline.h (inline_summary): Drop self_time.
28899 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
28901 * ipa-inline-analysis.c (account_size_time): Rename to ...
28902 (inline_summary::account_size_time): ... this one.
28903 (reset_ipa_call_summary): Turn to ...
28904 (ipa_call_summary::reset): ... this one.
28905 (reset_inline_summary): Turn to ...
28906 (inline_summary::reset): ... this one.
28907 (inline_summary_t::remove): Update.
28908 (inline_summary_t::duplicate): Update.
28909 (ipa_call_summary_t::remove): Update.
28910 (dump_inline_summary): Update.
28911 (estimate_function_body_sizes): Update.
28912 (compute_inline_parameters): Update.
28913 (estimate_node_size_and_time): Update.
28914 (inline_merge_summary): Update.
28915 (inline_update_overall_summary): Update.
28916 (inline_read_section): Update.
28917 (inline_write_summary): Update.
28918 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
28919 add account_size_time and reset member functions.
28920 (ipa_call_summary): Add reset function.
28921 * ipa-predicate.h (predicate::operator &): Constify.
28923 2017-05-22 Richard Biener <rguenther@suse.de>
28925 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
28927 2017-05-19 Jason Merrill <jason@redhat.com>
28929 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
28931 2017-05-19 Marek Polacek <polacek@redhat.com>
28934 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
28935 TYPE_OVERFLOW_WRAPS checks.
28937 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
28939 * tree-core.h (enum omp_clause_default_kind): Add
28940 "OMP_CLAUSE_DEFAULT_PRESENT".
28941 * tree-pretty-print.c (dump_omp_clause): Handle it.
28942 * gimplify.c (enum gimplify_omp_var_data): Add
28943 "GOVD_MAP_FORCE_PRESENT".
28944 (gimplify_adjust_omp_clauses_1): Map it to
28945 "GOMP_MAP_FORCE_PRESENT".
28946 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
28948 * gimplify.c (oacc_default_clause): Clarify.
28950 2017-05-19 Nathan Sidwell <nathan@acm.org>
28952 LANG_HOOK_REGISTER_DUMPS
28953 * toplev.c (general_init): Call register dump lang hook.
28954 * doc/invoke.texi: Document -fdump-lang option family.
28955 * dumpfile.c (dump_files): Remove class dump here.
28956 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
28957 * dumpfile.h (tree_dump_index): Remove TDI_class.
28958 * langhooks-def.h (lhd_register_dumps): Declare.
28959 (LANG_HOOKS_REGISTER_DUMPS): Define.
28960 (LANG_HOOKS_INITIALIZER): Add it.
28961 * langhooks.c (lhd_register_dumps): Define.
28962 * langhooks.h (struct lang_hooks): Add register_dumps.
28964 2017-05-19 Nathan Sidwell <nathan@acm.org>
28966 * context.h (context::set_passes): New.
28967 * context.c (context::context): Do not create pass manager.
28968 * toplev.c (general_init): Create pass manager here.
28970 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
28972 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
28973 use this splitter if two add or or instructions would also work for
28974 the constant we want to generate.
28976 2017-05-19 Richard Biener <rguenther@suse.de>
28979 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
28980 predicate evaluation.
28982 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
28984 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
28986 * ipa-inline.c (want_inline_small_function_p): Do not cast to
28989 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
28991 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
28992 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
28993 (inline_edge_summary_vec): Turn into ...
28994 (ipa_call_summaries): ... this one.
28995 (redirect_to_unreachable, edge_set_predicate,
28996 evaluate_properties_for_edge, inline_summary_alloc,
28997 reset_ipa_call_summary, reset_inline_summary,
28998 inline_summary_t::duplicate): Update.
28999 (inline_edge_duplication_hook): Turn to ...
29000 (ipa_call_summary_t::duplicate): ... this one.
29001 (inline_edge_removal_hook): Turn to ...
29002 (ipa_call_summary_t::remove): ... this one.
29003 (dump_inline_edge_summary): Turn to ...
29004 (dump_ipa_call_summary): ... this one.
29005 (estimate_function_body_sizes): Update.
29006 (inline_update_callee_summaries): Update.
29007 (remap_edge_change_prob): Update.
29008 (remap_edge_summaries): Update.
29009 (inline_merge_summary): Update.
29010 (do_estimate_edge_time): Update.
29011 (inline_generate_summary): Update.
29012 (inline_read_section): Update.
29013 (inline_read_summary): Update.
29014 (inline_free_summary): Update.
29015 * ipa-inline.c (can_inline_edge_p): Update.
29016 (compute_inlined_call_time): Update.
29017 (want_inline_small_function_p): Update.
29018 (edge_badness): Update.
29019 (early_inliner): Update.
29020 * ipa-inline.h (inline_edge_summary): Turn to ...
29021 (ipa_call_summary): ... this one.
29022 (ipa_call_summary_t): New class.
29023 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
29024 (ipa_call_summaries): New.
29025 (inline_edge_summary): Remove.
29026 (estimate_edge_growth): Update.
29027 * ipa-profile.c (ipa_propagate_frequency_1): Update.
29028 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
29029 * ipa-split.c (execute_split_functions): Update.
29030 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
29032 2017-05-19 Richard Biener <rguenther@suse.de>
29034 PR middle-end/80764
29035 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
29037 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
29039 * config/rs6000/rs6000.c (struct machine_function): Add field
29040 fpr_is_wrapped_separately.
29041 (rs6000_get_separate_components): Use 64 components. Handle the
29042 new FPR components.
29043 (rs6000_components_for_bb): Handle the FPR components.
29044 (rs6000_emit_prologue_components): Handle the FPR components.
29045 (rs6000_emit_epilogue_components): Handle the FPR components.
29046 (rs6000_set_handled_components): Handle the FPR components.
29047 (rs6000_emit_prologue): Don't output prologue code for those FPRs
29048 that are already separately shrink-wrapped.
29049 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
29050 that are already separately shrink-wrapped.
29052 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
29055 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
29058 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
29059 (define_peephole2 for Altivec d-form load): Add peepholes to catch
29060 cases where the register allocator uses a move and an offsettable
29061 memory operation to/from a FPR register on ISA 2.06/2.07.
29062 (define_peephole2 for Altivec d-form store): Likewise.
29064 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
29067 * config/i386/mmx.md (*mov<mode>_internal): Enable
29068 alternatives 11, 12, 13 and 14 also for 32bit targets.
29069 Remove alternatives 15, 16, 17 and 18.
29070 * config/i386/sse.md (vec_concatv2di): Change
29071 alternative (!x, *y) to (x, ?!*Yn).
29073 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
29075 * dumpfile.h (enum dump_kind): Remove stray comma.
29077 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
29079 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
29080 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
29081 predicate::num_conditions
29082 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
29083 (CHANGED): turn into predicate::changed.
29084 (agg_position_info): Move to ipa-predicate.h
29085 (add_condition, predicate::add_clause, predicate::operator &=,
29086 predicate::or_with, predicate::evaluate, predicate::probability,
29087 dump_condition, dump_clause, predicate::dump,
29088 predicate::remap_after_duplication, predicate::remap_after_inlining,
29089 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
29090 (evaluate_conditions_for_known_args): Update.
29091 (set_cond_stmt_execution_predicate): Update.
29092 * ipa-inline.h: Include ipa-predicate.h
29093 (condition, inline_param_summary, conditions, agg_position_info,
29094 predicate): Move to ipa-predicate.h
29095 * ipa-predicate.c: New file.
29096 * ipa-predicate.h: New file.
29098 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
29100 * final.c (leaf_function_p): Check we are not in a sequence.
29102 2017-05-18 Martin Liska <mliska@suse.cz>
29104 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
29105 * dumpfile.c (dump_register): Use new enum dump_kind.
29106 (get_dump_file_name): Likewise.
29107 (dump_enable_all): Likewise.
29108 (dump_switch_p_1): Likewise.
29109 (enable_rtl_dump_file): Remove usage of TDF_RTL.
29110 * dumpfile.h (enum dump_kind): New enum type.
29111 (struct dump_file_info): Create constructor and
29112 format fields and comments.
29113 * passes.c (pass_manager::register_one_dump_file):
29115 * statistics.c (statistics_early_init): Likewise.
29116 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
29117 TDF_TREE with TDF_SLIM.
29118 (gather_memory_references_ref): Likewise.
29120 2017-05-18 Martin Liska <mliska@suse.cz>
29122 * vec.h (struct vnull): Use it.
29124 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
29126 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
29127 (true_predicate, false_predicate, true_predicate_p,
29128 false_predicate_p): Remove.
29129 (single_cond_predicate, not_inlined_predicate): Turn to member function
29131 (add_condition): Update.
29132 (add_clause): Turn to...
29133 (predicate::add_clause): ... this one; update; allow passing NULL
29135 (and_predicates): Turn to ...
29136 (predicate::operator &=): ... this one.
29137 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
29138 (or_predicates): Turn to ...
29139 (predicate::or_with): ... this one.
29140 (evaluate_predicate): Turn to ...
29141 (predicate::evaluate): ... this one.
29142 (predicate_probability): Turn to ...
29143 (predicate::probability): ... this one.
29144 (dump_condition): Update.
29145 (dump_predicate): Turn to ...
29146 (predicate::dump): ... this one.
29147 (account_size_time): Update.
29148 (edge_set_predicate): Update.
29149 (set_hint_predicate): UPdate.
29150 (evaluate_conditions_for_known_args): Update.
29151 (evaluate_properties_for_edge): Update.
29152 (remap_predicate_after_duplication): Turn to...
29153 (predicate::remap_after_duplication): ... this one.
29154 (remap_hint_predicate_after_duplication): Update.
29155 (inline_summary_t::duplicate): UPdate.
29156 (dump_inline_edge_summary): Update.
29157 (dump_inline_summary): Update.
29158 (set_cond_stmt_execution_predicate): Update.
29159 (set_switch_stmt_execution_predicate): Update.
29160 (compute_bb_predicates): Update.
29161 (will_be_nonconstant_expr_predicate): Update.
29162 (will_be_nonconstant_predicate): Update.
29163 (phi_result_unknown_predicate): Update.
29164 (predicate_for_phi_result): Update.
29165 (array_index_predicate): Update.
29166 (estimate_function_body_sizes): Update.
29167 (estimate_node_size_and_time): Update.
29168 (estimate_ipcp_clone_size_and_time): Update.
29169 (remap_predicate): Rename to ...
29170 (predicate::remap_after_inlining): ... this one.
29171 (remap_hint_predicate): Update.
29172 (inline_merge_summary): Update.
29173 (inline_update_overall_summary): Update.
29174 (estimate_size_after_inlining): Update.
29175 (read_predicate): Rename to ...
29176 (predicate::stream_in): ... this one.
29177 (read_inline_edge_summary): Update.
29178 (write_predicate): Rename to ...
29179 (predicate::stream_out): ... this one.
29180 (write_inline_edge_summary): Update.
29181 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
29182 (clause_t): Turn to uint32_t
29183 (predicate): Turn to class; implement constructor and operators
29185 (size_time_entry): Update.
29186 (inline_summary): Update.
29187 (inline_edge_summary): Update.
29189 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
29191 * fold-const.c (fold_binary_loc): Move transformation...
29192 * match.pd (C - X CMP X): ... here.
29194 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
29196 * config/sparc/sparc.c (sparc_option_override): Set function
29197 alignment for -mcpu=niagara7 to 64 to match the I$ line.
29198 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
29200 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
29202 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
29204 2017-05-18 Marek Polacek <polacek@redhat.com>
29207 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
29208 (pass_ubsan::execute): Call gimple_assign_single_p instead of
29209 gimple_assign_load_p.
29211 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
29213 PR middle-end/80692
29214 * real.c (do_compare): Give decimal_do_compare preference over
29215 comparing just the signs.
29217 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
29219 * doc/md.texi (Canonicalization of Instructions): Describe the
29220 canonical form of instructions that inherently set a condition
29223 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
29225 PR middle-end/80775
29226 * tree-cfg.c: Move deletion of unreachable case statements to after
29227 the merging of consecutive case labels.
29229 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
29231 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
29232 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
29233 restoring of callee-saved registers.
29235 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
29237 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
29238 * config/visium/visium.c (single_set_and_flags): Likewise.
29239 * config/visium/visium.md (Substitutions): Likewise.
29241 2017-05-17 Martin Liska <mliska@suse.cz>
29243 * cfg.c: Introduce dump_flags_t type and
29244 use it instead of int type.
29246 * cfghooks.c: Likewise.
29247 * cfghooks.h (struct cfg_hooks): Likewise.
29248 * cfgrtl.c: Likewise.
29249 * cfgrtl.h: Likewise.
29250 * cgraph.c (cgraph_node::get_body): Likewise.
29251 * coretypes.h: Likewise.
29252 * domwalk.c: Likewise.
29253 * domwalk.h: Likewise.
29254 * dumpfile.c (struct dump_option_value_info): Likewise.
29255 (dump_enable_all): Likewise.
29256 (dump_switch_p_1): Likewise.
29257 (opt_info_switch_p): Likewise.
29258 * dumpfile.h (enum tree_dump_index): Likewise.
29259 (struct dump_file_info): Likewise.
29260 * genemit.c: Likewise.
29261 * generic-match-head.c: Likewise.
29262 * gengtype.c (open_base_files): Likewise.
29263 * gimple-pretty-print.c: Likewise.
29264 * gimple-pretty-print.h: Likewise.
29265 * graph.c (print_graph_cfg): Likewise.
29266 * graphite-scop-detection.c (dot_all_sese): Likewise.
29267 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
29268 * loop-unroll.c (report_unroll): Likewise.
29269 * passes.c (pass_manager::register_one_dump_file): Likewise.
29270 * print-tree.c: Likewise.
29271 * statistics.c: Likewise.
29272 * tree-cfg.c: Likewise.
29273 * tree-cfg.h: Likewise.
29274 * tree-dfa.c: Likewise.
29275 * tree-dfa.h: Likewise.
29276 * tree-dump.c (dump_function): Likewise.
29277 * tree-dump.h (struct dump_info): Likewise.
29278 * tree-pretty-print.c: Likewise.
29279 * tree-pretty-print.h: Likewise.
29280 * tree-ssa-live.c: Likewise.
29281 * tree-ssa-live.h: Likewise.
29282 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
29283 * tree-vect-loop.c: Likewise.
29284 * tree-vect-slp.c: Likewise.
29286 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
29287 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29289 PR tree-optimization/80457
29290 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
29291 of all arguments to a statement as scalar_to_vec operations.
29292 (vectorizable_call): Adjust call to vect_model_simple_cost for
29294 (vectorizable_conversion): Likewise.
29295 (vectorizable_assignment): Likewise.
29296 (vectorizable_shift): Likewise.
29297 (vectorizable_operation): Likewise.
29298 (vectorizable_comparison): Likewise.
29299 (vect_is_simple_cond): Record the def types for operands.
29300 (vectorizable_condition): Likewise, call vect_model_simple_cost.
29301 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
29302 for statement argument count.
29304 2017-05-16 Carl Love <cel@us.ibm.com>
29306 * config/rs6000/rs6000-c: Add support for built-in functions
29307 vector unsigned long long vec_bperm (vector unsigned long long,
29308 vector unsigned char)
29309 vector signed long long vec_mule (vector signed int,
29311 vector unsigned long long vec_mule (vector unsigned int,
29312 vector unsigned int)
29313 vector signed long long vec_mulo (vector signed int,
29315 vector unsigned long long vec_mulo (vector unsigned int,
29316 vector unsigned int)
29317 vector signed char vec_sldw (vector signed char,
29318 vector signed char,
29320 vector unsigned char vec_sldw (vector unsigned char,
29321 vector unsigned char,
29323 vector signed short vec_sldw (vector signed short,
29324 vector signed short,
29326 vector unsigned short vec_sldw (vector unsigned short,
29327 vector unsigned short,
29329 vector signed int vec_sldw (vector signed int,
29332 vector unsigned int vec_sldw (vector unsigned int,
29333 vector unsigned int,
29335 vector signed long long vec_sldw (vector signed long long,
29336 vector signed long long,
29338 vector unsigned long long vec_sldw (vector unsigned long long,
29339 vector unsigned long long,
29341 * config/rs6000/rs6000-c: Add support for built-in functions
29342 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
29343 * config/rs6000/altivec.h: Add defintion for vec_sldw.
29344 * doc/extend.texi: Update the built-in documentation for the
29345 new built-in functions.
29347 2017-05-16 Marek Polacek <polacek@redhat.com>
29351 * tree.c (save_expr): Don't fold the expression.
29353 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
29355 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
29356 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
29357 and (?*y,m). Update insn attributes.
29359 2017-05-16 Martin Liska <mliska@suse.cz>
29361 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
29362 flags argument of print_gimple_stmt, print_gimple_expr,
29363 print_generic_stmt and print_generic_expr.
29364 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
29365 * coretypes.h: Likewise.
29366 * except.c (dump_eh_tree): Likewise.
29367 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
29368 * gimple-pretty-print.h: Likewise.
29369 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
29370 (backprop::push_to_worklist): Likewise.
29371 (backprop::pop_from_worklist): Likewise.
29372 (backprop::process_use): Likewise.
29373 (backprop::intersect_uses): Likewise.
29374 (note_replacement): Likewise.
29375 * gimple-ssa-store-merging.c
29376 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
29377 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
29378 (pass_store_merging::execute): Likewise.
29379 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
29380 (ssa_base_cand_dump_callback): Likewise.
29381 (dump_incr_vec): Likewise.
29382 (replace_refs): Likewise.
29383 (replace_mult_candidate): Likewise.
29384 (create_add_on_incoming_edge): Likewise.
29385 (create_phi_basis): Likewise.
29386 (insert_initializers): Likewise.
29387 (all_phi_incrs_profitable): Likewise.
29388 (introduce_cast_before_cand): Likewise.
29389 (replace_one_candidate): Likewise.
29390 * gimplify.c (gimplify_expr): Likewise.
29391 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
29392 (set_rename): Likewise.
29393 (rename_uses): Likewise.
29394 (copy_loop_phi_nodes): Likewise.
29395 (add_close_phis_to_merge_points): Likewise.
29396 (copy_loop_close_phi_args): Likewise.
29397 (copy_cond_phi_args): Likewise.
29398 (graphite_copy_stmts_from_block): Likewise.
29399 (translate_pending_phi_nodes): Likewise.
29400 * graphite-poly.c (print_pdr): Likewise.
29401 (dump_gbb_cases): Likewise.
29402 (dump_gbb_conditions): Likewise.
29403 (print_scop_params): Likewise.
29404 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
29405 (build_cross_bb_scalars_use): Likewise.
29406 (gather_bbs::before_dom_children): Likewise.
29407 * hsa-dump.c (dump_hsa_immed): Likewise.
29408 * ipa-cp.c (print_ipcp_constant_value): Likewise.
29409 (get_replacement_map): Likewise.
29410 * ipa-inline-analysis.c (dump_condition): Likewise.
29411 (estimate_function_body_sizes): Likewise.
29412 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
29413 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
29414 * ipa-prop.c (ipa_dump_param): Likewise.
29415 (ipa_print_node_jump_functions_for_edge): Likewise.
29416 (ipa_modify_call_arguments): Likewise.
29417 (ipa_modify_expr): Likewise.
29418 (ipa_dump_param_adjustments): Likewise.
29419 (ipa_dump_agg_replacement_values): Likewise.
29420 (ipcp_modif_dom_walker::before_dom_children): Likewise.
29421 * ipa-pure-const.c (check_stmt): Likewise.
29422 (pass_nothrow::execute): Likewise.
29423 * ipa-split.c (execute_split_functions): Likewise.
29424 * omp-offload.c (dump_oacc_loop_part): Likewise.
29425 (dump_oacc_loop): Likewise.
29426 * trans-mem.c (tm_log_emit): Likewise.
29427 (tm_memopt_accumulate_memops): Likewise.
29428 (dump_tm_memopt_set): Likewise.
29429 (dump_tm_memopt_transform): Likewise.
29430 * tree-cfg.c (gimple_verify_flow_info): Likewise.
29431 (print_loop): Likewise.
29432 * tree-chkp-opt.c (chkp_print_addr): Likewise.
29433 (chkp_gather_checks_info): Likewise.
29434 (chkp_get_check_result): Likewise.
29435 (chkp_remove_check_if_pass): Likewise.
29436 (chkp_use_outer_bounds_if_possible): Likewise.
29437 (chkp_reduce_bounds_lifetime): Likewise.
29438 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
29439 (chkp_mark_completed_bounds): Likewise.
29440 (chkp_register_incomplete_bounds): Likewise.
29441 (chkp_mark_invalid_bounds): Likewise.
29442 (chkp_maybe_copy_and_register_bounds): Likewise.
29443 (chkp_build_returned_bound): Likewise.
29444 (chkp_get_bound_for_parm): Likewise.
29445 (chkp_build_bndldx): Likewise.
29446 (chkp_get_bounds_by_definition): Likewise.
29447 (chkp_generate_extern_var_bounds): Likewise.
29448 (chkp_get_bounds_for_decl_addr): Likewise.
29449 * tree-chrec.c (chrec_apply): Likewise.
29450 * tree-data-ref.c (dump_data_reference): Likewise.
29451 (dump_subscript): Likewise.
29452 (dump_data_dependence_relation): Likewise.
29453 (analyze_overlapping_iterations): Likewise.
29454 * tree-inline.c (expand_call_inline): Likewise.
29455 (tree_function_versioning): Likewise.
29456 * tree-into-ssa.c (dump_defs_stack): Likewise.
29457 (dump_currdefs): Likewise.
29458 (dump_names_replaced_by): Likewise.
29459 (dump_update_ssa): Likewise.
29460 (update_ssa): Likewise.
29461 * tree-object-size.c (pass_object_sizes::execute): Likewise.
29462 * tree-parloops.c (build_new_reduction): Likewise.
29463 (try_create_reduction_list): Likewise.
29464 (ref_conflicts_with_region): Likewise.
29465 (oacc_entry_exit_ok_1): Likewise.
29466 (oacc_entry_exit_single_gang): Likewise.
29467 * tree-pretty-print.h: Likewise.
29468 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
29469 (get_scalar_evolution): Likewise.
29470 (add_to_evolution): Likewise.
29471 (get_loop_exit_condition): Likewise.
29472 (analyze_evolution_in_loop): Likewise.
29473 (analyze_initial_condition): Likewise.
29474 (analyze_scalar_evolution): Likewise.
29475 (instantiate_scev): Likewise.
29476 (number_of_latch_executions): Likewise.
29477 (gather_chrec_stats): Likewise.
29478 (final_value_replacement_loop): Likewise.
29479 (scev_const_prop): Likewise.
29480 * tree-sra.c (dump_access): Likewise.
29481 (disqualify_candidate): Likewise.
29482 (create_access): Likewise.
29483 (reject): Likewise.
29484 (maybe_add_sra_candidate): Likewise.
29485 (create_access_replacement): Likewise.
29486 (analyze_access_subtree): Likewise.
29487 (analyze_all_variable_accesses): Likewise.
29488 (sra_modify_assign): Likewise.
29489 (initialize_constant_pool_replacements): Likewise.
29490 (find_param_candidates): Likewise.
29491 (decide_one_param_reduction): Likewise.
29492 (replace_removed_params_ssa_names): Likewise.
29493 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
29494 * tree-ssa-copy.c (dump_copy_of): Likewise.
29495 (copy_prop_visit_cond_stmt): Likewise.
29496 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
29497 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
29498 (record_equivalences_from_stmt): Likewise.
29499 * tree-ssa-dse.c (compute_trims): Likewise.
29500 (delete_dead_call): Likewise.
29501 (delete_dead_assignment): Likewise.
29502 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
29503 (forward_propagate_into_cond): Likewise.
29504 (pass_forwprop::execute): Likewise.
29505 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
29506 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
29508 (move_computations_worker): Likewise.
29509 (execute_sm): Likewise.
29510 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
29511 (remove_exits_and_undefined_stmts): Likewise.
29512 (remove_redundant_iv_tests): Likewise.
29513 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
29514 (adjust_iv_update_pos): Likewise.
29515 * tree-ssa-math-opts.c (bswap_replace): Likewise.
29516 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
29517 (value_replacement): Likewise.
29518 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
29519 * tree-ssa-pre.c (print_pre_expr): Likewise.
29520 (get_representative_for): Likewise.
29521 (create_expression_by_pieces): Likewise.
29522 (insert_into_preds_of_block): Likewise.
29523 (eliminate_insert): Likewise.
29524 (eliminate_dom_walker::before_dom_children): Likewise.
29525 (eliminate): Likewise.
29526 (remove_dead_inserted_code): Likewise.
29527 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
29528 * tree-ssa-reassoc.c (get_rank): Likewise.
29529 (eliminate_duplicate_pair): Likewise.
29530 (eliminate_plus_minus_pair): Likewise.
29531 (eliminate_not_pairs): Likewise.
29532 (undistribute_ops_list): Likewise.
29533 (eliminate_redundant_comparison): Likewise.
29534 (update_range_test): Likewise.
29535 (optimize_range_tests_var_bound): Likewise.
29536 (optimize_vec_cond_expr): Likewise.
29537 (rewrite_expr_tree): Likewise.
29538 (rewrite_expr_tree_parallel): Likewise.
29539 (linearize_expr): Likewise.
29540 (break_up_subtract): Likewise.
29541 (linearize_expr_tree): Likewise.
29542 (attempt_builtin_powi): Likewise.
29543 (attempt_builtin_copysign): Likewise.
29544 (transform_stmt_to_copy): Likewise.
29545 (transform_stmt_to_multiply): Likewise.
29546 (dump_ops_vector): Likewise.
29547 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
29548 (print_scc): Likewise.
29549 (set_ssa_val_to): Likewise.
29550 (visit_reference_op_store): Likewise.
29551 (visit_use): Likewise.
29552 (sccvn_dom_walker::before_dom_children): Likewise.
29553 (run_scc_vn): Likewise.
29554 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
29556 (expr_hash_elt::print): Likewise.
29557 (const_and_copies::pop_to_marker): Likewise.
29558 (const_and_copies::record_const_or_copy_raw): Likewise.
29559 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
29560 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
29561 (dump_predicates): Likewise.
29562 (find_uninit_use): Likewise.
29563 (warn_uninitialized_phi): Likewise.
29564 (pass_late_warn_uninitialized::execute): Likewise.
29565 * tree-ssa.c (verify_vssa): Likewise.
29566 (verify_ssa): Likewise.
29567 (maybe_optimize_var): Likewise.
29568 * tree-vrp.c (dump_value_range): Likewise.
29569 (dump_all_value_ranges): Likewise.
29570 (dump_asserts_for): Likewise.
29571 (register_edge_assert_for_2): Likewise.
29572 (vrp_visit_cond_stmt): Likewise.
29573 (vrp_visit_switch_stmt): Likewise.
29574 (vrp_visit_stmt): Likewise.
29575 (vrp_visit_phi_node): Likewise.
29576 (simplify_cond_using_ranges_1): Likewise.
29577 (fold_predicate_in): Likewise.
29578 (evrp_dom_walker::before_dom_children): Likewise.
29579 (evrp_dom_walker::push_value_range): Likewise.
29580 (evrp_dom_walker::pop_value_range): Likewise.
29581 (execute_early_vrp): Likewise.
29583 2017-05-16 Richard Biener <rguenther@suse.de>
29585 * dwarf2out.c (loc_list_from_tree_1): Do not create
29586 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
29588 2017-05-16 Richard Biener <rguenther@suse.de>
29590 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
29592 (note_variable_value_in_expr): If we resolved the decl ref
29593 do not push to the stack.
29595 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
29597 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
29598 operations in fast-math mode.
29599 (vaddq_f16): Likewise.
29600 (vmul_f16): Likewise.
29601 (vmulq_f16): Likewise.
29602 (vsub_f16): Likewise.
29603 (vsubq_f16): Likewise.
29604 * config/arm/neon.md (add<mode>3): New.
29606 (fma:<VH:mode>3): New. Also remove outdated comment.
29609 2017-05-16 Martin Liska <mliska@suse.cz>
29613 * ipa-devirt.c (warn_types_mismatch): Fix typo.
29614 (odr_types_equivalent_p): Likewise.
29616 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
29618 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
29620 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
29623 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
29624 non-interunit SSE move alternatives with '?'.
29625 (zero-extendsidi peephole2): New peephole to skip intermediate
29626 general register in SSE zero-extend sequence.
29628 2017-05-15 Jeff Law <law@redhat.com>
29630 * reorg.c (relax_delay_slots): Create a new variable to hold
29631 the temporary target rather than clobbering TARGET_LABEL.
29633 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
29634 missing argument to extract_bit_field call.
29635 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
29637 2017-05-15 Martin Liska <mliska@suse.cz>
29640 * gcc.c (process_command): Do not allow empty argument of -o option.
29642 2017-05-15 Renlin Li <renlin.li@arm.com>
29644 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
29645 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
29646 * config/aarch64/constraints.md (Usf): Add long call check.
29647 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
29648 (call_value): Likewise.
29649 (sibcall): Likewise.
29650 (sibcall_value): Likewise.
29652 (call_value_insn): New.
29653 (sibcall_insn): Update rtx pattern.
29654 (sibcall_value_insn): Likewise.
29655 (call_internal): Remove.
29656 (call_value_internal): Likewise.
29657 (sibcall_internal): Likewise.
29658 (sibcall_value_internal): Likewise.
29659 (call_reg): Likewise.
29660 (call_symbol): Likewise.
29661 (call_value_reg): Likewise.
29662 (call_value_symbol): Likewise.
29664 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
29667 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
29669 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
29671 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
29672 compatible with CCGOCmode and with CCZmode.
29674 2017-05-14 Martin Sebor <msebor@redhat.com>
29676 PR middle-end/77671
29677 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
29678 (gimple_fold_builtin_snprintf): Same.
29679 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
29680 (gimple_fold_builtin_snprintf): Same.
29681 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
29682 of character types.
29683 (is_call_safe): New function.
29684 (try_substitute_return_value): Call it.
29685 (try_simplify_call): New function.
29686 (pass_sprintf_length::handle_gimple_call): Call it.
29688 2017-05-14 Martin Sebor <msebor@redhat.com>
29690 PR middle-end/80669
29691 * builtins.c (expand_builtin_stpncpy): Simplify.
29693 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
29695 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
29696 * config/i386/i386.h
29697 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
29698 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
29699 (struct machine_function): Add new members call_ms2sysv,
29700 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
29701 (struct machine_frame_state): New fields sp_realigned and
29702 sp_realigned_offset.
29703 * config/i386/i386.c
29704 (enum xlogue_stub): New enum.
29705 (enum xlogue_stub_sets): New enum.
29706 (class xlogue_layout): New class.
29707 (struct ix86_frame): New fields stack_realign_allocate_offset,
29708 stack_realign_offset and outlined_save_offset. Modify comments to
29709 detail stack layout when using out-of-line stubs.
29710 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
29711 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
29712 -mcall-ms2sysv-xlogues.
29713 (stub_managed_regs): New static variable.
29714 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
29715 registers managed by out-of-line stub.
29716 (disable_call_ms2sysv_xlogues): New function.
29717 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
29718 m->call_ms2sysv when appropriate and compute frame layout for
29720 (sp_valid_at, fp_valid_at): New inline functions.
29721 (choose_basereg): New function.
29722 (choose_baseaddr): Add align parameter, use choose_basereg and modify
29724 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
29725 Use align parameter of choose_baseaddr to generated aligned SSE movs
29727 (pro_epilogue_adjust_stack): Modify to track
29728 machine_frame_state::sp_realigned.
29729 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
29730 (ix86_nsaved_sseregs): Likewise.
29731 (ix86_emit_save_regs): Likewise.
29732 (ix86_emit_save_regs_using_mov): Likewise.
29733 (ix86_emit_save_sse_regs_using_mov): Likewise.
29734 (get_scratch_register_on_entry): Likewise.
29735 (gen_frame_set): New function.
29736 (gen_frame_load): Likewise.
29737 (gen_frame_store): Likewise.
29738 (emit_outlined_ms2sysv_save): Likewise.
29739 (emit_outlined_ms2sysv_restore): Likewise.
29740 (ix86_expand_prologue): Modify stack re-alignment code and call
29741 emit_outlined_ms2sysv_save when appropriate.
29742 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
29743 parameter rtx_insn *insn, which allows the function to be used to only
29744 generate the notes.
29745 (ix86_expand_epilogue): Modify validity checks of frame and stack
29746 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
29747 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
29748 * config/i386/predicates.md
29749 (save_multiple): New predicate.
29750 (restore_multiple): Likewise.
29751 * config/i386/sse.md
29752 (save_multiple<mode>): New pattern.
29753 (save_multiple_realign<mode>): Likewise.
29754 (restore_multiple<mode>): Likewise.
29755 (restore_multiple_and_return<mode>): Likewise.
29756 (restore_multiple_leave_return<mode>): Likewise.
29757 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
29759 2017-05-14 Julia Koval <julia.koval@intel.com>
29761 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
29762 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
29763 (__builtin_ia32_xsetbv): New builtins.
29764 * config/i386/i386.c (ix86_expand_special_args_builtin):
29766 (ix86_expand_builtin): Special expand for new intrinsics.
29767 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
29768 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
29769 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
29771 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29773 * cfganal.c (inverted_post_order_compute): Change argument type
29775 * cfganal.h (inverted_post_order_compute): Adjust prototype.
29776 * df-core.c (rest_of_handle_df_initialize): Adjust.
29777 (rest_of_handle_df_finish): Likewise.
29778 (df_analyze_1): Likewise.
29779 (df_analyze): Likewise.
29780 (loop_inverted_post_order_compute): Change argument to be a vec *.
29781 (df_analyze_loop): Adjust.
29782 (df_get_n_blocks): Likewise.
29783 (df_get_postorder): Likewise.
29784 * df.h (struct df_d): Change field to be a vec.
29785 * lcm.c (compute_laterin): Adjust.
29786 (compute_available): Likewise.
29787 * lra-lives.c (lra_create_live_ranges_1): Likewise.
29788 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
29789 * tree-ssa-pre.c (compute_antic): Likewise.
29791 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29793 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
29794 (depth_first_search::depth_first_search): Change structure init
29795 function to this constructor.
29796 (depth_first_search::add_bb): Rename function to this member.
29797 (depth_first_search::execute): Likewise.
29798 (flow_dfs_compute_reverse_finish): Adjust.
29800 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29802 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
29803 (longest_simple_path): Likewise.
29804 * shrink-wrap.c (spread_components): Likewise.
29805 (disqualify_problematic_components): Likewise.
29806 (emit_common_heads_for_components): Likewise.
29807 (emit_common_tails_for_components): Likewise.
29808 (insert_prologue_epilogue_for_components): Likewise.
29810 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29812 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
29815 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29817 * df-core.c (df_set_blocks): Start using auto_bitmap.
29818 (df_compact_blocks): Likewise.
29819 * df-problems.c (df_rd_confluence_n): Likewise.
29820 * df-scan.c (df_insn_rescan_all): Likewise.
29821 (df_process_deferred_rescans): Likewise.
29822 (df_update_entry_block_defs): Likewise.
29823 (df_update_exit_block_uses): Likewise.
29824 (df_entry_block_bitmap_verify): Likewise.
29825 (df_exit_block_bitmap_verify): Likewise.
29826 (df_scan_verify): Likewise.
29827 * lra-constraints.c (lra_constraints): Likewise.
29828 (undo_optional_reloads): Likewise.
29829 (lra_undo_inheritance): Likewise.
29830 * lra-remat.c (calculate_gen_cands): Likewise.
29831 (do_remat): Likewise.
29832 * lra-spills.c (assign_spill_hard_regs): Likewise.
29833 (spill_pseudos): Likewise.
29834 * tree-ssa-pre.c (bitmap_set_and): Likewise.
29835 (bitmap_set_subtract_values): Likewise.
29837 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29839 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
29840 management with auto_bitmap.
29841 (fix_inter_tick): Likewise.
29842 (fix_recovery_deps): Likewise.
29843 * ira.c (add_store_equivs): Likewise.
29844 (find_moveable_pseudos): Likewise.
29845 (split_live_ranges_for_shrink_wrap): Likewise.
29846 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
29847 (rtx_reuse_manager::seen_def_p): Likewise.
29848 (rtx_reuse_manager::set_seen_def): Likewise.
29849 * print-rtl.h (class rtx_reuse_manager): Likewise.
29851 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29853 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
29855 (migrate_btr_def): Likewise.
29856 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
29857 * df-core.c (loop_post_order_compute): Likewise.
29858 (loop_inverted_post_order_compute): Likewise.
29859 * hsa-common.h: Likewise.
29860 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
29861 * init-regs.c (initialize_uninitialized_regs): Likewise.
29862 * ipa-inline.c (resolve_noninline_speculation): Likewise.
29863 (inline_small_functions): Likewise.
29864 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
29865 * ira.c (combine_and_move_insns): Likewise.
29866 (build_insn_chain): Likewise.
29867 * loop-invariant.c (find_invariants): Likewise.
29868 * lower-subreg.c (propagate_pseudo_copies): Likewise.
29869 * predict.c (tree_predict_by_opcode): Likewise.
29870 (predict_paths_leading_to): Likewise.
29871 (predict_paths_leading_to_edge): Likewise.
29872 (estimate_loops_at_level): Likewise.
29873 (estimate_loops): Likewise.
29874 * shrink-wrap.c (try_shrink_wrapping): Likewise.
29875 (spread_components): Likewise.
29876 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
29877 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
29878 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
29879 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
29880 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
29881 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
29882 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
29883 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
29884 (mark_threaded_blocks): Likewise.
29885 (thread_through_all_blocks): Likewise.
29886 * tree-ssa.c (verify_ssa): Likewise.
29887 (execute_update_addresses_taken): Likewise.
29888 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
29890 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29892 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
29894 (post_order_compute): Likewise.
29895 (inverted_post_order_compute): Likewise.
29896 (pre_and_rev_post_order_compute_fn): Likewise.
29898 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29900 * genrecog.c (int_set::int_set): Explicitly construct our
29901 auto_vec base class.
29902 * vec.h (auto_vec::auto_vec): New constructor.
29904 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29906 * bitmap.h (class auto_bitmap): New constructor taking
29907 bitmap_obstack * argument.
29909 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29911 * bitmap.h (class auto_bitmap): Change type of m_bits to
29912 bitmap_head, and adjust ctor / dtor and member operators.
29914 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
29916 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
29917 when returned register mode doesn't match original mode.
29919 2017-05-12 Jeff Law <law@redhat.com>
29920 Jakub Jelinek <jakub@redhat.com>
29922 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
29923 we look for cc setter after the compare-elim changes.
29924 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
29925 within the vector to match what compare-elim now expects.
29926 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
29927 (xorsi3_flags, one_cmplsi2_flags): Likewise.
29929 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
29930 after the compare-elim changes.
29931 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
29932 the vector to match what compare-elim now expects.
29933 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
29934 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
29935 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
29936 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
29937 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
29939 * config/visium/visium.c (single_set_and_flags): Fix where
29940 we look for cc setter after the compare-elim changes.
29941 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
29942 with the vector to match what compare-elim now expects.
29943 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
29944 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
29945 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
29946 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
29947 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
29948 (neg<mode>2_insn_set_overflow): Likewise.
29950 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
29952 PR middle-end/79794
29953 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
29954 maybe_expand_insn call, set ops[0].target. If still set after call,
29955 set alt_rtl. Add extra arg to recursive calls.
29956 (extract_bit_field): Add alt_rtl argument. Pass to
29958 * expmed.h (extract_bit_field): Fix prototype.
29959 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
29960 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
29961 to extract_bit_field_calls.
29962 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
29963 Pass alt_rtl to extract_bit_field calls.
29964 * calls.c (store_unaligned_arguments_into_psuedos)
29965 load_register_parameters): Pass extra NULL to extract_bit_field calls.
29966 * optabs.c (maybe_legitimize_operand): Clear op->target when call
29968 * optabs.h (struct expand_operand): Add target bitfield.
29970 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
29972 * compare-elim.c (try_eliminate_compare): Canonicalize
29973 operation with embedded compare to
29974 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
29975 (set (reg) (operation)].
29977 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
29979 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
29982 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
29983 cost of adding a carry flag for ADC instruction.
29984 [case MINUS]: Ignore the cost of subtracting a carry flag
29985 for SBB instruction.
29987 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
29989 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
29991 * config/rs6000/bmiintrin.h: New file.
29992 * config/rs6000/bmi2intrin.h: New file.
29993 * config/rs6000/x86intrin.h: New file.
29995 2017-05-12 Jeff Law <law@redhat.com>
29997 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
30000 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
30002 PR middle-end/80707
30003 * tree-cfg.c: Remove cfg edges of unreachable case statements.
30005 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30007 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
30008 early expansion of vector divide builtins.
30009 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
30010 builtins identified as having unsigned arguments.
30012 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30014 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
30015 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
30016 expansion of vector logical operations (and, andc, or, xor,
30019 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30021 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
30022 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
30024 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30026 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
30027 early GIMPLE expansion of vector multiplies.
30029 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
30031 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
30032 TARGET_HAVE_MOVT conditional.
30033 (movt splitter): Likewise.
30035 2017-05-12 Richard Biener <rguenther@suse.de>
30037 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
30038 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
30039 Fold all stmts not inplace.
30041 2017-05-12 Richard Biener <rguenther@suse.de>
30043 PR tree-optimization/80713
30044 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
30045 inserted_exprs bit for not removed stmts.
30047 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
30049 PR middle-end/69921
30050 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
30051 parallelized" attribute for parallelized OpenACC kernels.
30052 * omp-offload.c (execute_oacc_device_lower): Use it.
30054 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
30055 Set "oacc kernels" attribute.
30056 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
30057 parameter. Adjust all users.
30058 (oacc_fn_attrib_kernels_p): Remove function.
30059 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
30060 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
30061 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
30062 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
30063 assert "oacc kernels" attribute is set.
30065 2017-05-11 Carl Love <cel@us.ibm.com>
30067 * config/rs6000/rs6000-c: Add support for built-in functions
30068 vector unsigned char vec_popcnt (vector signed char)
30069 vector unsigned char vec_popcnt (vector unsigned char)
30070 vector unsigned short vec_popcnt (vector signed short)
30071 vector unsigned short vec_popcnt (vector unsigned short)
30072 vector unsigned int vec_popcnt (vector signed int)
30073 vector unsigned int vec_popcnt (vector unsigned int)
30074 vector unsigned long long vec_popcnt (vector signed long long)
30075 vector unsigned long long vec_popcnt (vector unsigned long long)
30076 vector signed long long vec_slo (vector signed long long,
30077 vector signed char)
30078 vector signed long long vec_slo (vector signed long long,
30079 vector unsigned char)
30080 vector unsigned long long vec_slo (vector unsigned long long,
30081 vector signed char)
30082 vector unsigned long long vec_slo (vector unsigned long long,
30083 vector unsigned char)
30084 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
30085 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
30086 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
30087 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
30088 * doc/extend.texi: Update the built-in documentation file for the
30089 new built-in functions.
30091 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
30093 * attribs.h (sorted_attr_string): Move machine independent
30094 functions for target clone support from the i386 port to common
30095 code. Rename ix86_function_versions to common_function_versions.
30096 Rename make_name to make_unique_name.
30097 (common_function_versions): Likewise.
30098 (make_unique_name): Likewise.
30099 (make_dispatcher_decl): Likewise.
30100 (is_function_default_version): Likewise.
30101 * attribs.c (attr_strcmp): Likewise.
30102 (sorted_attr_string): Likewise.
30103 (common_function_versions): Likewise.
30104 (make_unique_name): Likewise.
30105 (make_dispatcher_decl): Likewise.
30106 (is_function_default_version): Likewise.
30107 * config/i386/i386.c (attr_strcmp): Likewise.
30108 (sorted_attr_string): Likewise.
30109 (ix86_function_versions): Likewise.
30110 (make_name): Likewise.
30111 (make_dispatcher_decl): Likewise.
30112 (is_function_default_version): Likewise.
30113 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
30115 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30118 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
30119 Account for direct move costs for vec_construct of integer
30122 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
30125 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
30126 (UNSPEC_STX_ATOMIC): Ditto.
30127 (loaddi_via_sse): New insn.
30128 (storedi_via_sse): Ditto.
30129 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
30130 Update corresponding peephole2 patterns.
30131 (atomic_storedi_fpu): Ditto.
30133 2017-05-11 Julia Koval <julia.koval@intel.com>
30135 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
30136 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
30138 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
30139 (__builtin_ia32_rsqrt14ss_mask): New builtins.
30140 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
30142 2017-05-11 Nathan Sidwell <nathan@acm.org>
30144 * graphite-poly.c: Include dumpfile.h.
30146 * dumpfle.h (dump_function): Declare here ...
30147 * tree-dump.h (dump_function): ... not here.
30148 * dumpfile.c: #include tree-cfg.h.
30149 (dump_function): Move here from ...
30150 * tree-dump.c (dump_function): ... here.
30151 * gimplify.c: #include splay-tree.h, not tree-dump.h.
30152 * graphite-poly.c: Don't include tree-dump.h.
30153 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
30154 * print-tree.c: Likewise.
30155 * stor-layout.c: Likewise.
30156 * tree-nested.c: Likewise.
30158 * dumpfile.c (dump_start): Use TDF_FLAGS.
30159 (dump_enable_all): Fix TDF_KIND check thinko.
30161 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
30163 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
30164 array entries to represent two legal parameterizations of the
30165 overloaded __builtin_cmpb function, as represented by the
30166 P6_OV_BUILTIN_CMPB constant.
30167 (altivec_resolve_overloaded_builtin): Add special case handling
30168 for the __builtin_cmpb function, as represented by the
30169 P6_OV_BUILTIN_CMPB constant.
30170 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
30171 (BU_P6_64BIT_2): New macro.
30172 (BU_P6_OVERLOAD_2): New macro
30173 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
30174 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
30175 (CMPB): Add overload support to represent both 32-bit and 64-bit
30176 compare-bytes function.
30177 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
30178 support for TARGET_CMPB.
30179 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
30180 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
30181 documentation of the __builtin_cmpb overloaded built-in function.
30183 2017-05-11 Richard Biener <rguenther@suse.de>
30185 PR tree-optimization/80705
30186 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
30187 bases are not vectorizable.
30189 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30191 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
30192 when counting register pressure.
30194 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30196 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
30197 (iv_ca_more_deps): Renamed to ...
30198 (iv_ca_compare_deps): ... this.
30199 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
30201 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30203 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
30205 (determine_group_iv_costs): ... here.
30206 (find_inv_vars_cb): Record inv var if it's not recorded before.
30208 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30210 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
30211 (get_shiftadd_cost): Ditto.
30213 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30215 * tree-ssa-address.c: Include header file.
30216 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
30218 (add_to_parts): Refactor.
30219 (addr_to_parts): New parameter. Update use of move_hint_to_base.
30220 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
30223 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30225 PR tree-optimization/53090
30226 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
30228 (extract_cond_operands): Detect condition with IV on both sides
30229 and return COMP_IV_EXPR_2.
30230 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
30231 (rewrite_use_compare): Simplify by removing call to function
30232 extract_cond_operands.
30234 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30236 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
30237 (extract_cond_operands): Detect condition comparing against non-
30238 invariant bound and return appropriate enum value.
30239 (find_interesting_uses_cond): Update use of extract_cond_operands.
30240 Handle its return value accordingly.
30241 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
30243 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30245 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
30246 nonlinear iv_use computation in loop invariant sensitive way.
30248 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30250 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
30251 (find_iv_candidates): Call relate_compare_use_with_all_cands.
30253 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30255 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
30256 (dump_cand): Support iv_cand.inv_exprs.
30257 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
30259 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
30262 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30264 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
30266 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
30267 as local function. Include necessary header files.
30268 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
30270 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30272 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
30274 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30276 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
30277 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
30278 RSHIFT_EXPR and BIT_NOT_EXPR.
30280 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30282 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
30283 (adjust_setup_cost): New parameter supporting round up adjustment.
30284 (struct address_cost_data): Delete.
30285 (force_expr_to_var_cost): Don't bound cost with spill_cost.
30286 (split_address_cost, ptr_difference_cost): Delete.
30287 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
30288 (struct ainc_cost_data): New struct.
30289 (get_address_cost_ainc): New function.
30290 (get_address_cost, get_computation_cost): Reimplement.
30291 (determine_group_iv_cost_address): Record inv_expr for all uses of
30293 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
30294 (iv_ca_has_deps): Reimplemented to ...
30295 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
30297 (iv_ca_extend): Call iv_ca_more_deps.
30299 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30301 * tree-ssa-address.c (struct mem_address): Move to header file.
30302 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
30303 * tree-ssa-address.h (struct mem_address): Move from C file.
30304 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
30306 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30308 * tree-affine.h (aff_combination_type): New interface.
30309 (aff_combination_zero_p): Remove static.
30310 (aff_combination_const_p): New interface.
30311 (aff_combination_singleton_var_p): New interfaces.
30313 2017-05-11 Richard Biener <rguenther@suse.de>
30315 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
30316 Skip unreachable blocks and destinations.
30317 (eliminate): Move stmt removal and fixup ...
30318 (fini_eliminate): ... here. Skip inserted exprs.
30319 (pass_pre::execute): Move fini_pre after fini_eliminate.
30320 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
30321 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
30322 PRE to get rid of dead code that has invalid SSA form and
30323 split critical edges again.
30325 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30327 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
30329 2017-05-11 Richard Biener <rguenther@suse.de>
30331 * passes.c (execute_function_todo): Verify loops if they are
30332 said to be up-to-date.
30333 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
30334 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
30336 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
30339 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
30340 handle calling assemble_external ourself.
30343 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
30344 modes with zero size. Enhance comment.
30346 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30348 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
30349 built-ins for vec_xl and vec_xst with short and char pointer
30352 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
30354 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
30355 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
30356 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
30357 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
30358 (_mm_maskz_min_round_ss): New intrinsics.
30359 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
30360 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
30361 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
30362 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
30363 (__builtin_ia32_minss_mask_round): New builtins.
30364 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
30365 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
30366 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
30368 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
30369 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
30371 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
30374 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
30376 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
30377 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
30378 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
30379 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
30380 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
30381 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
30382 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
30383 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
30384 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
30385 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
30386 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
30387 (__builtin_ia32_mulss_mask_round): New builtins.
30388 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
30389 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
30390 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
30392 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
30393 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
30395 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
30398 2017-05-10 Julia Koval <julia.koval@intel.com>
30400 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
30401 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
30402 (_mm256_setr_m128i): New intrinsics.
30404 2017-05-10 Julia Koval <julia.koval@intel.com>
30406 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
30407 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
30408 (_mm_maskz_rcp14_ss): New intrinsics.
30409 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
30410 (__builtin_ia32_rcp14ss_mask): New builtins.
30411 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
30413 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
30415 PR tree-optimization/51513
30416 * tree-cfg.c (gimple_seq_unreachable_p): New function.
30417 (assert_unreachable_fallthru_edge_p): Use it.
30418 (group_case_labels_stmt): Likewise.
30419 * tree-cfg.h: Prototype it.
30420 * stmt.c: Include cfghooks.h and tree-cfg.h.
30421 (emit_case_dispatch_table) <gap_label>: New local variable.
30422 Use it to fill dispatch table gaps.
30423 Test for default_label before updating probabilities.
30424 (expand_case) <default_label>: Remove unneeded initialization.
30425 Test for unreachable default case statement and remove its edge.
30426 Set default_label accordingly.
30427 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
30429 2017-05-10 Carl Love <cel@us.ibm.com>
30431 * config/rs6000/rs6000-c: Add support for built-in functions
30432 vector signed char vec_neg (vector signed char)
30433 vector signed short int vec_neg (vector short int)
30434 vector signed int vec_neg (vector signed int)
30435 vector signed long long vec_neg (vector signed long long)
30436 vector float vec_neg (vector float)
30437 vector double vec_neg (vector double)
30438 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
30440 * config/rs6000/altivec.h: Add define for vec_neg
30441 * doc/extend.texi: Update the built-in documentation for the
30442 new built-in functions.
30444 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30446 PR tree-optimization/77644
30447 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
30449 2017-05-10 Nathan Sidwell <nathan@acm.org>
30451 * dumpfile.h (TDI_lang_all): New.
30452 (TDF_KIND): New. Renumber others
30453 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
30455 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
30457 (get_dump_file_name): Adjust suffix generation.
30458 (dump_enable_all): Use TDF_KIND.
30459 * doc/invoke.texi (-fdump-lang-all): Document.
30461 * dumpfile.h: Tabify.
30463 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
30466 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
30467 Move member access before delete.
30469 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
30471 * tree-inline.c (expand_call_inline): Split block at stmt
30474 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
30477 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
30478 are now unused after splitting mov{sf,sd}_hardfloat.
30479 (f32_lr2): Likewise.
30480 (f32_lm): Likewise.
30481 (f32_lm2): Likewise.
30482 (f32_li): Likewise.
30483 (f32_li2): Likewise.
30484 (f32_lv): Likewise.
30485 (f32_sr): Likewise.
30486 (f32_sr2): Likewise.
30487 (f32_sm): Likewise.
30488 (f32_sm2): Likewise.
30489 (f32_si): Likewise.
30490 (f32_si2): Likewise.
30491 (f32_sv): Likewise.
30492 (f32_dm): Likewise.
30493 (f32_vsx): Likewise.
30494 (f32_av): Likewise.
30495 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
30496 For movsf, order stores so the VSX stores occur before the GPR
30497 store which encourages the register allocator to use a traditional
30498 FPR instead of a GPR. For movsd, order the stores so that the GPR
30499 store comes before the VSX stores to allow the power6 to work.
30500 This is due to the power6 not having a 32-bit integer store
30501 instruction from a FPR.
30502 (movsf_hardfloat): Likewise.
30503 (movsd_hardfloat): Likewise.
30505 2017-05-09 Martin Sebor <msebor@redhat.com>
30507 PR translation/80280
30508 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
30511 PR translation/80280
30512 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
30513 data member added in r247778.
30514 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
30516 2017-05-09 Nathan Sidwell <nathan@acm.org>
30518 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
30520 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
30523 2017-05-09 Marek Polacek <polacek@redhat.com>
30525 * doc/invoke.texi: Fix typo.
30527 2017-05-09 Richard Biener <rguenther@suse.de>
30529 * tree-vrp.c (vrp_val_is_max): Adjust comment.
30530 (vrp_val_is_min): Likewise.
30531 (set_value_range_to_value): Likewise.
30532 (set_value_range_to_nonnegative): Likewise.
30533 (gimple_assign_nonzero_p): Likewise.
30534 (gimple_stmt_nonzero_p): Likewise.
30535 (vrp_int_const_binop): Likewise. Remove unreachable case.
30536 (adjust_range_with_scev): Adjust comments.
30537 (compare_range_with_value): Likewise.
30538 (extract_range_from_phi_node): Likewise.
30539 (test_for_singularity): Likewise.
30541 2017-05-09 Richard Biener <rguenther@suse.de>
30543 * tree-vrp.c (get_single_symbol): Add assert that we don't
30544 get overflowed constants as invariant part.
30545 (compare_values_warnv): Add comment before the TREE_NO_WARNING
30546 checks. Use wi::cmp instead of recursing for integer constants.
30547 (compare_values): Just ignore whether we assumed undefined
30548 overflow instead of failing the compare.
30549 (extract_range_for_var_from_comparison_expr): Add comment before the
30550 TREE_NO_WARNING sets.
30551 (test_for_singularity): Likewise.
30552 (extract_range_from_comparison): Do not disable optimization
30553 when we assumed undefined overflow.
30554 (extract_range_basic): Remove init of unused var.
30556 2017-05-09 Richard Biener <rguenther@suse.de>
30558 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
30559 (extract_range_from_multiplicative_op_1): Adjust.
30560 (extract_range_from_binary_expr_1): Use int_const_binop.
30562 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
30565 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
30566 rs6000_store_data_bypass_p in seven define_bypass directives and
30567 in several comments.
30568 * config/rs6000/rs6000-protos.h: Add prototype for
30569 rs6000_store_data_bypass_p function.
30570 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
30571 function implements slightly different (rs6000-specific) semantics
30572 than store_data_bypass_p, returning false rather than aborting
30573 with assertion error when arguments do not satisfy the
30574 requirements of store data bypass.
30575 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
30576 rs6000_store_data_bypass_p.
30578 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
30580 * config/xtensa/xtensa-protos.h
30581 (xtensa_initial_elimination_offset): New declaration.
30582 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
30583 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
30584 macro definition, add case for FRAME_POINTER_REGNUM when
30585 FRAME_GROWS_DOWNWARD.
30586 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
30587 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
30588 xtensa_initial_elimination_offset.
30590 2017-05-08 Nathan Sidwell <nathan@acm.org>
30592 * doc/invoke.texi: Alphabetize -fdump options.
30594 2017-05-08 Martin Sebor <msebor@redhat.com>
30596 PR translation/80280
30597 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
30599 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
30601 * target.def (compute_frame_layout): New optional target hook.
30602 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
30603 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
30604 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
30606 * reload1.c (verify_initial_elim_offsets): Likewise.
30607 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
30608 (use_simple_return_p): Call arm_compute_frame_layout if needed.
30609 (arm_get_frame_offsets): Split up into this ...
30610 (arm_compute_frame_layout): ... and this function.
30612 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
30614 * config/aarch64/constraints.md (Usa): New constraint.
30615 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
30617 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
30619 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
30620 with_multilib_list after it has been checked.
30622 2017-05-08 Richard Biener <rguenther@suse.de>
30624 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
30625 (bitmap_set_subtract_values): Likewise.
30627 2017-05-08 Richard Biener <rguenther@suse.de>
30629 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
30630 (gimple_assign_nonzero): ... this and remove strict_overflow_p
30632 (gimple_stmt_nonzero_warnv_p): Rename to ...
30633 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
30635 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
30636 (extract_range_basic): Adjust, do not disable propagation on
30637 strict overflow sensitive simplification.
30638 (vrp_visit_cond_stmt): Likewise.
30640 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
30642 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
30643 body size unconditionally.
30645 2017-05-07 Jeff Law <law@redhat.com>
30648 2017-05-06 Jeff Law <law@redhat.com>
30649 PR tree-optimization/78496
30650 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
30653 PR tree-optimization/78496
30654 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
30655 (simplify_stmt_using_ranges): Call it.
30656 (vrp_dom_walker::before_dom_children): Extract equivalences
30657 from an ASSERT_EXPR with an equality comparison against a
30660 2017-05-06 Jeff Law <law@redhat.com>
30662 PR tree-optimization/78496
30663 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
30666 PR tree-optimization/78496
30667 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
30668 (simplify_stmt_using_ranges): Call it.
30669 (vrp_dom_walker::before_dom_children): Extract equivalences
30670 from an ASSERT_EXPR with an equality comparison against a
30673 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
30675 * lra-constraints.c (lra_copy_reg_equiv): New function.
30676 (split_reg): Use it to copy equivalence information from the
30677 original register to the spill register.
30679 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
30681 PR rtl-optimization/75964
30682 * simplify-rtx.c (simplify_const_relational_operation): Remove
30683 invalid handling of comparisons of integer ABS.
30685 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
30687 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
30688 initialize to zero.
30689 (init_regs): Remove declaration.
30690 (function_arg_advance_32): Initialize error_p as boolean variable.
30692 2017-05-05 Nathan Sidwell <nathan@acm.org>
30694 * store-motion.c (remove_reachable_equiv_notes): Reformat long
30695 lines. Use for (;;).
30697 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30699 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
30700 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
30701 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
30702 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
30703 VF=2 that require versioning.
30705 2017-05-05 David Malcolm <dmalcolm@redhat.com>
30707 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
30710 2017-05-05 David Malcolm <dmalcolm@redhat.com>
30712 * diagnostic.h (diagnostic_override_option_index): Convert from
30713 macro to inline function.
30715 2017-05-05 David Malcolm <dmalcolm@redhat.com>
30717 * diagnostic.c (last_module_changed_p): New function.
30718 (set_last_module): New function.
30719 (diagnostic_report_current_module): Convert macro usage to
30720 the above functions.
30721 * diagnostic.h (diagnostic_context::last_module): Strengthen
30722 from const line_map * to const line_map_ordinary *.
30723 (diagnostic_last_module_changed): Delete macro.
30724 (diagnostic_set_last_module): Delete macro.
30726 2017-05-05 David Malcolm <dmalcolm@redhat.com>
30728 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
30729 with diagnostic_report_diagnostic.
30730 (diagnostic_n_impl_richloc): Likewise.
30731 * diagnostic.h (report_diagnostic): Delete macro.
30732 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
30733 with diagnostic_report_diagnostic.
30734 * substring-locations.c (format_warning_va): Likewise.
30736 2017-05-05 David Malcolm <dmalcolm@redhat.com>
30738 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
30739 save/restor of format_spec. Move option-printing code to...
30740 (print_option_information): ...this new function, and
30741 reimplement by simply printing to the pretty_printer,
30742 rather than appending to the format string.
30744 2017-05-05 David Malcolm <dmalcolm@redhat.com>
30746 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
30747 handling logic into...
30748 (update_effective_level_from_pragmas): ...this new function.
30750 2017-05-04 Andrew Waterman <andrew@sifive.com>
30752 * config/riscv/riscv.opt (mstrict-align): New option.
30753 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
30754 (SLOW_UNALIGNED_ACCESS): Define.
30755 (riscv_slow_unaligned_access): Declare.
30756 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
30758 (riscv_slow_unaligned_access): New variable.
30759 (rocket_tune_info): Set slow_unaligned_access to true.
30760 (optimize_size_tune_info): Set slow_unaligned_access to false.
30761 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
30762 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
30763 (riscv_option_override): Set riscv_slow_unaligned_access.
30764 * doc/invoke.texi: Add -mstrict-align to RISC-V.
30766 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
30768 * config/riscv/riscv.md: Unify indentation.
30770 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
30775 * config/rs6000/rs6000.md (u code attribute): Add FIX and
30777 (extendsi<mode>2): Add support for doing sign extension via
30778 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
30779 don't have ISA 3.0 instructions.
30780 (extendsi<mode>2 splitter): Likewise.
30781 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
30782 generate the normal insns since SImode can now go in vector
30783 registers. Disallow the special UNSPECs needed for previous
30784 machines to hide SImode being used. Add new insns
30785 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
30786 (fix_trunc<mode>si2_stfiwx): Likewise.
30787 (fix_trunc<mode>si2_internal): Likewise.
30788 (fixuns_trunc<mode>si2): Likewise.
30789 (fixuns_trunc<mode>si2_stfiwx): Likewise.
30790 (fctiw<u>z_<mode>_smallint): Likewise.
30791 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
30792 of floating point to 32-bit integer from doing a direct move to
30793 the GPR registers to do a store.
30794 (fctiwz_<mode>): Break long line.
30796 2017-05-05 Bin Cheng <bin.cheng@arm.com>
30798 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
30799 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
30800 (addr_list, addr_offset_valid_p): New.
30801 (split_address_groups): Check offset validity with above function.
30802 (gt-tree-ssa-loop-ivopts.h): Include header file.
30804 2017-05-05 Nathan Sidwell <nathan@acm.org>
30806 * config.gcc (arm*-*-*): Add missing 'fi'.
30808 2017-05-05 Steve Ellcey <sellcey@cavium.com>
30810 * doc/invoke.texi (-fopt-info): Explicitly say order of options
30811 included in -fopt-info does not matter.
30812 * doc/optinfo.texi (-fopt-info): Fix description of default
30813 behavour. Explicitly say order of options included in -fopt-info
30816 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
30818 * config.gcc: Allow combinations of aprofile and rmprofile values for
30819 --with-multilib-list.
30820 * config/arm/t-multilib: New file.
30821 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
30822 variables. Remove setting of ISA and floating-point ABI in
30823 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
30824 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
30825 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
30826 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
30828 * config/arm/t-rmprofile: Likewise except for the matches changes.
30829 * doc/install.texi (--with-multilib-list): Document the combination of
30830 aprofile and rmprofile values and warn about pitfalls in doing that.
30832 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
30834 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
30835 (movdi_aarch64): Likewise.
30837 2017-05-05 Jakub Jelinek <jakub@redhat.com>
30839 PR tree-optimization/80632
30840 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
30842 (build_arrays): Initialize it for virtual phis.
30843 (fix_phi_nodes): Use it for virtual phis.
30845 PR tree-optimization/80558
30846 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
30847 [x, y] op z into [x op, y op z] for op & or | if conditions
30850 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
30851 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
30854 * config/arm/arm.md (use_literal_pool): Remove.
30855 (64-bit immediate split): No longer takes cost into consideration
30856 if arm_disable_literal_pool is enabled.
30857 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
30858 used when arm_disable_literal_pool is enabled.
30859 (arm_max_const_double_inline_cost): Remove use of
30860 arm_disable_literal_pool.
30861 (push_minipool_fix): Add assert.
30862 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
30863 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
30864 (no_literal_pool_sf_immediate): New.
30866 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30868 PR tree-optimization/80613
30869 * tree-ssa-dce.c (propagate_necessity): Remove cases for
30870 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
30872 2017-05-05 Richard Biener <rguenther@suse.de>
30874 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
30876 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
30878 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
30879 of this flag from insn conditions due to removal from r247495.
30881 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
30883 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
30885 (arm_early_store_addr_dep_ptr): Likewise.
30886 * config/arm/aarch-common-protos.h
30887 (arm_early_load_addr_dep_ptr): Add prototype.
30888 (arm_early_store_addr_dep_ptr): Likewise.
30889 * config/arm/cortex-a53.md: Add new bypasses.
30891 2017-05-05 Jakub Jelinek <jakub@redhat.com>
30893 * tree.c (next_type_uid): Change type to unsigned.
30894 (type_hash_canon): Decrement back next_type_uid if
30895 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
30896 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
30899 2017-05-04 Martin Sebor <msebor@redhat.com>
30901 * builtins.c: Fix a trivial typo in a comment.
30903 PR middle-end/79234
30904 * builtins.c (check_sizes): Adjust to handle reading past the end.
30905 Avoid printing excessive upper bound of ranges. Use %E to print
30906 tree nodes instead of converting them to %wu.
30907 (expand_builtin_memchr): New function.
30908 (compute_dest_size): Rename...
30909 (compute_objsize): ...to this.
30910 (expand_builtin_memcpy): Adjust.
30911 (expand_builtin_mempcpy): Adjust.
30912 (expand_builtin_strcat): Adjust.
30913 (expand_builtin_strcpy): Adjust.
30914 (check_strncat_sizes): Adjust.
30915 (expand_builtin_strncat): Adjust.
30916 (expand_builtin_strncpy): Adjust and simplify.
30917 (expand_builtin_memset): Adjust.
30918 (expand_builtin_bzero): Adjust.
30919 (expand_builtin_memcmp): Adjust.
30920 (expand_builtin): Handle memcmp.
30921 (maybe_emit_chk_warning): Check strncat just once.
30923 2017-05-04 Martin Sebor <msebor@redhat.com>
30925 PR preprocessor/79214
30926 PR middle-end/79222
30927 PR middle-end/79223
30928 * builtins.c (check_sizes): Add inlining context and issue
30929 warnings even when -Wno-system-headers is set.
30930 (check_strncat_sizes): Same.
30931 (expand_builtin_strncat): Same.
30932 (expand_builtin_memmove): New function.
30933 (expand_builtin_stpncpy): Same.
30934 (expand_builtin): Handle memmove and stpncpy.
30936 2017-05-04 Bin Cheng <bin.cheng@arm.com>
30938 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
30939 which is not used any more.
30941 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
30943 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
30945 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
30947 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
30948 (cortexa53_tunings): Likewise.
30949 (cortexa57_tunings): Likewise.
30950 (cortexa72_tunings): Likewise.
30951 (cortexa73_tunings): Likewise.
30953 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
30955 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
30956 Set loop alignment to 8.
30958 2017-05-04 Martin Sebor <msebor@redhat.com>
30960 PR translation/80280
30961 * builtins.c (expand_builtin_object_size): Add missing quoting to
30962 %D and like directives.
30963 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
30964 (hsa_type_for_tree_type): Same.
30965 (verify_function_arguments): Same.
30966 * symtab.c (symbol_table::change_decl_assembler_name): Same.
30967 * varasm.c (get_section): Same.
30970 2017-05-04 Martin Sebor <msebor@redhat.com>
30972 PR translation/80280
30973 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
30975 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
30977 * config/aarch64/aarch64.c (generic_addrcost_table):
30978 Change HI/TI mode setting.
30980 2017-05-04 Martin Jambor <mjambor@suse.cz>
30982 PR tree-optimization/80622
30983 * tree-sra.c (comes_initialized_p): New function.
30984 (build_accesses_from_assign): Only set write lazily when
30985 comes_initialized_p is false.
30986 (analyze_access_subtree): Use comes_initialized_p.
30987 (propagate_subaccesses_across_link): Assert !comes_initialized_p
30988 instead of testing for PARM_DECL.
30990 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30992 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
30993 constraint on operand 0 to allow more general addressing modes.
30994 Adjust output template.
30995 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
30997 * config/aarch64/aarch64-protos.h
30998 (aarch64_address_valid_for_prefetch_p): Declare prototype.
30999 * config/aarch64/constraints.md (Dp): New address constraint.
31000 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
31003 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
31005 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
31006 update use of estimate_ipcp_clone_size_and_time.
31007 (estimate_local_effects): Update use of
31008 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
31009 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
31010 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
31011 Return nonspecialized time.
31013 2017-05-04 Richard Biener <rguenther@suse.de>
31015 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
31016 for the last VUSE which def dominates the PHI. Directly call
31018 (get_continuation_for_phi_1): Remove.
31020 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
31022 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
31023 to explain the use of truncating division. Cap the number of
31024 iterations to the maximum given by nb_iterations_upper_bound,
31027 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
31029 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
31030 * configure: Regenerate.
31031 * config.in: Regenerate.
31032 * config/i386/driver-mingw32.c: new file.
31033 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
31034 * config.host: Link driver-mingw32.o on MinGW host.
31035 * doc/install.texi: Document new --enable-mingw-wildcard configure
31038 2017-05-04 Marek Polacek <polacek@redhat.com>
31040 PR tree-optimization/80612
31041 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
31043 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31044 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
31046 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
31047 (movt splitter): Likewise.
31048 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
31049 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
31050 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
31051 block for Thumb-1 with MOVT.
31052 (thumb2_legitimate_address_p): Move code block ...
31053 (can_avoid_literal_pool_for_label_p): ... into this new function.
31054 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
31056 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
31057 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
31058 "M-profile targets with the MOVT instruction".
31060 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31062 * config/arm/arm-builtins.c (arm_init_builtins): Rename
31063 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
31064 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
31066 2017-05-04 Martin Liska <mliska@suse.cz>
31068 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
31069 variable cond_code.
31071 2017-05-04 Richard Biener <rguenther@suse.de>
31073 * tree.c (array_at_struct_end_p): Handle arrays at struct
31074 end with flexarrays more conservatively. Refactor and treat
31075 arrays of arrays or aggregates more strict. Fix
31076 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
31077 * tree.c (array_at_struct_end_p): Adjust prototype.
31078 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
31079 * gimple-fold.c (get_range_strlen): Likewise.
31080 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
31082 2017-05-04 Richard Biener <rguenther@suse.de>
31084 PR tree-optimization/31130
31085 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
31087 (supports_overflow_infinity): Likewise.
31088 (is_negative_overflow_infinity): Likewise.
31089 (is_positive_overflow_infinity): Likewise.
31090 (is_overflow_infinity): Likewise.
31091 (stmt_overflow_infinity): Likewise.
31092 (overflow_infinity_range_p): Likewise.
31093 (usable_range_p): Remove as always returning true.
31094 (make_overflow_infinity): Remove.
31095 (negative_overflow_infinity): Likewise.
31096 (positive_overflow_infinity): Likewise.
31097 (avoid_overflow_infinity): Likewise.
31098 (set_value_range): Adjust accordingly.
31099 (set_value_range_to_nonnegative): Likewise, remove now unused
31100 overflow_infinity arg.
31101 (vrp_operand_equal_p): Adjust.
31102 (update_value_range): Likewise.
31103 (range_int_cst_singleton_p): Likewise.
31104 (operand_less_p): Likewise.
31105 (compare_values_warnv): Likewise.
31106 (extract_range_for_var_from_comparison_expr): Likewise.
31107 (vrp_int_const_binop): Likewise.
31108 (zero_nonzero_bits_from_vr): Likewise.
31109 (extract_range_from_multiplicative_op_1): Likewise.
31110 (extract_range_from_binary_expr_1): Likewise.
31111 (extract_range_from_unary_expr): Likewise.
31112 (extract_range_from_comparison): Likewise.
31113 (extract_range_basic): Likewise.
31114 (adjust_range_with_scev): Likewise.
31115 (compare_ranges): Likewise.
31116 (compare_range_with_value): Likewise.
31117 (dump_value_range): Likewise.
31118 (test_for_singularity): Likewise, remove strict_overflow_p parameter
31120 (simplify_cond_using_ranges): Adjust.
31122 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
31124 * brig-builtins.def: Added a builtin for class_f64.
31125 * builtin-types.def: Added a builtin type needed by class_f64.
31127 2017-05-03 Jason Merrill <jason@redhat.com>
31129 * timevar.def: Add TV_CONSTEXPR.
31131 2017-05-03 David Malcolm <dmalcolm@redhat.com>
31133 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
31135 2017-05-03 Martin Jambor <mjambor@suse.cz>
31137 * ipa-prop.c (ipa_update_after_lto_read): Removed.
31138 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
31139 * ipa-cp.c (ipcp_propagate_stage): Do not call
31140 ipa_update_after_lto_read.
31141 * ipa-inline.c (ipa_inline): Likewise.
31143 2017-05-03 Martin Jambor <mjambor@suse.cz>
31145 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
31146 tag. Added a default constructor and a destructor.
31147 (ipa_edge_args_sum_t): New class;
31148 (ipa_edge_args_sum): Declare.
31149 (ipa_edge_args_vector): Remove declaration.
31150 (IPA_EDGE_REF): Use ipa_edge_args_sum.
31151 (ipa_free_edge_args_substructures): Remove declaration.
31152 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
31153 (ipa_edge_args_info_available_for_edge_p): Likewise.
31154 * ipa-prop.c (ipa_edge_args_vector): Removed.
31155 (edge_removal_hook_holder): Likewise.
31156 (edge_duplication_hook_holder): Likewise.
31157 (ipa_edge_args_sum): New variable.
31158 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
31159 ipa_edge_args_vector.
31160 (ipa_free_edge_args_substructures): Likewise.
31161 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
31162 ipa_edge_args_vector.
31163 (ipa_edge_removal_hook): Turned into method
31164 ipa_edge_args_sum_t::remove.
31165 (ipa_edge_duplication_hook): Turned into method
31166 ipa_edge_args_sum_t::duplicate.
31167 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
31168 registering edge hooks.
31169 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
31170 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
31171 ipa_edge_args_sum instead of ipa_edge_args_vector.
31172 * ipa-profile.c (ipa_profile): Likewise.
31174 2017-05-03 Martin Jambor <mjambor@suse.cz>
31176 * symbol-summary.h (function_summary): New method exists.
31177 (function_summary::symtab_removal): Deallocate through release.
31178 (call_summary): New class.
31179 (gt_ggc_mx): New overload.
31180 (gt_pch_nx): Likewise.
31181 (gt_pch_nx): Likewise.
31183 2017-05-03 Jeff Law <law@redhat.com>
31185 PR tree-optimization/78496
31186 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
31187 from simplify_cond_using_ranges. Split off code to walk
31188 backwards through casts into ...
31189 (simplify_cond_using_ranges_2): New function.
31190 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
31191 (execute_vrp): After identifying jump threads, call
31192 simplify_cond_using_ranges_2.
31194 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
31197 * ipa-inline.h (inline_summary): Add ctor.
31198 (create_ggc): Do not use ggc_cleared_alloc.
31200 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
31201 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
31203 * gcc.c (handle_braces): Support escaping in switch matching
31205 * doc/invoke.texi (Spec Files): Document it.
31206 Remove superfluous @code markup in items.
31208 2017-05-03 David Malcolm <dmalcolm@redhat.com>
31210 * diagnostic-show-locus.c (struct column_range): New struct.
31211 (get_affected_columns): New function.
31212 (get_printed_columns): New function.
31213 (struct correction): New struct.
31214 (correction::ensure_capacity): New function.
31215 (correction::ensure_terminated): New function.
31216 (struct line_corrections): New struct.
31217 (line_corrections::~line_corrections): New dtor.
31218 (line_corrections::add_hint): New function.
31219 (layout::print_trailing_fixits): Reimplement in terms of the new
31221 (selftest::test_overlapped_fixit_printing): New function.
31222 (selftest::diagnostic_show_locus_c_tests): Call it.
31224 2017-05-03 Nathan Sidwell <nathan@acm.org>
31226 Canonicalize canonical type hashing
31227 * tree.h (type_hash_canon_hash): Declare.
31228 * tree.c (type_hash_list, attribute_hash_list): Move into
31229 type_hash_canon_hash.
31230 (build_type_attribute_qual_variant): Break out hash code calc into
31231 type_hash_canon_hash.
31232 (type_hash_canon_hash): New. Generic type hash computation.
31233 (build_range_type_1, build_array_type_1, build_function_type,
31234 build_method_type_directly, build_offset_type, build_complex_type,
31235 make_vector_type): Call it.
31237 2017-05-03 Richard Biener <rguenther@suse.de>
31239 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
31240 When all DRs have unknown misaligned do not always peel
31241 when there is a store but apply the same costing model as if
31242 there were only loads.
31244 2017-05-03 Richard Biener <rguenther@suse.de>
31247 PR tree-optimization/80492
31248 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
31249 compare_base_decls returning dont-know properly.
31251 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
31253 * config/arm/iterators.md (CCSI): New mode iterator.
31254 (arch): New mode attribute.
31255 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
31256 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
31257 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
31258 code iterator for success result mode.
31259 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
31260 the corresponding new insn generators.
31262 2017-05-03 Bin Cheng <bin.cheng@arm.com>
31265 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31266 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
31268 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
31270 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
31271 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
31272 (DDR_A): Wrap DDR argument in brackets.
31273 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
31274 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
31275 (DDR_REVERSED_P): Likewise.
31277 2017-05-03 Jakub Jelinek <jakub@redhat.com>
31279 PR tree-optimization/79472
31280 * tree-switch-conversion.c (struct switch_conv_info): Add
31281 contiguous_range and default_case_nonstandard fields.
31282 (collect_switch_conv_info): Compute contiguous_range and
31283 default_case_nonstandard fields, don't clear final_bb if
31284 contiguous_range and only the default case doesn't have the required
31286 (check_all_empty_except_final): Set default_case_nonstandard instead
31287 of failing if contiguous_range and the default case doesn't have empty
31289 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
31290 and only the default case doesn't have the required constants. Skip
31292 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
31293 if default_case_nonstandard.
31294 (build_constructors): Build constant 1 just once. Assert that default
31295 values aren't inserted in between cases if contiguous_range. Skip
31297 (build_arrays): Skip virtual phis.
31298 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
31299 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
31300 Handle virtual phis.
31301 (gen_inbound_check): Handle default_case_nonstandard case.
31302 (process_switch): Adjust check_final_bb caller. Call
31303 gather_default_values with the first non-default case instead of
31304 default case if default_case_nonstandard.
31306 2017-05-02 Nathan Sidwell <nathan@acm.org>
31308 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
31309 check. Fix formatting.
31311 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
31313 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
31314 errors when comparing specialized and unspecialized times.
31316 2017-05-02 David Malcolm <dmalcolm@redhat.com>
31318 * diagnostic-show-locus.c
31319 (layout::should_print_annotation_line_p): Make private.
31320 (layout::print_annotation_line): Make private.
31321 (layout::annotation_line_showed_range_p): Make private.
31322 (layout::show_ruler): Make private.
31323 (layout::print_source_line): Make private. Pass in line and
31324 line_width, rather than calling location_get_source_line. Drop
31326 (layout::print_leading_fixits): New method.
31327 (layout::print_any_fixits): Rename to...
31328 (layout::print_trailing_fixits): ...this, and make private.
31329 Don't print newline fixits.
31330 (diagnostic_show_locus): Move logic for printing one row into...
31331 (layout::print_line): ...this new function. Move the
31332 location_get_source_line call and error-handling from
31333 print_source_line to here. Call print_leading_fixits, and rename
31334 print_any_fixits to print_trailing_fixits.
31335 (selftest::test_fixit_insert_containing_newline): Update now that
31336 newlines are partially supported.
31337 (selftest::test_fixit_insert_containing_newline_2): New test.
31338 (selftest::test_fixit_replace_containing_newline): Update comments.
31339 (selftest::diagnostic_show_locus_c_tests): Call the new test.
31340 * edit-context.c (class added_line): New class.
31341 (class edited_line): Describe newline handling in comment.
31342 (edited_line::actually_edited_p): New method.
31343 (edited_line::print_content): Delete redundant decl.
31344 (edited_line::m_predecessors): New field.
31345 (edited_file::print_content): Call edited_line::print_content.
31346 (edited_file::print_diff): Update to support newlines.
31347 (edited_file::print_diff_hunk): Likewise.
31348 (edited_file::print_run_of_changed_lines): New function.
31349 (edited_file::print_diff_line): Convert to...
31350 (print_diff_line): ...this.
31351 (edited_file::get_effective_line_count): New function.
31352 (edited_line::edited_line): Initialize new field m_predecessors.
31353 (edited_line::~edited_line): Clean up m_predecessors.
31354 (edited_line::apply_fixit): Handle newlines.
31355 (edited_line::get_effective_line_count): New function.
31356 (edited_line::print_content): New function.
31357 (edited_line::print_diff_lines): New function.
31358 (selftest::test_applying_fixits_insert_containing_newline): New
31360 (selftest::test_applying_fixits_replace_containing_newline): New
31362 (selftest::insert_line): New function.
31363 (selftest::test_applying_fixits_multiple_lines): Add example of
31365 (selftest::edit_context_c_tests): Call the new tests.
31367 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31369 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
31370 parameter cand. Update dump information.
31371 (get_computation_cost): Update uses.
31373 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31375 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
31376 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
31377 (get_computation_at, rewrite_use_address): Update use of
31378 get_computation_aff.
31380 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31382 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
31383 (get_computation): Delete.
31384 (get_computation_cost): Implement like get_computation_cost_at.
31385 Use get_computation_at.
31386 (get_computation_cost_at): Delete.
31387 (rewrite_use_nonlinear_expr): Use get_computation_at.
31388 (rewrite_use_compare, remove_unused_ivs): Ditto.
31390 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31392 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
31394 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31396 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
31397 (ivopts_global_cost_for_size): Rename parameter and update uses.
31398 (iv_ca_recount_cost): Update uses.
31399 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
31400 candidates seperately in n_invs and n_cands.
31401 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
31403 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31405 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
31406 (find_inv_vars_cb): New.
31407 (find_depends): Renamed to ...
31408 (find_inv_vars): ... this.
31409 (add_candidate_1, force_var_cost): Call find_inv_vars.
31410 (split_address_cost, determine_group_iv_cost_cond): Ditto.
31412 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31414 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
31415 inv_vars. Add inv_exprs.
31416 (struct iv_cand): Rename depends_on to inv_vars.
31417 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
31418 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
31419 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
31420 (dump_cand): Dump inv_vars.
31421 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
31422 (record_invariant, find_depends, add_candidate_1): Ditto.
31423 (set_group_iv_cost, force_var_cost): Ditto.
31424 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
31425 (get_computation_cost_at, get_computation_cost): Ditto.
31426 (determine_group_iv_cost_generic): Ditto.
31427 (determine_group_iv_cost_address): Ditto.
31428 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
31429 (determine_group_iv_costs): Ditto.
31430 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
31431 (iv_ca_set_remove_invariants): Renamed to ...
31432 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
31433 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
31434 (iv_ca_set_add_invariants): Renamed to ...
31435 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
31436 (iv_ca_set_cp): Use iv_ca_set_add_invs.
31437 (iv_ca_has_deps): Support inv_vars and inv_exprs.
31438 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
31439 (create_new_ivs): Remove useless dump.
31441 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31443 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
31445 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
31446 (iv_ca_set_no_cp, create_new_iv): Ditto.
31448 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31450 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
31452 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
31454 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
31455 function tree_check2.
31457 2017-05-02 Martin Liska <mliska@suse.cz>
31459 * doc/gcov.texi: Add missing preposition.
31460 * gcov.c (function_info::function_info): Properly fill up
31461 all member variables.
31463 2017-05-02 Tamar Christina <tamar.christina@arm.com>
31465 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
31467 2017-05-02 Tamar Christina <tamar.christina@arm.com>
31469 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
31471 2017-05-02 Martin Liska <mliska@suse.cz>
31474 * lto-streamer-in.c (lto_read_tree_1): Remove
31475 LTO_STREAMER_DEBUG.
31476 * lto-streamer.c (struct tree_hash_entry): Likewise.
31477 (struct tree_entry_hasher): Likewise.
31478 (tree_entry_hasher::hash): Likewise.
31479 (tree_entry_hasher::equal): Likewise.
31480 (lto_streamer_init): Likewise.
31481 (lto_orig_address_map): Likewise.
31482 (lto_orig_address_get): Likewise.
31483 (lto_orig_address_remove): Likewise.
31484 * lto-streamer.h: Likewise.
31485 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
31486 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
31488 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
31490 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
31491 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
31492 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
31493 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
31494 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
31495 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
31496 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
31497 (mm_maskz_sub_ss): New intrinsics.
31498 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31499 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
31500 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
31501 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
31502 (__builtin_ia32_subss_mask_round): New builtins.
31503 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31504 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
31505 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
31507 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
31508 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
31510 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
31513 2017-05-02 Martin Jambor <mjambor@suse.cz>
31515 PR tree-optimization/78687
31516 * tree-sra.c (access): New field parent.
31517 (process_subtree_disqualification): New function.
31518 (disqualify_candidate): Call it.
31519 (build_accesses_from_assign): Reset write flag if creating an
31521 (build_access_subtree): Fill in parent field and also prpagate
31522 down grp_write flag.
31523 (create_artificial_child_access): New parameter set_grp_write, set
31524 grp_write to its value.
31525 (propagate_subaccesses_across_link): Also propagate grp_write flag
31527 (propagate_all_subaccesses): Push the closest parent back to work
31528 queue if add_access_to_work_queue returned true.
31530 2017-05-02 Richard Biener <rguenther@suse.de>
31532 * common.opt (fstrict-overflow): Alias negative to fwrapv.
31533 * doc/invoke.texi (fstrict-overflow): Remove all traces of
31534 -fstrict-overflow documentation.
31535 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
31536 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
31537 flag_strict_overflow.
31538 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
31539 * lto-opts.c (lto_write_options): Do not stream it.
31540 * lto-wrapper.c (merge_and_complain): Do not handle it.
31541 * opts.c (default_options_table): Do not set -fstrict-overflow.
31542 (finish_options): Likewise do not clear it when sanitizing.
31543 * simplify-rtx.c (simplify_const_relational_operation): Do not
31544 test flag_strict_overflow.
31546 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
31548 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
31549 using enabled attribute.
31550 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
31551 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
31552 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
31553 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
31554 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
31555 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
31556 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
31557 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
31558 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
31559 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
31561 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
31563 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
31565 2017-05-02 Richard Biener <rguenther@suse.de>
31567 PR tree-optimization/80591
31569 2017-04-10 Richard Biener <rguenther@suse.de>
31571 * tree-ssa-structalias.c (find_func_aliases): Properly handle
31574 2017-05-02 Richard Biener <rguenther@suse.de>
31576 PR tree-optimization/80549
31577 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
31578 (cleanup_tree_cfg_noloop): Create forwarders to known loop
31579 headers if they do not have a preheader.
31581 2017-05-02 Martin Liska <mliska@suse.cz>
31584 * common.opt: Fix typo.
31585 * doc/invoke.texi: Likewise.
31587 2017-05-01 Jan Beulich <jbeulich@suse.com>
31589 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
31590 swapping, add (x,x,m,x,n) alternative.
31592 2017-05-01 Nathan Sidwell <nathan@acm.org>
31594 * calls.c (combine_pending_stack_adjustment_and_call): Remove
31595 unnecessary unadjusted_alignment check.
31597 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
31600 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
31602 * gimplify.c (gimplify_cilk_detach): New function.
31603 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
31604 * tree-core.h: Document EXPR_CILK_SPAWN.
31605 * tree.h (EXPR_CILK_SPAWN): Define.
31607 2017-05-01 David Malcolm <dmalcolm@redhat.com>
31609 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
31610 to use new fixit_hint representation, using the "replace" logic.
31611 (get_line_span_for_fixit_hint): Likewise.
31612 (layout::print_any_fixits): Likewise.
31613 (selftest::test_one_liner_many_fixits): Rename to...
31614 (selftest::test_one_liner_many_fixits_1): ...this, and update
31615 comment and expected output to reflect that the multiple fix-it
31616 hints are now consolidated into one insertion.
31617 (selftest::test_one_liner_many_fixits_2): New test.
31618 (selftest::test_diagnostic_show_locus_one_liner): Update for
31620 (selftest::test_fixit_consolidation): Update for fix-it API
31622 * diagnostic.c (print_parseable_fixits): Likewise.
31623 * edit-context.c (edited_line::m_line_events): Convert from
31624 auto_vec <line_event *> to auto_vec <line_event>.
31625 (class line_event): Convert from abstract base class to a concrete
31626 class, taking over the role of replace_event.
31627 (class insert_event): Delete.
31628 (class replace_event): Rename to class line_event. Convert to
31630 (edit_context::add_fixits): Reimplement.
31631 (edit_context::apply_insert): Delete.
31632 (edit_context::apply_replace): Rename to...
31633 (edit_context::apply_fixit): ...this. Convert to half-open range.
31634 (edited_file::apply_insert): Delete.
31635 (edited_file::apply_replace): Rename to...
31636 (edited_file::apply_fixit): ...this.
31637 (edited_line::~edited_line): Drop deletion of events.
31638 (edited_line::apply_insert): Delete.
31639 (edited_line::apply_replace): Rename to...
31640 (edited_line::apply_fixit): ...this. Convert to half-open range.
31641 Update for change to type of m_line_events.
31642 * edit-context.h (edit_context::apply_insert): Delete.
31643 (edit_context::apply_replace): Rename to...
31644 (edit_context::apply_fixit): ...this.
31646 2017-05-01 Martin Sebor <msebor@redhat.com>
31648 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
31651 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
31654 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
31655 __get_cpuid_max returns 0.
31656 (__get_cpuid_count): Ditto.
31658 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
31660 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
31661 replacement expression is another instance of one of its arguments.
31663 2017-05-01 Jakub Jelinek <jakub@redhat.com>
31666 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
31667 check for stack push/pop autoinc.
31668 * config/i386/i386.c (ix86_agi_dependent): Return false
31669 if the only reason why modified_in_p returned true is that
31670 addr is SP based and set_insn is a push or pop.
31672 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
31674 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
31677 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
31680 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
31681 (account_size_time): Use two predicates - exec_pred and
31683 (evaluate_conditions_for_known_args): Compute both clause and
31685 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
31686 (inline_summary_t::duplicate): Update.
31687 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
31689 (compute_inline_parameters): Likewise.
31690 (estimate_edge_size_and_time): Update caluclation of time.
31691 (estimate_node_size_and_time): Compute both time and nonspecialized
31693 (estimate_ipcp_clone_size_and_time): Update.
31694 (inline_merge_summary): Update.
31695 (do_estimate_edge_time): Update.
31696 (do_estimate_edge_size): Update.
31697 (do_estimate_edge_hints): Update.
31698 (inline_read_section, inline_write_summary): Stream both new predicates.
31699 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
31701 (compute_inlined_call_time): Cleanup.
31702 (big_speedup_p): Update.
31703 (edge_badness): Update.
31704 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
31705 (size_time_entry): Replace predicate by exec_predicate and
31706 nonconst_predicate.
31707 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
31708 (estimate_edge_time): Return also nonspec_time.
31709 (reset_edge_growth_cache): Update.
31711 2017-04-29 Jakub Jelinek <jakub@redhat.com>
31713 PR rtl-optimization/80491
31714 * ifcvt.c (noce_process_if_block): When looking for x setter
31715 with missing else_bb, don't check only the insn right before
31716 cond_earliest, but look for the last insn that x is modified in
31717 within the same bb.
31719 PR rtl-optimization/80491
31720 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
31722 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
31724 PR tree-optimization/80487
31725 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
31727 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31729 PR tree-optimization/79697
31730 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
31731 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
31732 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
31734 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
31735 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
31737 2017-04-28 Martin Sebor <msebor@redhat.com>
31739 PR tree-optimization/80523
31740 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
31741 (init_target_to_host_charmap, target_to_host, target_strtol10): New
31743 (maybe_warn, format_directive, parse_directive): Use new functions.
31744 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
31746 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
31748 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
31750 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
31752 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
31753 target_header_dir): Set correctly.
31754 * configure: Regenerated.
31755 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
31756 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
31757 instead of SYSTEM_HEADER_DIR.
31759 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
31761 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
31762 (estimate_local_effects): Likewise.
31763 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
31764 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
31765 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
31766 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
31767 do_estimate_edge_time, estimate_edge_time): Likewise.
31768 * ipa-inline-analysis.c (estimate_node_size_and_time,
31769 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
31770 (estimate_time_after_inlining): Remove.
31772 2017-04-28 Martin Liska <mliska@suse.cz>
31774 * doc/gcov.texi: Enhance documentation of gcov.
31776 2017-04-28 Martin Liska <mliska@suse.cz>
31778 * doc/gcov.texi: Sort options in alphabetic order.
31779 * doc/gcov-dump.texi: Likewise.
31780 * doc/gcov-tool.texi: Likewise.
31781 * gcov.c (print_usage): Likewise.
31782 * gcov-dump.c (print_usage): Likewise.
31783 * gcov-tool.c (print_merge_usage_message): Likewise.
31784 (print_rewrite_usage_message): Likewise.
31785 (print_overlap_usage_message): Likewise.
31787 2017-04-28 Martin Liska <mliska@suse.cz>
31789 PR gcov-profile/53915
31790 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
31792 2017-04-28 Martin Liska <mliska@suse.cz>
31794 PR gcov-profile/79891
31795 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
31796 is marked by compiler as living on a line.
31797 (get_cycles_count): Remove usage of the union.
31798 (output_intermediate_file): Likewise.
31799 (find_source): Fix GNU coding style.
31800 (accumulate_line_counts): Remove old non-all block mode.
31801 (output_lines): Remove usage of the union.
31802 * profile.c (output_location): Include all BBs, even if
31803 belonging to a same line (and file) as a previous BB.
31805 2017-04-28 Martin Liska <mliska@suse.cz>
31807 * gcov.c (process_args): Handle new argument 'w'.
31808 (read_graph_file): Assign ID to BBs.
31809 (output_branch_count): Display BB # if verbose flag is set.
31810 (output_lines): Likewise for arcs.
31811 (print_usage): Add '--verbose' option help.
31812 * doc/gcov.texi: Document --verbose (-w) option.
31814 2017-04-28 Martin Liska <mliska@suse.cz>
31816 * gcov.c (struct block_location_info): New struct.
31817 (process_file): Fill up the new structure.
31818 (read_graph_file): Replace usage of encoding by the newly added
31820 (add_line_counts): Likewise.
31821 (accumulate_line_counts): Remove usage of the union.
31822 (function_info::function_info): New function.
31823 (function_info::~function_info): Likewise.
31824 (process_file): Call delete instead of release_function.
31825 (release_function): Release the function.
31826 (release_structures): Call delete instead of release_function.
31827 (solve_flow_graph): Replace usage of num_blocks.
31828 (find_exception_blocks): Likewise.
31829 (output_lines): Fix GNU coding style.
31831 2017-04-28 Martin Liska <mliska@suse.cz>
31834 * coverage.c (coverage_remove_note_file): New function.
31835 * coverage.h: Declare the function.
31836 * toplev.c (finalize): Clean if an error has been seen.
31838 2017-04-28 Martin Liska <mliska@suse.cz>
31840 PR gcov-profile/80031
31841 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
31842 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
31843 * gcov.c (read_graph_file): Read just number of blocks.
31844 * profile.c (branch_prob): Do not stream 0 flags per a basic
31847 2017-04-28 Martin Liska <mliska@suse.cz>
31849 * gcov-dump.c (tag_*): Add new argument to declarations.
31850 (dump_gcov_file): Likewise.
31851 (tag_blocks): Add and use new argument depth.
31852 (tag_arcs): Likewise.
31853 (tag_lines): Likewise.
31854 (tag_counters): Likewise.
31855 (tag_summary): Likewise.
31856 (dump_working_sets): Use depth to do a proper indentation.
31858 2017-04-28 Jakub Jelinek <jakub@redhat.com>
31861 * cgraph.h (symtab_node::debug_symtab): No longer inline.
31862 * symtab.c (symtab_node::debug_symtab): Move definition here.
31864 2017-04-28 Richard Biener <rguenther@suse.de>
31866 * lto-streamer.h (LTO_major_version): Bump to 7.
31868 2017-04-28 Richard Biener <rguenther@suse.de>
31870 * tree-vrp.c (assert_info): New struct.
31871 (add_assert_info): New helper.
31872 (register_edge_assert_for_2): Refactor to add asserts to a vector
31874 (register_edge_assert_for_1): Likewise.
31875 (register_edge_assert_for): Likewise.
31876 (finish_register_edge_assert_for): New helper actually registering
31877 asserts where live on edge.
31878 (find_conditional_asserts): Adjust.
31879 (find_switch_asserts): Likewise.
31880 (evrp_dom_walker::try_find_new_range): Generalize.
31881 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
31883 2017-04-27 Marek Polacek <polacek@redhat.com>
31886 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
31887 arg10 and arg11 to itype.
31889 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
31891 * doc/extend.texi (Object Size Checking): Improve grammar.
31893 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
31896 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
31897 that the logic for permitting reciprocal estimates matches that
31900 2017-04-27 Jakub Jelinek <jakub@redhat.com>
31903 * tree.c (type_cache_hasher::equal): Only compare
31904 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
31905 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
31906 non-aggregate element types.
31907 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
31908 about the flag on ARRAY_TYPEs in the comment, formatting fix.
31910 2017-04-27 Richard Biener <rguenther@suse.de>
31912 PR middle-end/80533
31913 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
31914 stripping ARRAY_REFs from MEM_EXPR make sure we're not
31915 keeping a reference to a trailing array.
31917 2017-04-27 Richard Biener <rguenther@suse.de>
31919 PR middle-end/80539
31920 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
31921 being in loop-closed SSA form conservatively.
31922 (chrec_fold_multiply_poly_poly): Likewise.
31924 2017-04-27 Tamar Christina <tamar.christina@arm.com>
31926 PR middle-end/79665
31927 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
31928 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
31930 2017-04-27 Jakub Jelinek <jakub@redhat.com>
31933 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
31934 (aarch64_function_arg_alignment): Return unsigned int again, but still
31935 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
31936 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
31937 Don't emit -Wpsabi note.
31938 (aarch64_function_arg_boundary): Likewise.
31939 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
31942 2017-04-26 Nathan Sidwell <nathan@acm.org>
31944 * tree.h (crc32_unsigned_n): Declare.
31945 (crc32_unsigned, crc32_unsigned): Make inline.
31946 * tree.c (crc32_unsigned_bits): Replace with ...
31947 (crc32_unsigned_n): ... this.
31948 (crc32_unsigned, crc32_byte): Remove.
31949 (crc32_string): Remove unnecessary braces.
31951 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
31953 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
31954 * ipa-inline-analysis.c (MAX_TIME): Remove.
31955 (account_size_time): Use sreal for time.
31956 (dump_inline_summary): Update.
31957 (estimate_function_body_sizes): Update.
31958 (estimate_edge_size_and_time): Update.
31959 (estimate_calls_size_and_time): Update.
31960 (estimate_node_size_and_time): Update.
31961 (inline_merge_summary): Update.
31962 (inline_update_overall_summary): Update.
31963 (estimate_time_after_inlining): Update.
31964 (inline_read_section): Update.
31965 (inline_write_summary): Update.
31966 * ipa-inline.c (compute_uninlined_call_time): Update.
31967 (compute_inlined_call_time): Update.
31968 (recursive_inlining): Update.
31969 (inline_small_functions): Update.
31970 (dump_overall_stats): Update.
31971 * ipa-inline.h: Include sreal.h.
31972 (size_time_entry): Turn time to sreal.
31973 (inline_summary): Turn self_time nad time to sreal.
31975 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
31977 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
31979 (sreal::stream_out, sreal::stream_in): New.
31980 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
31982 2017-04-25 Jakub Jelinek <jakub@redhat.com>
31984 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
31987 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
31990 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
31992 (dimode_scalar_chain::compute_convert_gain): Ditto.
31993 (dimode_scalar_chain::make_vector_copies): Ditto.
31994 (dimode_scalar_chain::convert_reg): Ditto.
31995 (dimode_scalar_chain::convert_insn): Ditto.
31996 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
31997 (VI248_AVX512BW_1): New mode iterator.
31998 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
31999 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
32002 2017-04-25 Martin Sebor <msebor@redhat.com>
32004 PR tree-optimization/80497
32005 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
32006 constants are representable in HOST_WIDE_INT.
32007 (parse_directive): Ditto.
32009 2017-04-25 Martin Sebor <msebor@redhat.com>
32012 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
32013 (new_zero_array): Adjust signature.
32014 (dom_info::dom_init): Used unsigned rather that size_t.
32015 (dom_info::dom_info): Same.
32017 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
32018 Jakub Jelinek <jakub@redhat.com>
32021 * config/arm/arm.c: Include gimple.h.
32022 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
32023 returns negative, increment ncrn only if it returned positive.
32024 (arm_needs_doubleword_align): Return int instead of bool,
32025 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
32026 members, but if there is any such non-FIELD_DECL
32027 > PARM_BOUNDARY aligned decl, return -1 instead of false.
32028 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
32029 returns negative, increment nregs only if it returned positive.
32030 (arm_setup_incoming_varargs): Likewise.
32031 (arm_function_arg_boundary): Emit -Wpsabi note if
32032 arm_needs_doubleword_align returns negative, return
32033 DOUBLEWORD_ALIGNMENT only if it returned positive.
32035 2017-04-25 Marek Polacek <polacek@redhat.com>
32038 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
32039 first argument to type.
32041 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
32044 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
32045 type checks to test for compatibility instead of equality.
32047 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
32048 Jakub Jelinek <jakub@redhat.com>
32051 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
32053 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
32054 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
32055 the alignment computation, but return their maximum in warn_alignment.
32056 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
32057 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
32059 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
32060 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
32063 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32065 * config/arc/simdext.md (dmpyh): Fix typo.
32067 2017-04-25 Richard Biener <rguenther@suse.de>
32069 PR tree-optimization/80492
32070 * alias.c (compare_base_decls): Handle registers with asm
32071 specification conservatively.
32072 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
32073 compare_base_decls returning dont-know properly.
32075 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32077 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
32078 (legitimate_offset_address_p): New function.
32079 (arc_legitimate_address_p): Use above function.
32081 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32083 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
32085 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32087 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
32088 ACCH registers whenever they are available.
32090 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32092 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
32093 double regs fix when not used.
32095 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32097 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
32099 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
32100 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
32102 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32104 * config/arc/arc.c (arc_output_addsi): Check for h-register class
32105 when emitting short ADD instructions.
32107 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32109 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
32111 (cmpsi_cc_c_insn): Likewise.
32112 (cbranchsi4_scratch): Compute proper instruction length using
32113 compact_hreg_operand.
32114 * config/arc/predicates.md (compact_hreg_operand): New predicate.
32116 2017-04-25 Richard Biener <rguenther@suse.de>
32118 PR middle-end/80509
32119 * passes.c (pass_manager::pass_manager): Initialize
32120 m_name_to_pass_map.
32122 2017-04-25 Richard Biener <rguenther@suse.de>
32124 PR tree-optimization/79201
32125 * tree-ssa-sink.c (statement_sink_location): Handle calls.
32127 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32130 * config/s390/vector.md: Split MEM->GPR vector moves for
32131 non-s_operand addresses.
32133 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32136 * config/s390/predicates.md (reload_const_wide_int_operand): New
32138 * config/s390/s390.md ("movti"): Remove d/P alternative.
32139 ("movti_bigconst"): New pattern definition.
32141 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
32144 * s390-protos.h (s390_expand_cs_hqi): Removed.
32145 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
32146 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
32147 modes as well as CCZ1mode and CCZmode.
32148 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
32149 signature of s390_emit_compare_and_swap.
32150 (s390_expand_cs_hqi): Likewise, make static.
32151 (s390_expand_cs_tdsi): Generate an explicit compare before trying
32152 compare-and-swap, in some cases.
32153 (s390_expand_cs): Wrapper function.
32154 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
32156 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
32157 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
32158 patterns for small and large integers. Forbid symref memory operands.
32159 Move expander to s390.c. Require cc register.
32160 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
32161 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
32162 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
32163 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
32164 symref memory operands. Remove CC mode and call s390_match_ccmode
32166 ("atomic_exchange<mode>"): Allow and implement all integer modes.
32168 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
32170 * config/s390/s390.md (define_peephole2): New peephole to help
32171 combining the load-and-test pattern with volatile memory.
32173 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
32175 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
32176 with CCZmode for TARGET_Z196.
32178 2017-04-25 Jakub Jelinek <jakub@redhat.com>
32180 PR rtl-optimization/80501
32181 * combine.c (make_compound_operation_int): Set subreg_code to SET
32182 even for AND with mask of the sign bit of mode.
32184 PR rtl-optimization/80500
32185 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
32186 sum's initial value.
32188 2017-04-25 Julian Brown <julian@codesourcery.com>
32189 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32191 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
32193 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
32195 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
32197 2017-04-25 Julian Brown <julian@codesourcery.com>
32198 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32200 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
32201 (thunderx2t99_sha): New Reservation.
32203 2017-04-25 Julian Brown <julian@codesourcery.com>
32204 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32206 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
32207 type for 1-element load.
32209 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
32211 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
32213 2017-04-24 Martin Jambor <mjambor@suse.cz>
32215 PR tree-optimization/80293
32216 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
32217 char arrays not totally scalarizable if it is false.
32218 (analyze_all_variable_accesses): Pass correct value in the new
32219 parameter. Add a statistics counter.
32221 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
32223 PR middle-end/79931
32224 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
32226 2017-04-24 Richard Biener <rguenther@suse.de>
32228 PR tree-optimization/80494
32229 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
32230 out for complex types.
32232 2017-04-24 Richard Biener <rguenther@suse.de>
32234 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
32235 * tree-ssa-sccvn.c (print_scc): Print SCC size.
32236 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
32237 (DFS): Adjust and never fail.
32238 (sccvn_dom_walker::fail): Remove.
32239 (sccvn_dom_walker::before_dom_children): Adjust.
32240 (run_scc_vn): Likewise and never fail.
32241 * tree-ssa-pre.c (pass_pre::execute): Adjust.
32242 (pass_fre::execute): Likewise.
32244 2017-04-24 Richard Biener <rguenther@suse.de>
32246 PR tree-optimization/79725
32247 * tree-ssa-sink.c (statement_sink_location): Return whether
32248 failure reason was zero uses. Move that check later.
32249 (sink_code_in_bb): Deal with zero uses by removing the stmt
32252 2017-04-24 Richard Biener <rguenther@suse.de>
32255 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
32256 pointer-based references.
32258 2017-04-24 Richard Biener <rguenther@suse.de>
32261 * pass_manager.h (pass_manager::operator new): Remove.
32262 (pass_manager::operator delete): Likewise.
32263 * passes.c (pass_manager::operator new): Remove.
32264 (pass_manager::operator delete): Likewise.
32265 (pass_manager::pass_manager): Zero individual pass members.
32267 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
32270 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
32271 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
32272 Check "XEXP (src, 1)" operand here.
32273 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
32274 Check "XEXP (src, 1)" operand here.
32275 (dimode_scalar_chain::make_vector_copies): Detect count register
32276 of a shift instruction. Zero extend count register from QImode
32277 to DImode to satisfy vector shift pattern count operand predicate.
32278 Substitute vector shift count operand with a DImode copy.
32279 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
32282 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
32284 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
32285 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
32286 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
32287 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
32288 (UNSPEC_NOREX_MEM): Remove definition.
32290 2017-04-21 Richard Biener <rguenther@suse.de>
32292 PR tree-optimization/79547
32293 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
32294 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
32295 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
32296 without any constraints.
32298 2017-04-21 Richard Biener <rguenther@suse.de>
32300 PR tree-optimization/78847
32301 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
32303 2017-04-21 Richard Biener <rguenther@suse.de>
32305 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
32306 (build_distinct_type_copy): Likewise.
32307 (build_variant_type_copy): Likewise.
32308 * tree.c (build_qualified_type): Pass down mem-stat info.
32309 (build_distinct_type_copy): Likewise.
32310 (build_variant_type_copy): Likewise.
32312 2017-04-21 Richard Biener <rguenther@suse.de>
32314 PR tree-optimization/80237
32315 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
32317 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
32318 for a simplified result.
32320 2016-04-21 Richard Biener <rguenther@suse.de>
32322 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
32323 sth as strict as a simple_iv but a chrec without symbols and an
32324 operand defined in the loop we are peeling (and not some subloop).
32325 (propagate_constants_for_unrolling): Propagate all constants.
32327 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
32330 * config/i386/i386.c (print_reg): Remove assert for disalowed
32331 regno values, call output_operand_lossage instead.
32333 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
32336 * config/i386/constraints.md (Yc): New register constraint.
32337 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
32338 Use Yc constraint for alternative 2 of operand 0. Remove
32339 preferred_for_speed attribute.
32341 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
32343 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
32344 lastprivate clauses in SIMT case.
32346 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
32348 * doc/invoke.texi (-Wextra-semi): Document new warning option.
32350 2017-04-20 Richard Biener <rguenther@suse.de>
32352 PR tree-optimization/57796
32353 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
32354 as N scalar stores.
32355 (vect_model_load_cost): Cost gathers as N scalar loads.
32357 2017-04-20 Richard Biener <rguenther@suse.de>
32359 * ggc-page.c (ggc_allocated_p): Rename to ...
32360 (safe_lookup_page_table_entry): ... this and return the lookup
32362 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
32364 2017-04-20 Richard Biener <rguenther@suse.de>
32366 PR tree-optimization/80453
32367 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
32368 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
32369 from the conditions.
32370 (vn_phi_eq): Pass them down.
32371 (vn_phi_lookup): Record them.
32372 (vn_phi_insert): Likewise.
32374 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
32376 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
32377 uninitialized variable warning to avoid buffer overrun.
32379 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
32382 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
32383 is suppressed for '{ 0 }' in C.
32385 2017-04-20 Jakub Jelinek <jakub@redhat.com>
32387 * BASE-VER: Set to 8.0.0.
32389 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
32391 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
32392 priority .init_array and .fini_array section with SECTION_NOTYPE
32395 2017-04-20 Jakub Jelinek <jakub@redhat.com>
32397 PR middle-end/80423
32398 * tree.h (build_array_type): Add typeless_storage default argument.
32399 * tree.c (type_cache_hasher::equal): Also compare
32400 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
32401 (build_array_type): Add typeless_storage argument, set
32402 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
32404 (build_nonshared_array_type): Adjust build_array_type_1 caller.
32405 (build_array_type): Likewise. Add typeless_storage argument.
32407 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
32408 Jakub Jelinek <jakub@redhat.com>
32410 PR tree-optimization/80426
32411 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
32412 operation on symbolic operands, also compute the overflow for the
32413 invariant part when the operation degenerates into a negation.
32415 2017-04-19 Jakub Jelinek <jakub@redhat.com>
32418 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
32419 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
32422 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
32424 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
32427 * config/avr/avr.c (tree.h): Include it.
32428 (cgraph.h): Include it.
32429 (avr_encode_section_info): Don't warn for uninitialized progmem
32430 variable if it's just an alias.
32432 2017-04-19 Richard Biener <rguenther@suse.de>
32435 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
32436 when needed by AutoPGO.
32438 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
32441 * doc/lto.texi: Remove an extra 'that'.
32443 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
32445 PR rtl-optimization/80429
32446 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
32447 are only used in debug insns.
32449 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
32450 Vladimir Makarov <vmakarov@redhat.com>
32452 * config/sparc/predicates.md (input_operand): Add comment. Return
32453 true for any memory operand when LRA is in progress.
32454 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
32456 2017-04-18 Jeff Law <law@redhat.com>
32459 * mips.md ({return,simple_return}_internal): Do not overwrite
32462 2017-04-18 Jakub Jelinek <jakub@redhat.com>
32464 PR tree-optimization/80443
32465 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
32466 instead of adding 1, subtract -1 and similarly instead of subtracting
32469 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
32471 PR rtl-optimization/80357
32472 * haifa-sched.c (tmp_bitmap): New variable.
32473 (model_recompute): Handle duplicate use records.
32474 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
32475 (free_global_sched_pressure_data): Free it.
32477 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
32480 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
32481 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
32482 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
32483 instead of SYSTEM_HEADER_DIR.
32485 2017-04-18 Jeff Law <law@redhat.com>
32487 PR middle-end/80422
32488 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
32489 predecessors after walking up the insn chain.
32491 2017-04-18 Jakub Jelinek <jakub@redhat.com>
32494 * dwarf2out.c (modified_type_die): Try harder not to emit internal
32495 sizetype type into debug info.
32497 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
32500 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
32501 unneeded test for TARGET_UPPER_REGS_SF.
32502 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
32504 2017-04-18 Jakub Jelinek <jakub@redhat.com>
32507 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
32508 instead of gsi_after_labels.
32510 2017-04-18 Jeff Law <law@redhat.com>
32512 * regcprop.c (maybe_mode_change): Avoid creating copies of the
32516 2017-04-13 Jeff Law <law@redhat.com>
32517 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
32518 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
32520 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
32523 * config/avr/avr.c (intl.h): Include it.
32524 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
32526 2017-04-18 Martin Liska <mliska@suse.cz>
32528 PR gcov-profile/78783
32529 * gcov-tool.c (gcov_output_files): Validate that destination
32530 file is either removed by the tool or by a user.
32532 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
32533 Guy Benyei <guybe@mellanox.com>
32535 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
32536 block, and do not negate it, the stored id is already negative.
32538 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
32540 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
32542 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
32545 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
32546 masks of options that should be turned off if the VSX vector
32547 options are turned off.
32548 (OTHER_P8_VECTOR_MASKS): Likewise.
32549 (OTHER_VSX_VECTOR_MASKS): Likewise.
32550 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
32551 rs6000_disable_incompatible_switches to validate no type switches
32553 (rs6000_incompatible_switch): New function to disallow turning on
32554 other vector options if -mno-vsx, -mno-power8-vector, or
32555 -mno-power9-vector are specified.
32557 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
32559 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
32561 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
32563 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
32564 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
32565 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
32566 (ARG_POINTER_CFA_OFFSET): Likewise.
32568 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
32570 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
32571 conditions to take advantage of various optimizations.
32573 2017-04-13 Jeff Law <law@redhat.com>
32575 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
32576 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
32577 (zero_extendsidi2_dext): Likewise.
32579 2017-04-13 Jakub Jelinek <jakub@redhat.com>
32582 * fold-const.c (fold_ternary_loc): Revert
32583 use op0 instead of fold_convert_loc (loc, type, arg0) part of
32586 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
32588 PR rtl-optimization/80343
32589 * lra-remat.c (update_scratch_ops): Assign original hard reg to
32590 new scratch pseudo.
32592 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
32595 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
32596 to ubsan_encode_value.
32598 2017-04-13 Jeff Law <law@redhat.com>
32600 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
32601 appearing in DEBUG_INSNs.
32603 2017-04-13 Martin Liska <mliska@suse.cz>
32605 PR gcov-profile/80413
32606 * gcov-io.c (gcov_write_string): Copy to buffer just when
32607 allocated size is greater than zero.
32609 2017-04-13 Jakub Jelinek <jakub@redhat.com>
32612 * dwarf2out.c (decls_for_scope): Ignore declarations of
32613 current_function_decl in BLOCK_NONLOCALIZED_VARS.
32615 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
32618 * ipa-visibility.c (non_local_p): Fix typos.
32619 (localize_node): When localizing symbol in same comdat group,
32620 dissolve the group only when we know external symbols are going
32622 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
32624 2017-04-12 Jakub Jelinek <jakub@redhat.com>
32626 PR tree-optimization/79390
32627 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
32628 order does not result in usable sequence, retry with reversed operand
32634 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
32635 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
32636 op0 instead of fold_convert_loc (loc, type, arg0).
32638 2017-04-12 Jeff Law <law@redhat.com>
32640 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
32641 has a delay slot in the generated code.
32643 * config/cris/cris.md (cris_preferred_reload_class): Return
32644 GENNONACR_REGS rather than GENERAL_REGS.
32646 2017-04-12 Jakub Jelinek <jakub@redhat.com>
32649 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
32650 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
32651 signedness of the result type.
32653 2017-04-12 Richard Biener <rguenther@suse.de>
32654 Jeff Law <law@redhat.com>
32656 PR tree-optimization/80359
32657 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
32658 trim stores to TARGET_MEM_REFs.
32660 2017-04-12 Richard Biener <rguenther@suse.de>
32662 PR tree-optimization/79390
32663 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
32664 threading case even more.
32666 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
32669 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
32670 for quad_address_p for TImode, instead of just not indexed_address.
32672 2017-04-12 Richard Biener <rguenther@suse.de>
32673 Bernd Edlinger <bernd.edlinger@hotmail.de>
32675 PR middle-end/79671
32676 * alias.c (component_uses_parent_alias_set_from): Handle
32677 TYPE_TYPELESS_STORAGE.
32678 (get_alias_set): Likewise.
32679 * tree-core.h (tree_type_common): Add typeless_storage flag.
32680 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
32681 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
32682 for types containing members with TYPE_TYPELESS_STORAGE.
32683 (place_field): Likewise.
32684 (layout_type): Likewise for ARRAY_TYPE.
32685 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
32686 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
32687 TYPE_TYPELESS_STORAGE.
32688 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
32690 2017-04-12 Jakub Jelinek <jakub@redhat.com>
32693 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
32694 first argument to type.
32696 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32700 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
32701 CONST0_RTX (mode) rather than const0_rtx where appropriate.
32702 (rs6000_expand_binop_builtin): Likewise.
32703 (rs6000_expand_ternop_builtin): Likewise; also add missing
32704 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
32705 vshasigma built-ins.
32706 * doc/extend.texi: Document that vec_xxpermdi's third argument
32707 must be a constant.
32709 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
32711 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
32712 Use shift_const cost parameter when calculating gain of STV shifts.
32714 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
32716 PR rtl-optimization/70478
32717 * lra-constraints.c (process_alt_operands): Check memory for
32718 disfavoring memory insn operand.
32720 2017-04-11 Jakub Jelinek <jakub@redhat.com>
32722 PR middle-end/80100
32723 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
32724 left shift in unsigned HOST_WIDE_INT type.
32726 PR rtl-optimization/80385
32727 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
32728 (not (neg X)) into (plus X -1) for complex or non-integral modes.
32731 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
32732 if they have any depend clauses.
32734 2017-04-11 Martin Liska <mliska@suse.cz>
32737 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
32738 * ipa-split.c (split_function): Create a local comdat symbol
32739 if caller is in a comdat group.
32741 2017-04-11 Martin Liska <mliska@suse.cz>
32744 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
32747 2017-04-11 Martin Sebor <msebor@redhat.com>
32749 PR middle-end/80364
32750 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
32751 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
32753 (directive::set_width, directive::set_precision, format_character):
32755 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
32758 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
32761 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
32762 conflict, set target->arch_name instead of target->cpu_name.
32764 2017-04-11 Richard Biener <rguenther@suse.de>
32766 PR tree-optimization/80374
32767 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
32768 build_zero_cst, remove fold_convertible_p check again.
32770 2017-04-11 Martin Liska <mliska@suse.cz>
32773 * ubsan.c (instrument_object_size): Do not instrument register
32776 2017-04-11 Jakub Jelinek <jakub@redhat.com>
32779 * config/i386/i386-builtin-types.def
32780 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
32781 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
32782 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
32783 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
32784 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
32785 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
32786 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
32787 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
32788 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
32789 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
32790 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
32791 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
32792 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
32793 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
32794 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
32795 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
32796 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
32797 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
32798 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
32799 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
32800 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
32801 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
32802 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
32803 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
32804 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
32805 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
32806 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
32807 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
32808 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
32809 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
32810 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
32811 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
32812 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
32813 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
32814 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
32815 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
32816 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
32817 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
32818 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
32819 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
32820 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
32821 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
32822 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
32823 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
32824 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
32825 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
32827 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
32828 flag to second_arg_count, handle 4 argument function type _COUNT
32829 aliases, handle second_arg_count on second argument rather than last.
32831 2017-04-10 Jeff Law <law@redhat.com>
32833 PR tree-optimization/80374
32834 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
32835 record anything if we can not convert integer_zero_node to the
32838 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
32841 * config/rs6000/rs6000.c (rs6000_option_override_internal):
32842 Enhance special handling given to the TARGET_P9_MINMAX option in
32843 relation to certain other options.
32845 2017-04-10 Bin Cheng <bin.cheng@arm.com>
32847 PR tree-optimization/80153
32848 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
32849 remove POINTER_PLUS_EXPR's base part directly, rather than through
32852 2017-04-10 Richard Biener <rguenther@suse.de>
32853 Bin Cheng <bin.cheng@arm.com>
32855 PR tree-optimization/80153
32856 * tree-affine.c (aff_combination_to_tree): Get base pointer from
32857 the first element of pointer type aff_tree. Build result expr in
32859 (add_elt_to_tree): Convert to type unconditionally. Remove other
32860 fold_convert calls.
32861 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
32862 (rewrite_use_nonlinear_expr): Check invariant using iv information.
32864 2017-04-10 Richard Biener <rguenther@suse.de>
32866 * tree-ssa-structalias.c (find_func_aliases): Properly handle
32869 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
32871 PR rtl-optimization/70478
32872 * lra-constraints.c (curr_small_class_check): New.
32873 (update_and_check_small_class_inputs): New.
32874 (process_alt_operands): Update curr_small_class_check. Disfavor
32875 alternative insn memory operands. Check available regs for small
32878 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
32881 * config/mips/mips.opt (-mvirt): Update description.
32882 * doc/invoke.texi (-mvirt): Likewise.
32884 2017-04-10 Richard Biener <rguenther@suse.de>
32886 PR middle-end/80362
32887 * fold-const.c (fold_binary_loc): Look at unstripped ops when
32888 looking for NEGATE_EXPR in -A / -B to A / B folding.
32890 2017-04-10 Martin Liska <mliska@suse.cz>
32892 PR gcov-profile/80224
32893 * gcov.c (print_usage): Fix usage string.
32894 (get_gcov_intermediate_filename): Remove.
32895 (output_gcov_file): Use both for normal and intermediate format.
32896 (generate_results): Do not initialize special file for
32897 intermediate format.
32899 2017-04-10 Richard Biener <rguenther@suse.de>
32901 PR tree-optimization/80304
32902 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
32905 2017-04-10 Nathan Sidwell <nathan@acm.org>
32908 * config/rs6000/rs6000.c (rs6000_vector_type): New.
32909 (rs6000_init_builtins): Use it.
32911 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32913 * config/arm/arm.md (<mrc>): Add mode to SET source.
32914 (<mrrc>): Likewise.
32916 2017-04-10 Richard Biener <rguenther@suse.de>
32918 PR middle-end/80344
32919 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
32921 2017-04-10 Jakub Jelinek <jakub@redhat.com>
32924 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
32925 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
32926 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
32927 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
32928 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
32929 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
32930 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
32931 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
32932 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
32933 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
32934 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
32935 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
32936 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
32937 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
32938 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
32939 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
32940 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
32941 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
32942 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
32943 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
32944 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
32945 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
32946 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
32948 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
32950 PR rtl-optimization/70478
32951 * lra-constraints.c: Reverse the last patch.
32953 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
32955 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
32956 Add comment for WCHAR_T.
32958 2017-04-08 Martin Liska <mliska@suse.cz>
32961 2017-04-07 Martin Liska <mliska@suse.cz>
32964 * ipa-split.c (split_function): Add function part to a same comdat
32967 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32970 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
32972 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
32974 * rs6000/rs6000.c (vec_load_pendulum): Rename...
32975 (vec_pairing): ...to this.
32976 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
32977 (rs6000_sched_init): Adjust for name change.
32978 (struct rs6000_sched_context): Likewise.
32979 (rs6000_init_sched_context): Likewise.
32980 (rs6000_set_sched_context): Likewise.
32982 2017-04-07 Jakub Jelinek <jakub@redhat.com>
32988 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
32990 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
32991 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
32992 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
32994 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
32996 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
32998 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
33000 PR rtl-optimization/70703
33001 * ira-color.c (update_conflict_hard_regno_costs): Use
33002 int64_t instead of HOST_WIDE_INT.
33004 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
33006 PR rtl-optimization/70478
33007 * lra-constraints.c (process_alt_operands): Disfavor alternative
33008 insn memory operands.
33010 2017-04-07 Jeff Law <law@redhat.com>
33012 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
33013 CALL and NOTE_INSN_CALL_ARG_LOCATION.
33015 2017-04-07 Martin Liska <mliska@suse.cz>
33018 * config/aarch64/aarch64.c (aarch64_process_target_attr):
33019 Show error message instead of an ICE.
33021 2017-04-07 Martin Liska <mliska@suse.cz>
33024 * ipa-split.c (split_function): Add function part to a same comdat
33027 2017-04-07 Richard Biener <rguenther@suse.de>
33029 PR middle-end/80341
33030 * tree.c (get_unwidened): Also handle ! for_type case for
33032 * convert.c (do_narrow): Split out from ...
33033 (convert_to_integer_1): ... here. Do not pass final truncation
33034 type to get_unwidened for TRUNC_DIV_EXPR.
33036 2017-04-07 Richard Biener <rguenther@suse.de>
33038 * tree-affine.c (wide_int_ext_for_comb): Take type rather
33040 (aff_combination_const): Adjust.
33041 (aff_combination_scale): Likewise.
33042 (aff_combination_add_elt): Likewise.
33043 (aff_combination_add_cst): Likewise.
33044 (aff_combination_convert): Likewise.
33045 (add_elt_to_tree): Likewise. Remove unused argument.
33046 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
33048 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
33050 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
33052 * config/arm/arm.c (arm_default_short_enums): Use
33053 ARM_DEFAULT_SHORT_ENUMS.
33054 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
33056 2017-04-06 Jakub Jelinek <jakub@redhat.com>
33059 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
33060 members with redundant out-of-class redeclaration.
33062 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
33065 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
33066 * config/i386/i386.md (*zero_extendsidi2):
33067 Add (?*x,*x) and (?*v,*v) alternatives.
33069 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
33072 * config/i386/i386.c (ix86_expand_builtin)
33073 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
33074 mode from insn data. Convert operands to insn operand mode.
33075 Copy operands that don't satisfy insn predicate to a register.
33077 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
33079 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
33082 2017-04-06 Richard Biener <rguenther@suse.de>
33084 PR tree-optimization/80334
33085 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
33086 preserve alignment of accesses.
33088 2017-04-06 Richard Biener <rguenther@suse.de>
33090 PR tree-optimization/80262
33091 * tree-sra.c (build_ref_for_offset): Preserve address-space
33093 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
33094 Drop useless address-space information on MEM_REF offsets.
33096 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
33098 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
33100 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
33102 PR rtl-optimization/70703
33103 * ira-color.c (update_conflict_hard_regno_costs): Use
33104 HOST_WIDE_INT instead of long.
33106 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
33109 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
33110 not defined for x86_64 target. Add -mmmx target option when __SSE2__
33112 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
33113 for x86_64 target. Handle -m3dnowa option.
33115 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
33117 PR rtl-optimization/70703
33118 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
33119 (update_conflict_hard_regno_costs): Use long instead of unsigned
33120 arithmetic for cost calculation.
33122 2017-04-05 Jakub Jelinek <jakub@redhat.com>
33123 Bernd Edlinger <bernd.edlinger@hotmail.de>
33126 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
33129 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
33132 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
33133 ptr_mode with Pmode throughout.
33134 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
33135 into probe_stack_range and use DImode.
33137 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
33140 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
33141 call_eh_return is true.
33143 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33145 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
33146 Initialize last_match_fntype_index.
33148 2017-04-05 Jakub Jelinek <jakub@redhat.com>
33151 * tree-nvr.c: Include internal-fn.h.
33152 (pass_return_slot::execute): Ignore internal calls without
33155 2017-04-04 Jakub Jelinek <jakub@redhat.com>
33156 Richard Biener <rguenther@suse.de>
33159 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
33160 captures used multiple times, except for the last use.
33161 * generic-match-head.c: Include gimplify.h.
33163 2017-04-04 Jakub Jelinek <jakub@redhat.com>
33165 PR tree-optimization/79390
33166 * target.h (struct noce_if_info): Declare.
33167 * targhooks.h (default_noce_conversion_profitable_p): Declare.
33168 * target.def (noce_conversion_profitable_p): New target hook.
33169 * ifcvt.h (struct noce_if_info): New type, moved from ...
33170 * ifcvt.c (struct noce_if_info): ... here.
33171 (noce_conversion_profitable_p): Renamed to ...
33172 (default_noce_conversion_profitable_p): ... this. No longer
33174 (noce_try_store_flag_constants, noce_try_addcc,
33175 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
33176 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
33177 instead of noce_conversion_profitable_p.
33178 * config/i386/i386.c: Include ifcvt.h.
33179 (ix86_option_override_internal): Don't override
33180 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
33181 (ix86_noce_conversion_profitable_p): New function.
33182 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
33183 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
33184 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
33185 * doc/tm.texi: Regenerated.
33187 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33189 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
33192 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
33195 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
33196 instructions for small multiply cores.
33198 2017-04-04 Jeff Law <law@redhat.com>
33200 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
33202 (mips_expand_vec_perm_const): Initialize elements in orig_perm
33203 that are not set by the loop over the elements.
33205 2017-04-04 Jakub Jelinek <jakub@redhat.com>
33208 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
33209 int mode, convert_modes it to mode as unsigned, otherwise use
33210 lowpart_subreg to mode rather than SImode.
33211 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
33212 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
33213 Use DImode instead of SImode for the shift count operand.
33214 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
33217 2017-04-04 Richard Biener <rguenther@suse.de>
33219 PR middle-end/80281
33220 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
33221 arithmetic done for the negate or the plus. Simplify.
33222 (A - (-B) -> A + B): Likewise.
33223 * fold-const.c (split_tree): Make sure to not negate pointers.
33225 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
33227 PR rtl-optimization/60818
33228 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
33229 a compare of comparisons with the thing compared if this results
33230 in a different machine mode.
33232 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
33234 * alias.c (base_alias_check): Fix typo in comment.
33235 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
33236 * cgraphunit.c (symbol_table::compile): Likewise.
33237 * collect2.c (maybe_run_lto_and_relink): Likewise.
33238 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
33239 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
33240 * config/avr/avr.c (avr_map_op_t): Likewise.
33241 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
33242 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
33243 * config/epiphany/epiphany.md (movcc): Likewise.
33244 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
33245 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
33247 * config/mips/mips.c (mips_save_restore_reg): Likewise.
33248 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
33249 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
33250 * config/sh/sh.c (sh_rtx_costs): Likewise.
33251 * fold-const.c (fold_truth_andor): Likewise.
33252 * genautomata.c (collapse_flag): Likewise.
33253 * gengtype.h (struct type::u::s): Likewise.
33254 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
33255 * input.c (FORMAT_AMOUNT): Likewise.
33256 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
33257 (known_aggs_to_agg_replacement_list): Likewise.
33258 * ipa-inline-analysis.c: Likewise.
33259 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
33260 * ipa-polymorphic-call.c
33261 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
33262 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
33263 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
33265 * modulo-sched.c (apply_reg_moves): Likewise.
33266 * omp-expand.c (build_omp_regions_1): Likewise.
33267 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
33268 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
33269 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
33270 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
33271 * value-prof.c: Likewise.
33272 * var-tracking.c (val_reset): Likewise.
33274 2017-04-03 Richard Biener <rguenther@suse.de>
33276 PR tree-optimization/80275
33277 * fold-const.c (split_address_to_core_and_offset): Handle
33280 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
33282 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
33283 descriptors is at least equal to that of functions.
33285 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
33287 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
33289 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
33292 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
33293 (mov<IMOD4:mode>): New expander.
33294 (*mov<IMOD4:mode>_internal): New insn and split pattern.
33296 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
33298 PR rtl-optimization/79405
33299 * fwprop.c (propagations_left): New variable.
33300 (forward_propagate_into): Decrement it.
33301 (fwprop_init): Initialize it.
33302 (fw_prop): If the variable has reached zero, stop propagating.
33303 (fwprop_addr): Ditto.
33305 2017-03-31 Jakub Jelinek <jakub@redhat.com>
33308 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
33309 a FUNCTION_DECL, pass it as decl instead of origin to
33312 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
33314 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
33317 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
33320 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
33321 TARGET_VSX_SMALL_INTEGER.
33323 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33325 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
33326 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
33328 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
33330 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
33331 extraction from odd-numbered MSA register.
33333 2017-03-31 Jakub Jelinek <jakub@redhat.com>
33335 PR middle-end/80173
33336 * expmed.c (store_bit_field_1): Don't attempt to create
33337 a word subreg out of hard registers wider than word if they
33338 have HARD_REGNO_NREGS of 1 for their mode.
33340 PR middle-end/80163
33341 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
33342 conversions to integer types wider than word and pointer.
33345 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
33346 (rtx_equal_for_cselib_p): Pass 0 to it.
33347 * cselib.c (cselib_hasher::equal): Likewise.
33348 (rtx_equal_for_cselib_1): Add depth argument. If depth
33349 is 128, don't look up VALUE locs and punt. Increment
33350 depth in recursive calls when walking VALUE locs.
33352 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
33354 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
33355 (make_gcov_file_name): Use the canonical path name for generating
33357 (read_line): Fix handling of files with ascii null bytes.
33359 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
33361 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
33362 to initialise a vector register instead
33363 of using a const_int.
33365 2017-03-30 Jakub Jelinek <jakub@redhat.com>
33367 PR translation/80189
33368 * gimplify.c (omp_default_clause): Use %qs instead of %s in
33369 diagnostic messages.
33371 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
33374 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
33375 (dfp_diex_<mode>): Update mode of operand 1.
33376 * doc/extend.texi (dxex, dxexq): Document change to return type.
33377 (diex, diexq): Document change to argument type.
33379 2017-03-30 Martin Jambor <mjambor@suse.cz>
33382 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
33383 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
33384 it reflects the signature changes performed at the callee side.
33385 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
33386 to cgraph_build_function_type_skip_args.
33387 (build_function_decl_skip_args): Adjust call to the above function.
33389 2017-03-30 Jakub Jelinek <jakub@redhat.com>
33392 * config/i386/sse.md
33393 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
33394 register as dest whenever it is a MEM not rtx_equal_p to the
33395 corresponding dup operand, and when forcing into reg move the
33396 reg into the memory afterwards.
33397 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
33398 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
33399 for the force_reg mode.
33400 (avx512vl_vextractf128<mode>): Use register as dest either
33401 always when a MEM, or when it is a MEM not rtx_equal_p to the
33402 corresponding dup operand, or even not when it is a CONST_VECTOR
33403 depending on the mode and lo vs. hi.
33404 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
33406 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
33407 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
33408 Likewise. Require that operands[2] is even.
33409 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
33410 Remove extraneous parens. Require that operands[2] is a multiple
33412 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
33413 operands[0] is a MEM if <mask_applied>, the predicates/constraints
33414 disallow memory then.
33416 2017-03-30 Richard Biener <rguenther@suse.de>
33418 PR tree-optimization/77498
33419 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
33420 to non-constants over backedges.
33422 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
33424 PR rtl-optimization/80233
33425 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
33426 as last_combined_insn. Do not test for BARRIER_P separately.
33428 2017-03-29 Andreas Schwab <schwab@suse.de>
33431 * calls.c (prepare_call_address): Convert funexp to Pmode before
33432 copying to temp reg.
33434 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33436 PR tree-optimization/80158
33437 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
33438 Handle possible future case of more than one alternate
33440 (replace_rhs_if_not_dup): Likewise.
33441 (replace_one_candidate): Likewise.
33443 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
33445 PR rtl-optimization/80193
33446 * ira.c (ira): Do not check allocation for LRA.
33448 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
33450 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
33451 (nvptx_output_simt_exit): Declare.
33452 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
33453 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
33454 (init_softstack_frame): Move initialization of crtl->is_leaf to...
33455 (nvptx_declare_function_name): ...here. Emit declaration of local
33456 memory space buffer for omp_simt_enter insn.
33457 (nvptx_output_unisimt_switch): New.
33458 (nvptx_output_softstack_switch): New.
33459 (nvptx_output_simt_enter): New.
33460 (nvptx_output_simt_exit): New.
33461 * config/nvptx/nvptx.h (struct machine_function): New fields
33462 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
33463 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
33464 (UNSPECV_SIMT_EXIT): Ditto.
33465 (omp_simt_enter_insn): New insn.
33466 (omp_simt_enter): New expansion.
33467 (omp_simt_exit): New insn.
33468 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
33470 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
33471 (expand_GOMP_SIMT_ENTER_ALLOC): New.
33472 (expand_GOMP_SIMT_EXIT): New.
33473 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
33474 (GOMP_SIMT_ENTER_ALLOC): Ditto.
33475 (GOMP_SIMT_EXIT): Ditto.
33476 * target-insns.def (omp_simt_enter): New insn.
33477 (omp_simt_exit): Ditto.
33478 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
33480 (lower_rec_simd_input_clauses): Implement SIMT privatization.
33481 (lower_rec_input_clauses): Likewise.
33482 (lower_lastprivate_clauses): Handle SIMT privatization.
33484 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
33485 (ompdevlow_adjust_simt_enter): New.
33486 (find_simtpriv_var_op): New.
33487 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
33488 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
33490 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
33491 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
33492 (copy_decl_for_dup_finish): Ditto.
33494 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
33496 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
33499 * config/i386/i386.c (ix86_option_override_internal): Always
33500 allow -mpreferred-stack-boundary=3 for 64-bit targets.
33502 2017-03-28 Bin Cheng <bin.cheng@arm.com>
33504 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
33506 2017-03-28 Bin Cheng <bin.cheng@arm.com>
33508 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
33509 mark new edge's irreducible flag accordign to it.
33510 (vect_do_peeling): Check loop preheader edge's irreducible flag
33511 and pass it to function slpeel_add_loop_guard.
33513 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
33515 PR tree-optimization/80218
33516 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
33517 Update block frequencies and counts.
33519 2017-03-28 Richard Biener <rguenther@suse.de>
33521 PR tree-optimization/78644
33522 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
33523 of a simplification result we may not use it at all.
33525 2017-03-28 Richard Biener <rguenther@suse.de>
33528 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
33529 without arguments, generate default definition of a SSA name.
33531 2017-03-28 Richard Biener <rguenther@suse.de>
33533 PR middle-end/80222
33534 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
33535 TYPE_REF_CAN_ALIAS_ALL references.
33536 * fold-const.c (fold_indirect_ref_1): Likewise.
33538 2017-03-28 Martin Liska <mliska@suse.cz>
33541 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
33542 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
33544 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
33545 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
33547 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
33548 (EXTRA_SPECS): Define.
33549 (SUBTARGET_EXTRA_SPECS): Likewise.
33550 (SUBTARGET_CPP_SPEC): Likewise.
33551 * config/arc/elf.h (EXTRA_SPECS): Renamed to
33552 SUBTARGET_EXTRA_SPECS.
33553 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
33555 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
33557 * config/arc/simdext.md (vst64_insn): Update pattern.
33558 (vld32wh_insn): Likewise.
33559 (vld32wl_insn): Likewise.
33560 (vld64_insn): Likewise.
33561 (vld32_insn): Likewise.
33563 2017-03-28 Marek Polacek <polacek@redhat.com>
33566 * fold-const.c (fold_comparison): Use protected_set_expr_location
33567 instead of SET_EXPR_LOCATION.
33569 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
33571 * tree.c (add_expr): Avoid name lookup warning.
33573 2017-03-27 Jeff Law <law@redhat.com>
33575 PR tree-optimization/80216
33576 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
33577 function name. Limit recursion depth.
33578 (record_temporary_equivalences): Corresponding changes.
33580 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
33582 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
33585 2017-03-27 Jakub Jelinek <jakub@redhat.com>
33588 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
33590 * cfgcleanup.c (reg_note_cfa_p): New array.
33591 (insns_have_identical_cfa_notes): New function.
33592 (old_insns_match_p): Don't cross-jump in between /f
33593 and non-/f instructions. If both i1 and i2 are frame related,
33594 verify all CFA notes, their order and content.
33596 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
33599 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
33600 HImode and SImode with zero extend to DImode to one insn.
33601 (bswap<mode>2_extenddi): Likewise.
33602 (bswapsi2_extenddi): Likewise.
33603 (bswaphi2_extendsi): Likewise.
33604 (bswaphi2): Combine bswap HImode and SImode into one insn.
33605 Separate memory insns from swapping register.
33606 (bswapsi2): Likewise.
33607 (bswap<mode>2): Likewise.
33608 (bswaphi2_internal): Delete, no longer used.
33609 (bswapsi2_internal): Likewise.
33610 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
33611 store, and gpr<-gpr swap insns.
33612 (bswap<mode>2_store): Likewise.
33613 (bswaphi2_reg): Register only splitter, combine with the splitter.
33614 (bswaphi2 splitter): Likewise.
33615 (bswapsi2_reg): Likewise.
33616 (bswapsi2 splitter): Likewise.
33617 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
33618 the insns into load, store, and register/register insns.
33619 (bswapdi2_ldbrx): Likewise.
33620 (bswapdi2_load): Likewise.
33621 (bswapdi2_store): Likewise.
33622 (bswapdi2_reg): Likewise.
33624 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
33626 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
33627 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
33629 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
33632 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
33634 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
33635 special handling for target option conflicts between dform
33636 options (-mpower9-dform, -mpower9-dform-vector,
33637 -mpower9-dform-scalar) and -mno-direct-move.
33639 2017-03-27 Richard Biener <rguenther@suse.de>
33641 PR tree-optimization/80181
33642 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
33644 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
33646 * config/arc/predicates.md (move_double_src_operand): Replace the
33647 call to move_double_src_operand with a call to address_operand.
33649 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
33651 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
33652 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
33653 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
33655 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
33657 * config/arc/predicates.md (long_immediate_loadstore_operand):
33658 Consider scaled addresses cases.
33660 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
33662 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
33663 restored when in interrupt.
33664 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
33665 doesn't have delay slot.
33667 2017-03-27 Richard Biener <rguenther@suse.de>
33670 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
33671 inlined thunk clones.
33673 2017-03-27 Jakub Jelinek <jakub@redhat.com>
33676 * asan.c (instrument_derefs): Copy over last operand from
33677 original COMPONENT_REF to the new COMPONENT_REF with
33678 DECL_BIT_FIELD_REPRESENTATIVE.
33679 * ubsan.c (instrument_object_size): Likewise.
33681 2017-03-27 Richard Biener <rguenther@suse.de>
33683 PR tree-optimization/80170
33684 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
33685 sure DR/SCEV didnt fold in constants we do not see when looking
33686 at the reference base alignment.
33688 2017-03-27 Richard Biener <rguenther@suse.de>
33690 PR middle-end/80171
33691 * gimple-fold.c (fold_ctor_reference): Properly guard against
33692 NULL return value from canonicalize_constructor_val.
33694 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
33697 * config/i386/i386.c (ix86_expand_builtin)
33698 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
33699 flags reg setting and flags reg using instructions.
33700 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
33701 clobbering instructions to zero extend op2.
33703 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
33705 * doc/install.texi (Configuration) <--with-aix-soname>:
33706 Update link to AIX ld.
33708 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
33710 PR rtl-optimization/80160
33711 PR rtl-optimization/80159
33712 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
33713 reg_alternate_class into account.
33715 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
33718 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
33719 to consider in curr_insn_transform.
33721 2017-03-24 Jakub Jelinek <jakub@redhat.com>
33723 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
33724 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
33725 and emit_mode_inner.
33727 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33729 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
33730 argument to the overloaded builtin variants. Use the new flag to
33731 deprecate certain builtin variants.
33732 * config/s390/s390-builtin-types.def: Add new builtin types.
33733 * config/s390/s390-builtins.h: Support new flags field for
33734 overloaded builtins.
33735 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
33736 (s390_macro_to_expand): Enable vector float data type.
33737 (s390_cpu_cpp_builtins_internal): Indicate support of the new
33738 builtins by incrementing the __VEC__ version number.
33739 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
33741 (s390_resolve_overloaded_builtin): Emit error messages depending
33742 on the builtin flags.
33743 * config/s390/s390.c (s390_expand_builtin): Support additional
33744 flags argument. Change error message to match the messages
33745 emitted in s390-c.c.
33746 * config/s390/s390.md: New UNSPEC_* constants.
33747 (op_type): Add new instruction types.
33748 * config/s390/vecintrin.h: Add new builtins and test data class
33750 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
33751 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
33752 (VEC_INEXACT, VEC_NOINEXACT): New constants.
33753 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
33754 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
33755 ("vec_mergel<mode>"): V_HW -> VEC_HW.
33757 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
33758 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
33759 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
33760 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
33762 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
33763 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
33764 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
33765 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
33767 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
33768 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
33769 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
33770 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
33771 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
33772 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
33773 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
33775 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
33776 ("vec_scatter_element<V_HW_4:mode>_DI")
33777 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
33778 ("vec_fpint<mode>", "vflls")
33779 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
33780 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
33781 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
33782 ("*vec_cmphe<mode>_cc"): ... these.
33784 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
33785 mode constant instead of magic value.
33787 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33789 * config/s390/s390.c (s390_expand_vec_compare): Support other
33790 vector floating point modes than just V2DF.
33791 (s390_expand_vcond): Likewise.
33792 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
33793 (s390_cannot_change_mode_class): Prevent mode changes between TF
33794 and V1TF in vector registers.
33795 * config/s390/s390.md (DF, SF): New mode attributes.
33796 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
33797 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
33798 SFmode support for VRs.
33799 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
33801 (VFT, VF_HW): New mode iterators.
33802 (vw, sdx): New mode attributes.
33803 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
33804 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
33805 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
33806 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
33807 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
33808 also the new vector floating point modes. Renaming to ...
33810 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
33811 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
33812 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
33813 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
33814 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
33815 ("vec_unordered<mode>"): ... these.
33817 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
33818 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
33819 ("*vec_extendv2df"): New insn definitions.
33821 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33823 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
33824 ("mulditi3_2", "*muldi3_sign"): New patterns.
33825 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
33826 rename the pattern definition.
33828 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33830 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
33832 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
33834 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33836 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
33837 instruction if possible.
33838 * config/s390/vector.md (vec_halfnumelts): New mode
33840 ("*vec_vllezlf<mode>"): New pattern.
33842 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33844 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
33845 ("popcountv4si2", "popcountv2di2"): Rename to ...
33846 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
33847 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
33849 ("popcount<mode>2_vxe"): New pattern.
33851 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33853 * common/config/s390/s390-common.c (processor_flags_table): Add
33855 * config.gcc: Add arch12.
33856 * config/s390/driver-native.c (s390_host_detect_local_cpu):
33857 Default to arch12 for unknown CPU model numbers.
33858 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
33859 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
33860 PROCESSOR_max sanity check.
33861 * config/s390/s390-opts.h (enum processor_type): Add
33863 * config/s390/s390.c (processor_table): Add arch12.
33864 (s390_expand_builtin): Add check for B_VXE flag.
33865 (s390_issue_rate): Add PROCESSOR_ARCH12.
33866 (s390_get_sched_attrmask): Likewise.
33867 (s390_get_unit_mask): Likewise.
33868 (s390_sched_score): Enable z13 scheduling for arch12.
33869 (s390_sched_reorder): Likewise.
33870 (s390_sched_variable_issue): Likewise.
33871 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
33873 (s390_tune_attr): Use z13 scheduling also for arch12.
33874 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
33875 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
33876 (TARGET_VXE_P): New macros.
33877 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
33878 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
33879 * config/s390/s390.opt: Add arch12 as processor_type.
33881 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33883 * config/s390/s390.md
33884 ("fixuns_truncdddi2", "fixuns_trunctddi2")
33885 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
33886 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
33888 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
33889 Rename expanders to ...
33891 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
33892 ("fixuns_truncdddi2_emu"): ... these.
33894 ("fixuns_trunc<mode>si2_emu"): New expander.
33896 ("*fixuns_truncdfdi2_z13"): Rename to ...
33897 ("*fixuns_truncdfdi2_vx"): ... this.
33899 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33901 * config/s390/2964.md: Remove the single element vector compare
33902 instructions which are no longer used.
33903 * config/s390/s390.c (s390_select_ccmode): Remove handling of
33905 (s390_canonicalize_comparison): Remove handling of DFmode
33907 (s390_expand_vec_compare_scalar): Remove function.
33908 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
33909 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
33911 ("*cmp<mode>_ccs"): Add wfcdb instruction.
33913 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33915 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
33917 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
33918 will anyway by matched by mov<mode>_64dfp.
33920 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33922 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
33923 vlef/vstef. Add missing operand to vleif.
33925 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33927 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
33928 pair for all vector types with 64 bit elements.
33929 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
33930 * config/s390/vector.md (V_HW_64): ... here.
33931 (V_128_NOSINGLE): New mode iterator.
33932 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
33933 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
33934 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
33935 ("*vec_load_pairv2di"): Change to ...
33936 ("*vec_load_pair<mode>"): ... this one.
33938 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33940 * config/s390/constraints.md: Add comments.
33941 (jKK): Reject element sizes > 8 bytes.
33942 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
33944 * config/s390/s390.md: Add the s_operand checks formerly in
33945 s390_split_ok_p to various splitters where they are still
33947 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
33948 for 128 bit vectors. Plus two splitters.
33950 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33952 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
33955 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33958 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
33959 error if the boundary argument is not constant.
33961 2017-03-24 Jakub Jelinek <jakub@redhat.com>
33963 PR rtl-optimization/80112
33964 * loop-doloop.c (doloop_condition_get): Don't check condition
33965 if cmp isn't SET with IF_THEN_ELSE src.
33967 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33969 PR tree-optimization/80158
33970 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
33971 replacing a candidate statement, also replace it for the
33972 candidate's alternate interpretation.
33973 (replace_rhs_if_not_dup): Likewise.
33974 (replace_one_candidate): Likewise.
33976 2017-03-24 Richard Biener <rguenther@suse.de>
33978 PR tree-optimization/80167
33979 * graphite-isl-ast-to-gimple.c
33980 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
33982 (translate_isl_ast_to_gimple::get_rename): Likewise.
33984 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
33986 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
33987 handling of certain combinations of target options, including the
33988 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
33989 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
33991 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33994 * config/arm/arm.md (*load_multiple): Add reload_completed to
33995 matching condition.
33997 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33998 Richard Biener <rguenth@suse.de>
34000 PR tree-optimization/79908
34001 PR tree-optimization/80136
34002 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
34003 been cast away, gimplify_and_add suffices.
34005 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
34007 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
34009 2017-03-23 Richard Biener <rguenther@suse.de>
34011 PR tree-optimization/80032
34012 * gimplify.c (gimple_push_cleanup): Forced unconditional
34013 cleanups still have to go to the conditional_cleanups
34016 2017-03-22 Jakub Jelinek <jakub@redhat.com>
34018 PR tree-optimization/80072
34019 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
34021 (next_operand_entry_id): Change type to unsigned int.
34022 (sort_by_operand_rank): Make sure to return the right return value
34023 even if unsigned fields are bigger than INT_MAX.
34024 (struct oecount): Change cnt and id type to unsigned int.
34025 (oecount_hasher::equal): Formatting fix.
34026 (oecount_cmp): Make sure to return the right return value
34027 even if unsigned fields are bigger than INT_MAX.
34028 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
34031 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
34032 TREE_READONLY on result if writing it more than once.
34035 * doc/invoke.texi (-fsanitize=thread): Document that with
34036 -fnon-call-exceptions atomics are not able to throw
34040 * tsan.c: Include tree-eh.h.
34041 (instrument_builtin_call): Call maybe_clean_eh_stmt or
34042 maybe_clean_or_replace_eh_stmt where needed.
34043 (instrument_memory_accesses): Add cfg_changed argument.
34044 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
34045 if it returned true.
34046 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
34048 PR rtl-optimization/63191
34049 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
34050 wrapper function, moved the whole old content into ...
34051 (ix86_delegitimize_address_1): ... this. New inline function.
34052 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
34053 true as last argument instead of ix86_delegitimize_address.
34055 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
34057 * config/aarch64/aarch64.c (generic_branch_cost): Copy
34058 cortexa57_branch_cost.
34060 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
34062 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
34064 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34067 * doc/md.texi (Constraints): Document wA constraint.
34068 * config/rs6000/constraints.md (wA): New.
34069 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
34070 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
34071 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
34072 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
34074 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
34077 * gimplify.c (is_oacc_declared): New function.
34078 (oacc_default_clause): Use it to set default flags for acc declared
34079 variables inside parallel regions.
34080 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
34081 declared variables.
34082 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
34083 declare attribute to any decl as necessary.
34085 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
34088 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
34089 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
34090 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
34091 (arm_arch_lpae): This.
34092 * config/arm/arm.c (arm_arch7ve): Rename into ...
34093 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
34094 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
34097 2017-03-22 Martin Liska <mliska@suse.cz>
34100 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
34101 error message instead of an ICE.
34103 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34105 * doc/extend.texi (6.11 Additional Floating Types): Revise.
34107 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
34109 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
34111 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
34114 2017-03-21 Martin Sebor <msebor@redhat.com>
34116 * doc/extend.texi: Use "cannot" instead of "can't."
34117 * doc/hostconfig.texi: Same.
34118 * doc/install.texi: Same.
34119 * doc/invoke.texi: Same.
34120 * doc/loop.texi: Same.
34121 * doc/md.texi: Same.
34122 * doc/objc.texi: Same.
34123 * doc/rtl.texi: Same.
34124 * doc/tm.texi: Same.
34125 * doc/tm.texi.in: Same.
34126 * doc/trouble.texi: Same.
34128 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
34131 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
34132 (collect_checksum_attributes): Set it.
34133 (die_checksum_ordered): Use it.
34135 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34137 PR tree-optimization/79908
34138 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
34139 change: For a VA_ARG whose LHS has been cast away, use
34140 force_gimple_operand to construct the side effects.
34142 2017-03-21 David Malcolm <dmalcolm@redhat.com>
34144 PR translation/80001
34145 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
34146 more amenable to translation.
34147 (oacc_loop_auto_partitions): Likewise.
34149 2017-03-21 Marek Polacek <polacek@redhat.com>
34150 Martin Sebor <msebor@redhat.com>
34152 PR tree-optimization/80109
34153 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
34154 on INTEGRAL_TYPE_P.
34156 2017-03-21 Jakub Jelinek <jakub@redhat.com>
34157 Segher Boessenkool <segher@kernel.crashing.org>
34160 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
34161 check reg_used_between_p between insn and one of succ or succ2
34162 depending on if succ is artificial insn not inserted into insn
34165 2017-03-21 Martin Liska <mliska@suse.cz>
34167 PR gcov-profile/80081
34168 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
34169 * doc/gcc.texi: Include gcov-dump stuff.
34170 * doc/gcov-dump.texi: New file.
34172 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
34174 PR rtl-optimization/79150
34175 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
34176 conditional jump, if the jump is the last insn of the loop.
34178 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34179 Richard Biener <rguenth@suse.de>
34181 PR tree-optimization/79908
34182 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
34183 been cast away, use force_gimple_operand to construct the side
34186 2017-03-21 Martin Liska <mliska@suse.cz>
34188 PR libfortran/79956
34189 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
34192 2017-03-21 Brad Spengler <spender@grsecurity.net>
34195 * plugin.c (htab_hash_plugin): New function.
34196 (add_new_plugin): Use it and adjust.
34197 (parse_plugin_arg_opt): Adjust.
34198 (init_one_plugin): Likewise.
34200 2017-03-21 Richard Biener <rguenther@suse.de>
34202 PR tree-optimization/80032
34203 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
34204 if set force the cleanup to happen unconditionally.
34205 (gimplify_target_expr): Push inserted clobbers with force_uncond
34206 to avoid them being removed by control-dependent DCE.
34208 2017-03-21 Richard Biener <rguenther@suse.de>
34210 PR tree-optimization/80122
34211 * tree-inline.c (copy_bb): Do not expans va-arg packs or
34212 va_arg_pack_len when the inlined call stmt requires pack
34214 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
34216 2017-03-21 Jakub Jelinek <jakub@redhat.com>
34219 * tsan.c (instrument_builtin_call): If the memory model argument
34220 is not a constant, assume it is valid.
34223 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
34226 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
34228 PR rtl-optimization/79910
34229 * combine.c (can_combine_p): Do not allow combining an I0 or I1
34230 if its dest is used by an insn before I2 (other than the combined
34231 insns themselves, which are properly handled already).
34233 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
34236 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
34238 * combine.c (record_used_regs): New static function.
34239 (try_combine): Handle situations where there is an additional
34240 instruction between I2 and I3 which needs to have a LOG_LINK
34244 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
34246 * combine.c (try_combine): Delete redundant i1 test. Call
34247 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
34249 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34252 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
34253 alternatives 13/14.
34255 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34257 PR tree-optimization/80054
34258 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
34259 the optimization if a PHI or any of its arguments is not dominated
34260 by the candidate's basis. Use gphi* rather than gimple* as
34262 (replace_profitable_candidates): Clean up a gimple* variable that
34263 should be a gphi* variable.
34265 2017-03-20 Martin Sebor <msebor@redhat.com>
34268 * doc/extend.texi (attribute constructor): Document present limitation.
34270 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
34273 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
34274 __POWER9_VECTOR__ #ifdef control, change template definition to
34275 use Power9-specific built-in function.
34276 (vec_any_eq): Likewise.
34277 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
34278 to control outcomes from this test.
34279 (vector_ae_<mode>p): For VEC_F modes, likewise.
34281 2017-03-20 Ian Lance Taylor <iant@google.com>
34283 * config/i386/i386.c (ix86_function_regparm): Save an extra
34284 register for -fsplit-stack with DECL_STATIC_CHAIN.
34286 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
34289 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
34290 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
34292 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
34294 * config/riscv/riscv.c (riscv_print_operand): Use "fence
34296 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
34299 2017-03-20 Marek Polacek <polacek@redhat.com>
34302 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
34304 2017-03-20 Richard Biener <rguenther@suse.de>
34306 PR tree-optimization/80113
34307 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
34308 allocate extra SSA name for PHI def.
34309 (add_close_phis_to_outer_loops): Likewise.
34310 (add_close_phis_to_merge_points): Likewise.
34311 (copy_loop_close_phi_args): Likewise.
34312 (copy_cond_phi_nodes): Likewise.
34314 2017-03-20 Martin Liska <mliska@suse.cz>
34316 PR middle-end/79753
34317 * tree-chkp.c (chkp_build_returned_bound): Do not build
34318 returned bounds for a LHS that's not a BOUNDED_P type.
34320 2017-03-20 Martin Liska <mliska@suse.cz>
34324 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
34325 COMPLEX_CST and VECTOR_CST.
34327 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34330 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
34331 target operand. A new splitter adds the clobber statement in case
34332 the target operand is dead anyway.
34334 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
34336 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
34337 to age-old versions of binutils and glibc.
34339 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
34341 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
34343 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
34345 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
34347 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
34349 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
34350 requirement for binutils 2.13.
34352 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
34354 * combine.c (try_combine): Delete redundant i1 test. Call
34355 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
34357 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
34359 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
34360 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
34362 <riscv64-*-elf>: Re-arrange section
34363 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
34364 <riscv32-*-linux>: Likewise.
34365 <riscv64-*-elf>: Likewise
34366 <riscv64-*-linux>: Likewise.
34368 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
34371 * aarch64.opt(verbose-cost-dump): Fix typo.
34373 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
34376 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
34377 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
34379 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
34381 * reload.c (find_reloads): When reloading a nonoffsettable address,
34382 use RELOAD_OTHER for it and its address reloads.
34384 PR rtl-optimization/79910
34385 * combine.c (record_used_regs): New static function.
34386 (try_combine): Handle situations where there is an additional
34387 instruction between I2 and I3 which needs to have a LOG_LINK
34390 2017-03-17 Jeff Law <law@redhat.com>
34392 PR tree-optimization/71437
34393 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
34394 conditional in the hash table first.
34395 (vrp_dom_walker::before_dom_children): Extract condition from
34396 ASSERT_EXPR. Record condition, its inverion and any implied
34397 conditions as well.
34399 2017-03-17 Marek Polacek <polacek@redhat.com>
34400 Markus Trippelsdorf <markus@trippelsdorf.de>
34402 PR tree-optimization/80079
34403 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
34406 2017-03-17 Richard Biener <rguenther@suse.de>
34408 PR middle-end/80075
34409 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
34410 Properly verify the LHS before the RHS possibly claims to be
34412 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
34415 2017-03-17 Martin Jambor <mjambor@suse.cz>
34417 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
34418 (List of -O2 options): Likewise.
34419 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
34422 2017-03-17 Tom de Vries <tom@codesourcery.com>
34424 * gcov-dump.c (print_usage): Print bug_report_url.
34426 2017-03-17 Richard Biener <rguenther@suse.de>
34428 PR middle-end/80050
34429 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
34430 (parser::peek): Likewise.
34432 2017-03-17 Richard Biener <rguenther@suse.de>
34434 PR tree-optimization/80048
34435 * sese.c (free_sese_info): Properly release rename_map and
34436 copied_bb_map elements.
34438 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
34440 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
34441 Add linked-list forward and backlinks. Insert on
34442 construction, remove on destruction.
34443 (class pass_store_merging): Add m_stores_head field.
34444 (pass_store_merging::terminate_and_process_all_chains):
34445 Iterate over m_stores_head list.
34446 (pass_store_merging::terminate_all_aliasing_chains):
34448 (pass_store_merging::execute): Check for debug stmts first.
34449 Push new chains onto the m_stores_head stack.
34451 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
34454 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
34455 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
34456 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
34458 2017-03-16 Jeff Law <law@redhat.com>
34460 PR tree-optimization/71437
34461 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
34462 member function. Implementation moved into after_dom_children
34463 member function and into the threader's thread_outgoing_edges
34465 (dom_opt_dom_walker::after_dom_children): Simplify by moving
34466 some code into new thread_outgoing_edges.
34467 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
34468 definition. Simplify marker handling (do it here). Assume we always
34469 have the available expression and the const/copies tables.
34470 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
34472 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
34473 * tree-vrp.c (equiv_stack): No longer file scoped.
34474 (vrp_dom_walker): New class.
34475 (vrp_dom_walker::before_dom_children): New member function.
34476 (vrp_dom_walker::after_dom_children): Likewise.
34477 (identify_jump_threads): Setup domwalker. Use it rather than
34478 walking edges in a random order by hand. Simplify setup/finalization.
34479 (finalize_jump_threads): Remove.
34480 (vrp_finalize): Do not call identify_jump_threads here.
34481 (execute_vrp): Do it here instead and call thread_through_all_blocks
34484 PR tree-optimization/71437
34485 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
34487 (simplify_stmt_for_jump_threading): Add basic_block argument. All
34489 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
34490 (dom_opt_dom_walker::thread_across_edge): Remove
34491 handle_dominating_asserts argument. All callers changed.
34492 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
34493 changes. Remove calls to lhs_of_dominating_assert. Other
34494 uses of handle_dominating_asserts turn into unconditional code
34495 (simplify_control_stmt_condition_1): Likewise.
34496 (simplify_control_stmt_condition): Likewise.
34497 (thread_through_normal_block, thread_across_edge): Likewise.
34498 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
34499 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
34500 object if it is not an SSA_NAME.
34501 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
34502 before calling into the VRP specific simplifiers.
34503 (identify_jump_threads): Remove handle_dominating_asserts
34506 2017-03-16 Jakub Jelinek <jakub@redhat.com>
34509 * tree-diagnostic.c (default_tree_printer): No longer static.
34510 * tree-diagnostic.h (default_tree_printer): New prototype.
34512 2017-03-16 Tamar Christina <tamar.christina@arm.com>
34514 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
34515 Change ins into fmov.
34517 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34519 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
34520 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
34521 Use h_con constraint for operand 1.
34522 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
34523 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
34525 2017-03-15 Jeff Law <law@redhat.com>
34527 PR tree-optimization/71437
34528 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
34529 (record_temporary_equivalences): Use it.
34531 PR tree-optimization/71437
34532 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
34533 tree-ssa-scopedtables.
34534 (lookup_avail_expr, build_and_record_new_cond): Likewise.
34535 (record_conditions, record_cond, vuse_eq): Likewise.
34536 (record_edge_info): Adjust to API tweak of record_conditions.
34537 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
34538 (record_temporary_equivalences, optimize_stmt): Likewise.
34539 (eliminate_redundant_computations): Likewise.
34540 (record_equivalences_from_stmt): Likewise.
34541 * tree-ssa-scopedtables.c: Include options.h and params.h.
34542 (vuse_eq): New function, moved from tree-ssa-dom.c
34543 (build_and_record_new_cond): Likewise.
34544 (record_conditions): Likewise. Accept vector of conditions rather
34545 than edge_equivalence structure for first argument.
34546 for the first argument.
34547 (avail_exprs_stack::lookup_avail_expr): New member function, moved
34548 from tree-ssa-dom.c.
34549 (avail_exprs_stack::record_cond): Likewise.
34550 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
34551 from tree-ssa-dom.c.
34552 (avail_exprs_stack): Add new member functions lookup_avail_expr
34554 (record_conditions): Declare.
34556 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
34559 * lra-constraints.c (process_alt_operands): Increase reject for
34560 reloading an input/output operand.
34562 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
34565 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
34566 insns to convert from signed/unsigned char/short to IEEE 128-bit
34568 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
34570 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
34573 * config/i386/i386.c (ix86_vector_duplicate_value): Create
34574 subreg of inner mode for values already in registers.
34576 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
34578 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
34579 iteration reg is used after the loop.
34581 2017-03-14 Martin Sebor <msebor@redhat.com>
34583 PR tree-optimization/79800
34584 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
34585 precision in negative-positive range.
34586 (format_floating): Call non-const overload with adjusted precision.
34588 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
34591 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
34594 2017-03-14 Martin Sebor <msebor@redhat.com>
34596 PR middle-end/80020
34597 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
34598 * builtins.def (aligned_alloc): Use it.
34601 * Makefile.in (GTFILES): Add calls.c.
34602 * calls.c: Include "gt-calls.h".
34604 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
34606 PR rtl-optimization/79728
34607 * regs.h (struct target_regs): New field
34608 x_contains_allocatable_regs_of_mode.
34609 (contains_allocatable_regs_of_mode): New macro.
34610 * reginfo.c (init_reg_sets_1): Initialize it, and change
34611 contains_reg_of_mode so it includes global regs as well.
34612 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
34613 rather than contains_regs_of_mode.
34615 2017-03-14 Martin Liska <mliska@suse.cz>
34617 * doc/invoke.texi: Document options that can't be combined with
34618 -fcheck-pointer-bounds.
34620 2017-03-14 Martin Liska <mliska@suse.cz>
34622 PR middle-end/79831
34623 * doc/invoke.texi (-Wchkp): Document the option.
34625 2017-03-14 Martin Liska <mliska@suse.cz>
34627 * Makefile.in: Install gcov-dump.
34629 2017-03-14 Martin Liska <mliska@suse.cz>
34631 * multiple_target.c (expand_target_clones): Bail out for
34632 an invalid attribute.
34634 2017-03-14 Richard Biener <rguenther@suse.de>
34636 * alias.c (struct alias_set_entry): Pack properly.
34637 * cfgloop.h (struct loop): Likewise.
34638 * cse.c (struct set): Likewise.
34639 * ipa-utils.c (struct searchc_env): Likewise.
34640 * loop-invariant.c (struct invariant): Likewise.
34641 * lra-remat.c (struct cand): Likewise.
34642 * recog.c (struct change_t): Likewise.
34643 * rtl.h (struct address_info): Likewise.
34644 * symbol-summary.h (function_summary): Likewise.
34645 * tree-loop-distribution.c (struct partition): Likewise.
34646 * tree-object-size.c (struct object_size_info): Likewise.
34647 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
34648 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
34649 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
34650 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
34651 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
34652 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
34653 (struct _stmt_vec_info): Likewise.
34655 2017-03-14 Martin Liska <mliska@suse.cz>
34658 * multiple_target.c (create_dispatcher_calls): Check that
34659 a target can create a function dispatcher.
34661 2017-03-14 Martin Liska <mliska@suse.cz>
34664 * multiple_target.c (expand_target_clones): Drop local.local
34665 flag for default implementation.
34667 2017-03-14 Richard Biener <rguenther@suse.de>
34669 PR tree-optimization/80030
34670 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
34672 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
34674 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
34675 gcc_fallthrough() instead of __attribute__((fallthrough));
34677 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
34679 * doc/gcc.texi: Remove "up" link to (DIR).
34680 * doc/gccint.texi: Ditto.
34682 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
34684 * doc/install.texi (Specific) <avr>: Remove reference to
34687 2017-03-13 Jeff Law <law@redhat.com>
34689 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
34690 attribute rather than comments.
34692 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
34693 match_scratch operand is highest.
34695 2017-03-13 Martin Liska <mliska@suse.cz>
34697 PR middle-end/78339
34698 * ipa-pure-const.c (warn_function_noreturn): If the declarations
34699 is a CHKP clone, use original declaration.
34701 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
34703 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
34704 (arc_conditional_register_usage): Use a different allocation order
34705 when optimizing for size.
34706 * common/config/arc/arc-common.c (arc_option_optimization_table):
34707 Section anchors default on when optimizing for size.
34709 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
34711 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
34713 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
34715 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
34716 * config/arc/arc.md (cpu_facility): Add cd variant.
34717 (*movqi_insn): Add code density variant.
34718 (*movhi_insn): Likewise.
34719 (*movqi_insn): Likewise.
34720 (*addsi3_mixed): Likewise.
34721 (subsi3_insn): Likewise.
34723 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
34725 * config/arc/arc.md (movsi_cond_exec): Update constraint.
34727 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
34729 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
34730 expressions with MINUS and UNARY ops.
34732 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34735 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
34737 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
34738 between vec_select and vector argument.
34739 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
34740 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
34741 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
34742 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
34743 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
34744 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
34746 2017-03-13 Richard Biener <rguenther@suse.de>
34749 * params.def (vect-max-peeling-for-alignment): Fix typo.
34751 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
34753 * doc/install.texi (Specific) <mips-*-*>: Remove description of
34754 issue that only occurred with binutils below 2.18.
34756 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
34758 * doc/install.texi (Specific) <cris-axis-elf>: No longer
34759 refer to binutils 2.11/2.12 minimum.
34761 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
34763 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
34764 ftp.kernel.org and simplify binutils requirement.
34766 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
34768 * doc/invoke.texi (Warning Options): Fix spelling of link-time
34770 (Optimize Options): Ditto. Also remove redundancy.
34772 2017-03-10 David Malcolm <dmalcolm@redhat.com>
34774 PR translation/79848
34775 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
34777 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
34778 to G_ to avoid double translation.
34780 2017-03-10 David Malcolm <dmalcolm@redhat.com>
34782 PR translation/79923
34783 * auto-profile.c (get_combined_location): Convert leading
34784 character of diagnostics to lower case and remove trailing period.
34785 (read_profile): Likewise for various diagnostics.
34786 * config/arm/arm.c (arm_option_override): Remove trailing period
34787 from various diagnostics.
34788 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
34789 (msp430_expand_delay_cycles): Likewise.
34791 2017-03-10 David Malcolm <dmalcolm@redhat.com>
34794 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
34795 full command-line argument, rather than just "str".
34796 (aarch64_validate_march): Likewise.
34797 (aarch64_validate_mtune): Likewise.
34799 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
34801 PR rtl-optimization/78911
34802 * lra-assigns.c (must_not_spill_p): New function.
34803 (spill_for): Use it.
34805 2017-03-10 Jakub Jelinek <jakub@redhat.com>
34807 PR tree-optimization/79981
34808 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
34809 ATOMIC_COMPARE_EXCHANGE ifn result.
34810 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
34811 IFN_ATOMIC_COMPARE_EXCHANGE.
34813 2017-03-10 David Malcolm <dmalcolm@redhat.com>
34816 * opts.c (parse_sanitizer_options): Add missing question mark to
34817 "did you mean" message.
34819 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34821 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
34823 (VMULEUH_UNS): Likewise.
34824 (VMULOUB_UNS): Likewise.
34825 (VMULOUH_UNS): Likewise.
34826 * config/rs6000/rs6000.c (builtin_function_type): Remove
34827 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
34829 2017-03-10 David Malcolm <dmalcolm@redhat.com>
34832 * read-rtl-function.c (function_reader::read_rtx_operand): Update
34833 x with result of extra_parsing_for_operand_code_0.
34834 (function_reader::extra_parsing_for_operand_code_0): Convert
34835 return type from void to rtx, returning x. When reading
34836 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
34837 larger size containing struct block_symbol.
34839 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
34841 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
34842 -mfloat128-hardware without -m64.
34844 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
34847 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
34848 entries to the case statement that marks unsigned arguments to
34849 overloaded functions.
34851 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
34853 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
34854 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
34856 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
34859 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
34860 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
34862 2017-03-10 Martin Liska <mliska@suse.cz>
34866 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
34867 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
34870 2017-03-10 Olivier Hainque <hainque@adacore.com>
34872 * tree-switch-conversion (array_value_type): Start by resetting
34873 candidate type to it's main variant.
34875 2017-03-10 Jakub Jelinek <jakub@redhat.com>
34877 PR rtl-optimization/79909
34878 * combine.c (try_combine): Use simplify_replace_rtx on individual
34879 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
34880 of the whole CALL_INSN_FUNCTION_USAGE.
34882 PR tree-optimization/79972
34883 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
34884 get_range_info on SSA_NAMEs. Formatting fixes.
34886 2017-03-10 Richard Biener <rguenther@suse.de>
34887 Jakub Jelinek <jakub@redhat.com>
34889 PR tree-optimization/77975
34890 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
34891 edge to be constant.
34892 (get_val_for): For constant x return it. Formatting fix.
34893 (loop_niter_by_eval): Avoid pointless looping if the next iteration
34894 would use the same bases as the current one.
34896 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34898 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
34899 instead of vec_select for V1TImode.
34900 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
34902 (VSX_LE_128): Add V1TI to this mode iterator.
34903 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
34904 (*vsx_le_perm_store_<mode>): Likewise.
34905 (pre-reload splitter for VSX stores): Likewise.
34906 (post-reload splitter for VSX stores): Likewise.
34907 (*vsx_xxpermdi2_le_<mode>): Likewise.
34908 (*vsx_lxvd2x2_le_<mode>): Likewise.
34909 (*vsx_stxvd2x2_le_<mode>): Likewise.
34911 2017-03-09 Michael Eager <eager@eagercon.com>
34913 Correct failures with --enable-checking=yes,rtl.
34915 * config/microblaze/microblaze.c (microblaze_expand_shift):
34916 Replace GET_CODE test with CONST_INT_P and INTVAL test with
34917 test for const0_rtx.
34918 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
34919 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
34921 2017-03-09 Richard Biener <rguenther@suse.de>
34923 PR tree-optimization/79977
34924 * graphite-scop-detection.c (scop_detection::merge_sese):
34925 Handle the case of extra exits to blocks dominating the entry.
34927 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
34929 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
34932 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
34934 PR rtl-optimization/79949
34935 * lra-constraints.c (process_alt_operands): Check memory when
34936 trying to predict a cycle. Print about the overall increase.
34938 2017-03-09 Richard Biener <rguenther@suse.de>
34940 PR middle-end/79971
34941 * gimple-expr.c (useless_type_conversion_p): Preserve
34942 TYPE_SATURATING for fixed-point types.
34944 2017-03-09 Richard Biener <rguenther@suse.de>
34947 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
34948 alignment of BLKmode params.
34950 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34953 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
34954 (VALL_NO_V2Q): Likewise.
34956 * config/aarch64/aarch64-simd.md
34957 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
34959 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
34960 VALL_NO_V2Q mode iterator.
34961 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
34963 2017-03-09 Martin Liska <mliska@suse.cz>
34965 PR tree-optimization/79631
34966 * tree-chkp-opt.c (chkp_is_constant_addr): Call
34967 tree_int_cst_sign_bit just for INTEGER constants.
34969 2017-03-09 Martin Liska <mliska@suse.cz>
34973 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
34976 2017-03-09 Marek Polacek <polacek@redhat.com>
34979 * tree.c (inchash::add_expr): Handle TREE_VEC.
34981 2017-03-09 Martin Liska <mliska@suse.cz>
34984 (chkp_narrow_size_and_offset): New function.
34985 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
34986 (void chkp_parse_bit_field_ref): New function.
34987 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
34988 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
34990 2017-03-09 Martin Liska <mliska@suse.cz>
34993 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
34994 (chkp_find_bounds_1): Remove gcc_unreachable.
34996 2017-03-09 Jakub Jelinek <jakub@redhat.com>
34999 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
35000 BUILT_IN_SYNC*, determine the access type from the size suffix and
35001 always build a MEM_REF with that type. Handle forgotten
35002 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
35005 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
35006 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
35007 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
35008 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
35009 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
35010 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
35011 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
35012 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
35013 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
35014 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
35015 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
35016 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
35017 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
35018 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
35019 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
35020 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
35021 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
35022 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
35023 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
35024 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
35025 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
35026 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
35027 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
35028 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
35029 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
35030 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
35031 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
35032 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
35033 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
35034 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
35035 definitions outside of __OPTIMIZE__ guarded section.
35038 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
35039 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
35040 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
35041 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
35044 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35046 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
35047 ("vfenez<mode>"): Add missing constraints.
35049 2017-03-08 Martin Sebor <msebor@redhat.com>
35052 * config/nds32/nds32.c (nds32_option_override):
35053 Fix misspelled diagnostic.
35055 2017-03-08 Jakub Jelinek <jakub@redhat.com>
35058 * gimplify.c (gimplify_omp_for): Replace index var in outer
35059 taskloop statement with an artificial variable and add
35060 OMP_CLAUSE_PRIVATE clause for it.
35062 2017-03-08 Richard Biener <rguenther@suse.de>
35064 PR tree-optimization/79955
35065 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
35066 for accesses that are completely outside of the variable.
35068 2017-03-08 Andrew Haley <aph@redhat.com>
35070 PR tree-optimization/79943
35071 * tree-ssa-loop-split.c (compute_new_first_bound): When
35072 calculating the new upper bound, (END-BEG) should be added, not
35075 2017-03-08 Jakub Jelinek <jakub@redhat.com>
35077 * config/avr/avr.md (setmemhi): Make sure match_dup
35078 operand number comes before match_scratch.
35080 2017-03-08 Richard Biener <rguenther@suse.de>
35082 PR tree-optimization/79920
35083 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
35084 with ncopies == 1 to ...
35085 (vect_transform_slp_perm_load): ... here. Properly compute
35086 all element loads by iterating VF times over the group. Do
35087 not handle ncopies (computed in a broken way) in
35088 vect_create_mask_and_perm.
35090 2017-03-08 Jakub Jelinek <jakub@redhat.com>
35093 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
35094 is a uniform vector, use uniform_vector_p return value instead of
35095 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
35097 2017-03-07 Marek Polacek <polacek@redhat.com>
35099 PR middle-end/79809
35100 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
35101 (alloca_call_type): Likewise.
35103 2017-03-07 Martin Liska <mliska@suse.cz>
35105 * gcov.c (process_args): Put comment to correct location.
35107 2017-03-07 Martin Liska <mliska@suse.cz>
35109 PR middle-end/68270
35110 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
35111 Use array_at_struct_end_p instead of DECL_CHAIN (field).
35112 (chkp_narrow_bounds_for_field): Likewise.
35113 (chkp_parse_array_and_component_ref): Pass one more argument to
35116 2017-03-07 Richard Biener <rguenther@suse.de>
35118 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
35121 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
35123 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
35124 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
35126 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35129 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
35130 to end of description.
35131 (PARAM_MAX_STORES_TO_MERGE): Likewise.
35133 2017-03-07 Jakub Jelinek <jakub@redhat.com>
35135 PR rtl-optimization/79901
35136 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
35138 (*avx512f_<code><mode>3<mask_name>): ... this.
35139 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
35140 iterator instead of VI8_AVX2_AVX512BW.
35142 PR rtl-optimization/79901
35143 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
35144 min/max expander, expand it using expand_vec_cond_expr.
35147 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
35150 2017-03-06 Jakub Jelinek <jakub@redhat.com>
35153 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
35154 to void * for PCH reasons.
35155 * dwarf2out.c (output_loc_operands, output_die): Cast
35156 v.val_vec.array to unsigned char *.
35158 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
35161 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
35164 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
35166 PR rtl-optimization/79571
35167 * lra-constraints.c (process_alt_operands): Calculate static
35168 reject and subtract it from overall when only addresses will be
35171 2017-03-06 Julia Koval <julia.koval@intel.com>
35174 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
35175 incoming stack boundary to 128 for 64-bit targets.
35177 2017-03-06 Richard Biener <rguenther@suse.de>
35179 PR tree-optimization/79894
35180 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
35181 to NULL after folding it.
35183 2017-03-06 Richard Biener <rguenther@suse.de>
35185 PR tree-optimization/79824
35186 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
35187 check disabling peeling for gaps.
35189 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
35191 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
35192 attributes): Document gettimeofday.
35194 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
35196 * config/s390/s390.c (s390_option_override_internal): Set
35197 PARAM_MIN_VECT_LOOP_BOUND
35199 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
35201 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
35202 * config/s390/s390.md: Likewise.
35204 2017-03-06 Jakub Jelinek <jakub@redhat.com>
35207 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
35208 (<avx2_avx512>_perm<mode>): Rename to ...
35209 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
35211 (<avx512>_perm<mode>_mask): Rename to ...
35212 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
35214 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
35215 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
35216 instead of VI8F_256_512.
35217 (avx512f_perm<mode>): New define_expand.
35218 (avx512f_perm<mode>_mask): Likewise.
35219 (avx512f_perm<mode>_1<mask_name>): New define_insn.
35220 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
35222 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
35224 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
35225 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
35227 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
35229 2017-03-06 Martin Liska <mliska@suse.cz>
35232 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
35233 when having a SSA NAME w/o VAR_DECL assigned to it.
35235 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
35237 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
35238 msa_dpsub_<su>_d): Fix MODE for vec_select.
35240 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
35242 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
35244 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
35246 2017-03-06 Richard Biener <rguenther@suse.de>
35248 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
35249 * plugin.c (register_plugin_info): Likewise.
35250 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
35252 2017-03-05 Jakub Jelinek <jakub@redhat.com>
35254 * config/i386/sse.md (sse_storehps, sse_storelps,
35255 avx_<castmode><avxsizesuffix>_<castmode>,
35256 avx512f_<castmode><avxsizesuffix>_<castmode>,
35257 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
35258 in condition that at least one operand is not a MEM.
35260 2017-03-03 Jakub Jelinek <jakub@redhat.com>
35262 PR middle-end/79805
35263 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
35264 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
35266 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
35267 gimple_call_nothrow_p flag based on whether original builtin can throw.
35268 If it can, emit following stmts on the fallthrough edge.
35269 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
35270 don't create new bb if inserting just debug stmts on the edge, try to
35271 insert them on the fallthru bb or just reset debug stmts.
35273 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
35276 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
35277 restore recog_data (including the operand rtxes inside it) around
35278 the call to get_insn_template.
35280 2017-03-03 Martin Sebor <msebor@redhat.com>
35282 PR tree-optimization/79699
35283 * context.c (context::~context): Free MPFR caches to avoid
35284 a memory leak on program exit.
35286 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35288 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
35289 Use wide_int::ulow () instead of .elt (0).
35291 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
35293 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
35294 (*pushxf): Limit oF constraint to 32bit targets and add oC
35295 constraint for 64bit targets.
35296 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
35297 (*pushdf): Change rmF constraint to rmC.
35299 2017-03-03 Martin Liska <mliska@suse.cz>
35301 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
35302 Remove unused variable.
35304 2017-03-03 Jakub Jelinek <jakub@redhat.com>
35307 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
35308 is a memory operand, increase num_memory.
35309 (ix86_expand_args_builtin): Likewise.
35311 2017-03-03 Jan Hubicka <jh@suse.cz>
35314 * ipa-devirt.c (maybe_record_node): Properly handle
35315 __cxa_pure_virtual visibility.
35317 2017-03-03 Martin Liska <mliska@suse.cz>
35319 PR tree-optimization/79803
35320 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
35322 (pass_loop_prefetch::execute): Disabled optimization if an
35323 assumption about L1 cache size is not met.
35325 2017-03-03 Martin Liska <mliska@suse.cz>
35327 PR rtl-optimization/79574
35328 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
35329 (hash_scan_set): Likewise.
35330 (dump_hash_table): Likewise.
35331 (hoist_code): Likewise.
35333 2017-03-03 Richard Biener <rguenther@suse.de>
35335 * fixed-value.c (fixed_from_string): Restore use of elt (1)
35336 in place of uhigh ().
35337 (fixed_convert_from_real): Likewise.
35339 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
35342 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
35344 2017-03-03 Richard Biener <rguenther@suse.de>
35346 PR middle-end/79818
35347 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
35348 TYPE_OVERFLOW_UNDEFINED check.
35350 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35352 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
35354 (vector_ae_<mode>_p): Likewise.
35355 (vector_nez_<mode>_p): Likewise.
35356 (vector_ne_v2di_p): Likewise.
35357 (vector_ae_v2di_p): Likewise.
35358 (vector_ne_<mode>_p): Likewise.
35359 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
35361 (vsx_tsqrt<mode>2_fe): Likewise.
35363 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
35366 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
35368 2017-03-02 Jakub Jelinek <jakub@redhat.com>
35370 PR rtl-optimization/79780
35371 * cprop.c (one_cprop_pass): When second and further conditional trap
35372 in a single basic block is turned into an unconditional trap, turn it
35373 into a deleted note to avoid RTL verification failures.
35375 2017-03-02 Richard Biener <rguenther@suse.de>
35377 * fold-const.c (const_binop): Use ulow () instead of elt (0).
35379 2017-03-02 Richard Biener <rguenther@suse.de>
35381 PR tree-optimization/79345
35383 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
35384 param and abort the walk, returning -1 if it is hit.
35385 (walk_aliased_vdefs): Take a limit param and pass it on.
35386 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
35387 defaulting to 0 and return a signed int.
35388 * tree-ssa-uninit.c (struct check_defs_data): New struct.
35389 (check_defs): New helper.
35390 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
35391 about uninitialized memory.
35392 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
35393 bogus uninitialized warning.
35394 (fixed_convert_from_real): Likewise.
35396 2017-03-02 Bin Cheng <bin.cheng@arm.com>
35398 PR tree-optimization/66768
35399 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
35400 iv_use if base object can't be determined.
35402 2017-03-02 Jakub Jelinek <jakub@redhat.com>
35404 PR tree-optimization/79345
35405 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
35406 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
35407 (get_pattern_stats): Initialize it.
35408 * genemit.c (gen_expand): Verify match_scratch numbers come after
35409 match_operand/match_dup numbers.
35410 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
35411 match_scratch numbers.
35412 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
35414 * config/s390/s390.md (trunctdsd2): Likewise.
35416 2017-03-02 Richard Biener <rguenther@suse.de>
35418 * wide-int.h (wide_int_storage::operator=): Implement in terms
35421 2017-03-02 Richard Biener <rguenther@suse.de>
35423 PR tree-optimization/79777
35424 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
35425 the to insert expression to sth existing.
35427 2017-03-01 Martin Sebor <msebor@redhat.com>
35429 PR middle-end/79692
35430 * gimple-ssa-sprintf.c
35431 (directive::known_width_and_precision): New function.
35432 (format_integer): Use it.
35433 (get_mpfr_format_length): Consider the full range of precision
35434 when computing %g output with the # flag. Set the likely byte
35435 count to 3 rather than 1 when precision is indeterminate.
35436 (format_floating): Correct the lower bound of precision.
35438 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35440 * doc/invoke.texi: Document default code model for 64-bit Linux.
35442 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35445 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
35446 udiv rather than div since input pattern is unsigned.
35448 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
35450 * config/i386/i386.c (print_reg): Warn for values of
35451 unsupported size in integer register.
35453 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
35456 * config/rs6000/predicates.md (current_file_function_operand): Do
35457 not allow self calls to be local if the function is replaceable.
35459 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
35462 * config/rs6000/altivec.h (vec_ctz and others): Change the
35463 preprocessor macro that controls conditional compilation from
35464 _ARCH_PWR9 to __POWER9_VECTOR__.
35465 (vec_all_ne): Change parameterization of __altivec_scalar_pred
35466 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
35467 control (instead of _ARCH_PWR9 control) so that template
35468 definition uses power9-specific function.
35469 (vec_any_eq): Likewise.
35470 (vec_all_ne): Change macro definition to use a power9-specific
35471 expansion under #ifdef __POWER9_VECTOR__ control (instead of
35472 _ARCH_PWR9 control).
35473 (vec_any_eq) Likewise.
35474 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
35475 expansion for CMPNEF to remove support for xvcmpnesp instruction.
35476 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
35477 support for xvcmpnedp instruction.
35478 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
35479 macro expansion so that Power9 implementation of vec_all_ne does
35480 not use the AltiVec predicate framework.
35481 (VCMPNEH_P): Likewise.
35482 (VCMPNEW_P): Likewise.
35483 (VCMPNED_P): Likewise.
35484 (VCMPNEFP_P): Likewise.
35485 (VCMPNEDP_P): Likewise.
35486 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
35487 implementation of vec_any_eq to not use AltiVec predicate
35489 (VCMPAEH_P): Likewise.
35490 (VCMPAEW_P): Likewise.
35491 (VCMPAED_P): Likewise.
35492 (VCMPAEFP_P): Likewise.
35493 (VCMPAEDP_P): Likewise.
35494 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
35495 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
35496 not use the AltiVec predicate framework.
35497 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
35498 of vec_any_eq to not use AltiVec predicate framework.
35499 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
35500 support for predefined __POWER9_VECTOR__ macro to indicate that
35501 Power9 instruction selection is enabled.
35502 (altivec_overloaded_builtins): Remove extraneous
35503 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
35504 function argument types RS6000_BTI_bool_V16QI and
35505 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
35506 entry for overloaded function argument types RS6000_BTI_bool_V4SI
35507 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
35508 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
35509 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
35510 Power9 for implementations of vec_cmpne. Change the signature for
35511 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
35512 (representing vec_all_ne) to remove the previously described first
35513 argument of type RS6000_BTI_INTSI, as this was an artifact of
35514 reliance on the AltiVec predicate framework, which is no longer
35515 used in the implementation of these functions. Add
35516 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
35517 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
35518 since, unlike the AltiVec predicate framework implementation, we
35519 do not share function descriptors between vec_alle and vec_anyeq.
35520 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
35521 set of modes that receive special treatment even when
35522 TARGET_P9_VECTOR is true. The special treatment emits code that
35523 does not depend on Power9 instructions.
35524 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
35525 define_expand to not rely on AltiVec predicate framework.
35526 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
35528 (vector_ne_v2di_p): Change this define_expand to not rely on
35529 AltiVec predicate framework.
35530 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
35532 (vector_ne_<mode>_p): Change this define_expand to not rely on
35533 AltiVec predicate framework.
35534 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
35536 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
35537 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
35538 define_insn pattern.
35539 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
35540 define_insn pattern because the xvcmpne<VSs>. instruction is not
35542 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
35543 instruction is not supported.
35545 2017-03-01 Jakub Jelinek <jakub@redhat.com>
35547 * config/nvptx/nvptx.c: Include intl.h.
35549 2017-03-01 Martin Jambor <mjambor@suse.cz>
35552 * ipa-prop.h (ipa_bits): Removed field known.
35553 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
35554 to pointers. Adjusted their comments to warn about their sharing.
35555 (ipcp_transformation_summary): Change bits to a vector of pointers.
35556 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
35557 (ipa_get_ipa_bits_for_value): Declare.
35558 * tree-vrp.h (value_range): Mark as GTY((for_user)).
35559 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
35560 (ipa_bits_hash_table): Likewise.
35561 (ipa_vr_ggc_hash_traits): Likewise.
35562 (ipa_vr_hash_table): Likewise.
35563 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
35564 being pointers and vr_known being removed.
35565 (ipa_set_jf_unknown): Likewise.
35566 (ipa_get_ipa_bits_for_value): New function.
35567 (ipa_set_jfunc_bits): Likewise.
35568 (ipa_get_value_range): New overloaded functions.
35569 (ipa_set_jfunc_vr): Likewise.
35570 (ipa_compute_jump_functions_for_edge): Use the above functions to
35571 construct bits and vr parts of jump functions.
35572 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
35573 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
35575 (ipcp_grow_transformations_if_necessary): Also allocate
35576 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
35578 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
35579 them. Fix too long lines.
35580 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
35581 vr_known being removed.
35582 (ipa_read_jump_function): Use new setter functions to construct bits
35583 and vr parts of jump functions or set them to NULL.
35584 (write_ipcp_transformation_info): Adjust for bits being pointers.
35585 (read_ipcp_transformation_info): Likewise.
35586 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
35588 Include gt-ipa-prop.h.
35589 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
35591 (ipcp_store_bits_results): Likewise.
35592 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
35593 Do not write to existing jump functions but use a temporary instead.
35595 2017-03-01 Jakub Jelinek <jakub@redhat.com>
35598 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
35599 attempt to use its first operand as BIT_FIELD_REF base.
35601 2017-03-01 Richard Biener <rguenther@suse.de>
35603 PR middle-end/79721
35604 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
35605 interpolating formula in wrapping arithmetic.
35606 (chrec_apply): Convert chrec_evaluate return value to wanted type.
35608 2017-03-01 Jakub Jelinek <jakub@redhat.com>
35610 PR tree-optimization/79734
35611 * tree-vect-generic.c (expand_vector_condition): Optimize
35612 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
35613 Handle VEC_COND_EXPR where comparison has different inner width from
35614 type's inner width.
35616 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
35618 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
35619 markup, and similar issues. Remove @opindex entries for things
35620 that aren't options. Add missing -mmpy-option entries.
35622 2017-02-28 Jakub Jelinek <jakub@redhat.com>
35624 PR tree-optimization/79737
35625 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
35626 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
35627 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
35628 instead of byte_size. Formatting fix.
35629 (shift_bytes_in_array_right): Formatting fix.
35631 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
35634 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
35635 condition on optimize for the leaf function test.
35637 2017-02-28 Martin Liska <mliska@suse.cz>
35640 * read-rtl-function.c (function_reader::handle_unknown_directive):
35641 Bail out when one uses -flto.
35643 2017-02-28 Martin Liska <mliska@suse.cz>
35645 * common.opt: Replace space with tabular for options of <number>
35647 * config/i386/i386.opt: Show <number> value for
35648 -mlarge-data-threshold.
35649 * opts.c (print_filtered_help): Do not display number in hexadecimal
35652 2017-02-28 Martin Liska <mliska@suse.cz>
35654 * common.opt: Fix --help=option -Q for options which are of
35657 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
35659 * config/i386/i386.c (print_reg): Error out for values
35660 of 8-bit size in invalid integer register.
35662 2017-02-28 Martin Sebor <msebor@redhat.com>
35664 PR tree-optimization/79691
35665 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
35667 2017-02-28 Jakub Jelinek <jakub@redhat.com>
35670 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
35671 gcc_unreachable with output_operand_lossage.
35673 2017-02-28 Richard Biener <rguenther@suse.de>
35675 PR tree-optimization/79740
35676 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
35678 (visit_nary_op): Insert the nary into the hashtable if we
35679 pattern-matched sth.
35680 * tree-ssa-pre.c (eliminate_insert): Robustify.
35682 2017-02-28 Richard Biener <rguenther@suse.de>
35684 PR middle-end/79731
35685 * fold-const.c (decode_field_reference): Reject out-of-bound
35688 2017-02-28 Jakub Jelinek <jakub@redhat.com>
35690 * config/i386/i386.c: Include intl.h.
35691 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
35692 instead of just cond ? "..." : "...".
35693 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
35694 * coverage.c (read_counts_file): Likewise.
35695 * omp-offload.c: Include intl.h.
35696 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
35697 of just cond ? "..." : "...".
35698 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
35699 of just cond ? "..." : "...".
35701 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
35704 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
35706 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
35707 'tune for' CPU name.
35708 * config/arm/arm-cpu-data.h: Regenerated.
35710 2017-02-28 Richard Biener <rguenther@suse.de>
35712 PR tree-optimization/79732
35713 * tree-inline.c (expand_call_inline): Do not shadow var.
35715 2017-02-28 Richard Biener <rguenther@suse.de>
35717 PR tree-optimization/79723
35718 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
35719 address-space properly.
35721 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
35723 * doc/optinfo.texi (Optimization groups): Fix option used for
35725 * doc/invoke.texi (-fopt-info): Document "omp".
35726 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
35727 (OPTGROUP_ALL): Add OPTGROUP_OMP.
35728 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
35729 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
35730 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
35732 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
35734 * dumpfile.c (optgroup_options): Instead of "openmp", associate
35735 OPTGROUP_OMP with "omp".
35737 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
35740 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
35741 for arithmetic shift of unsigned V2DI.
35743 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
35745 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
35746 arc/linux.h headers.
35747 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
35748 (LINK_SPEC): Likewise.
35749 (ARC_TLS_EXTRA_START_SPEC): Likewise.
35750 (EXTRA_SPECS): Likewise.
35751 (STARTFILE_SPEC): Likewise.
35752 (ENDFILE_SPEC): Likewise.
35753 (LIB_SPEC): Likewise.
35754 (TARGET_SDATA_DEFAULT): Likewise.
35755 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
35756 (MULTILIB_DEFAULTS): Likewise.
35757 (DWARF2_UNWIND_INFO): Likewise.
35758 * config/arc/big.h: New file.
35759 * config/arc/elf.h: Likewise.
35760 * config/arc/linux.h: Likewise.
35761 * config/arc/t-uClibc: Remove.
35763 2017-02-27 Bin Cheng <bin.cheng@arm.com>
35765 PR tree-optimization/77536
35766 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
35767 (tree_transform_and_unroll_loop): Use above function to compute the
35768 estimated niter of unrolled loop and use it when scaling profile.
35769 Also use count info rather than frequency if it's non-zero.
35770 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
35771 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
35772 (vect_transform_loop): Call above function.
35774 2017-02-27 Richard Biener <rguenther@suse.de>
35776 PR tree-optimization/45397
35777 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
35778 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
35779 (visit_nary_op): Add pattern matching for CSEing sign-changed
35780 or truncated operations with wider ones.
35782 2017-02-27 Richard Biener <rguenther@suse.de>
35784 PR tree-optimization/79690
35785 * tree-vect-stmts.c (vectorizable_store): Use vector type
35786 built from the DR with address-space.
35788 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
35790 * doc/invoke.texi (Optimize Options): Refine the description
35791 of asan-use-after-return.
35793 2017-02-25 Alan Modra <amodra@gmail.com>
35795 PR rtl-optimization/79584
35796 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
35797 base, not ad->base_term, the reg within base. Remove assertion
35798 that ad->base == ad->base_term. Replace gen_int_mode using
35799 bogus mode with const0_rtx.
35801 2017-02-25 Jakub Jelinek <jakub@redhat.com>
35803 PR middle-end/79396
35804 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
35805 FMA_EXPR like tcc_binary or tcc_unary.
35807 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
35810 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
35812 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
35813 (output_loc_operands): Handle DW_OP_call_ref and
35814 DW_OP_GNU_variable_value.
35815 (struct variable_value_struct): New type.
35816 (struct variable_value_hasher): Likewise.
35817 (variable_value_hash): New variable.
35818 (string_types): Remove.
35819 (copy_loc_descr): New function.
35820 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
35821 (prepend_loc_descr_to_each): New function.
35822 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
35823 instead of add_loc_descr_to_each if the first argument is single
35824 location list and the second has multiple.
35825 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
35826 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
35827 when looking for variable value which doesn't have other location info.
35828 (loc_list_from_tree): Formatting fix.
35829 (gen_array_type_die): Simplify DW_AT_string_length handling.
35830 (adjust_string_types): Remove.
35831 (gen_subprogram_die): Don't call adjust_string_types nor test/set
35832 string_types. Call resolve_variable_values.
35833 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
35834 (resolve_addr_in_expr): Likewise. Add A argument.
35835 (copy_deref_exprloc): Remove deref argument. Adjust for the
35836 original expression being DW_OP_GNU_variable_value with optionally
35837 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
35838 optionally after it.
35839 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
35840 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
35841 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
35842 (variable_value_hasher::hash, variable_value_hasher::equal): New
35844 (resolve_variable_value_in_expr, resolve_variable_value,
35845 resolve_variable_values, note_variable_value_in_expr,
35846 note_variable_value): New functions.
35847 (dwarf2out_early_finish): Call note_variable_value on all toplevel
35850 2017-02-24 Jakub Jelinek <jakub@redhat.com>
35853 * opts.h (handle_generated_option): Add GENERATED_P argument.
35854 * opts-common.c (handle_option): Adjust function comment.
35855 (handle_generated_option): Add GENERATED_P argument, pass it to
35857 (control_warning_option): Pass false to handle_generated_option
35859 * opts.c (maybe_default_option): Pass true to handle_generated_option
35861 * optc-gen.awk: Likewise.
35863 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
35865 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
35866 a REG, look at the REG it is a SUBREG of.
35867 (splitter for cmpeqsi_t): Ditto.
35869 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
35871 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
35872 the special USEs with the pattern of the insn, not the insn itself.
35874 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
35877 * doc/invoke.texi: Document -mload-store-pairs.
35879 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
35880 Sandra Loosemore <sandra@codesourcery.com>
35882 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
35883 argument isn't a CONST_INT.
35884 (nios2_alternate_compare_const): Assert op is a CONST_INT.
35885 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
35886 (nios2_validate_compare): Bypass alternate compare logic if *op2
35887 is not a CONST_INT.
35888 (ldstwm_operation_p): Return false if first_base is not a REG or
35889 if first_offset is not a CONST_INT.
35891 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
35893 * config/cris/cris.md: Use correct operand in a define_peephole2.
35895 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
35897 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
35899 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
35901 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
35902 this_insn if it is an INSN or JUMP_INSN.
35903 (force_offsettable): Look at base, not at addr.
35904 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
35905 on things that aren't necessarily CONST_INTs.
35907 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
35909 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
35910 -mfpmath=sse is the default also for x86-32 targets with SSE2
35911 instruction set when @option{-ffast-math} is enabled
35913 2017-02-24 Jeff Law <law@redhat.com>
35915 PR rtl-optimizatoin/79286
35916 * ira.c (update_equiv_regs): Drop may_trap_p exception to
35919 2017-02-24 Richard Biener <rguenther@suse.de>
35921 PR tree-optimization/79389
35922 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
35925 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
35927 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
35928 function comment to reflect reality.
35929 (loop_exits_before_overflow): Fix typo in function description.
35931 2017-02-24 Richard Biener <rguenther@suse.de>
35933 PR tree-optimization/79389
35934 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
35935 properly that a threading opportunity exists. Detect conditional
35936 copy/constant propagation opportunities.
35938 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
35940 * config/visium/visium.md (type): Add trap.
35941 (b): New mode attribute.
35942 (*btst): Rename into...
35943 (*btst<mode>): ...this and adjust.
35944 (*cbranchsi4_btst_insn): Rename into...
35945 (*cbranch<mode>4_btst_insn): ...this and adjust.
35946 (trap): New define_insn.
35948 2017-02-23 Jakub Jelinek <jakub@redhat.com>
35950 PR tree-optimization/79389
35951 * ifcvt.c (struct noce_if_info): Add rev_cond field.
35952 (noce_reversed_cond_code): New function.
35953 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
35954 reversed_comparison_code. Formatting fix.
35955 (noce_try_store_flag): Test rev_cond != NULL in addition to
35956 reversed_comparison_code.
35957 (noce_try_store_flag_constants): Likewise.
35958 (noce_try_store_flag_mask): Likewise.
35959 (noce_try_addcc): Use rev_cond if non-NULL instead of
35960 reversed_comparison_code.
35961 (noce_try_cmove_arith): Likewise. Formatting fixes.
35962 (noce_try_minmax, noce_try_abs): Clear rev_cond.
35963 (noce_find_if_block): Initialize rev_cond.
35964 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
35965 instead of false as last argument never attempt to reverse it
35968 2017-02-23 Bin Cheng <bin.cheng@arm.com>
35970 PR tree-optimization/79663
35971 * tree-predcom.c (combine_chains): Process refs in reverse order
35972 only for ZERO length chains, and add explaining comment.
35974 2017-02-23 Jeff Law <law@redhat.com>
35976 PR tree-optimization/79578
35977 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
35978 in call to operand_equal_p.
35980 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
35983 * config/i386/cpuid.h: Fix another undefined behavior.
35985 2017-02-23 Richard Biener <rguenther@suse.de>
35987 PR tree-optimization/79683
35988 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
35989 vector types for data-refs.
35991 2017-02-23 Martin Liska <mliska@suse.cz>
35993 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
35995 2017-02-23 Jakub Jelinek <jakub@redhat.com>
35997 PR middle-end/79665
35998 * internal-fn.c (get_range_pos_neg): Moved to ...
35999 * tree.c (get_range_pos_neg): ... here. No longer static.
36000 * tree.h (get_range_pos_neg): New prototype.
36001 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
36002 are known to be in between 0 and signed maximum inclusive, try to
36003 expand both unsigned and signed divmod and use the cheaper one from
36006 2017-02-22 Jeff Law <law@redhat.com>
36008 PR tree-optimization/79578
36009 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
36010 to compare base operands.
36012 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
36015 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
36016 gpc_reg_operand instead of fpr_reg_operand.
36018 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
36020 * config/mips/mips.c (mips_return_in_memory): Force FP
36021 vector types to be returned in memory for o32 ABI.
36023 2017-02-22 Jakub Jelinek <jakub@redhat.com>
36025 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
36026 instead of DW_TAG_member for static data member declarations and don't
36027 set no_linkage_name for static inline data members.
36028 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
36031 2017-02-22 Martin Liska <mliska@suse.cz>
36033 * doc/invoke.texi: Replace inequality signs with square brackets
36036 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
36039 * lra-constraints.c (simplify_operand_subreg): Handle
36040 WORD_REGISTER_OPERATIONS targets.
36042 2017-02-22 Jakub Jelinek <jakub@redhat.com>
36045 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
36046 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
36047 elimination by swapping fld*.
36049 2017-02-22 Richard Biener <rguenther@suse.de>
36051 PR tree-optimization/79673
36052 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
36053 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
36054 irrelevant address-space qualifiers and avoiding a
36055 ADDR_SPACE_CONVERT_EXPR from fold_convert.
36057 2017-02-22 Richard Biener <rguenther@suse.de>
36059 PR tree-optimization/79666
36060 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
36061 to not symbolically negate if that may introduce undefined
36064 2017-02-22 Martin Liska <mliska@suse.cz>
36067 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
36068 * data-streamer-out.c (streamer_write_gcov_count_stream):
36070 * value-prof.c (stream_out_histogram_value): Make assert more
36071 precise based on type of counter.
36073 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
36076 * config/i386/i386.md (standard_x87sse_constant_load splitter):
36077 Use nonimmediate_operand instead of memory_operand for operand 1.
36078 (float-extend standard_x87sse_constant_load splitter): Ditto.
36080 2017-02-21 Jeff Law <law@redhat.com>
36082 PR tree-optimization/79621
36083 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
36084 blocks with edges to themselves.
36086 2017-02-21 Jakub Jelinek <jakub@redhat.com>
36089 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
36090 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
36091 Use gimple_call_builtin_p.
36094 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
36095 on temporarily removed DEBUG_INSNs.
36097 PR tree-optimization/79649
36098 * tree-loop-distribution.c (classify_partition): Give up on
36099 non-generic address space loads/stores.
36101 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
36103 * doc/loop.texi (Loop manipulation): Remove nonexistent
36104 tree_ssa_loop_version from the documentation.
36105 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
36107 2017-02-21 Jakub Jelinek <jakub@redhat.com>
36110 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
36111 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
36112 * config/rs6000/rs6000.c: Include except.h.
36113 (rs6000_expand_split_stack_prologue): Call
36114 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
36116 2017-02-21 Martin Jambor <mjambor@suse.cz>
36119 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
36120 have been analyzed.
36122 2017-02-21 Martin Jambor <mjambor@suse.cz>
36124 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
36125 for backward compatibility only.
36126 * doc/invoke.texi (Option Summary): Remove all references to
36127 -fipa-cp-alignment.
36129 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
36133 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36135 * lra-constraints.c (curr_insn_transform): Handle
36136 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
36138 2017-02-21 Martin Liska <mliska@suse.cz>
36140 * config/i386/i386.opt: Replace -masm-dialect with -masm.
36142 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
36144 PR translation/79638
36145 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
36147 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
36150 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
36151 (arm_function_ok_for_sibcall): Return false for an indirect call by
36152 descriptor if all the argument registers are used.
36153 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
36154 alignment of the function.
36156 2017-02-21 Jakub Jelinek <jakub@redhat.com>
36158 PR tree-optimization/61441
36159 * simplify-rtx.c (simplify_const_unary_operation): For
36160 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
36161 the sNaN unmodified.
36163 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
36165 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
36166 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
36167 instead of SYSTEM_HEADER_DIR.
36169 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
36170 Martin Liška <mliska@suse.cz>
36172 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
36173 Fix typos and grammar, use active voice, and clarify.
36175 2017-02-20 Marek Polacek <polacek@redhat.com>
36177 PR middle-end/79537
36178 * gimplify.c (gimplify_expr): Handle unused *&&L;.
36181 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
36183 2017-02-20 Jakub Jelinek <jakub@redhat.com>
36186 * config/i386/i386.c (ix86_expand_builtin): Handle
36187 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
36188 ix86_builtins_isa[fcode].isa as a requirement of those
36189 flags and any other flag in the bitmask.
36190 (ix86_init_mmx_sse_builtins): Use 0 instead of
36191 ~OPTION_MASK_ISA_64BIT as mask.
36192 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
36193 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
36194 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
36195 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
36197 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36200 * lra-constraints.c (split_reg): Check requested split mode
36201 is supported by the register.
36203 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36205 * lra-constraints.c (simplify_operand_subreg): Remove early
36208 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36211 * lra-constraints.c (curr_insn_transform): Tighten condition
36212 for converting SUBREG reloads from OP_OUT to OP_INOUT.
36214 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36217 * lra-constraints.c (curr_insn_transform): Handle
36218 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
36220 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
36223 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
36225 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
36227 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
36230 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
36233 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
36235 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
36238 2017-02-18 Jakub Jelinek <jakub@redhat.com>
36240 * final.c (last_columnnum, override_columnnum): New variables.
36241 (final_start_function): Set last_columnnum, pass it to begin_prologue
36242 hook and pass 0 to dwarf2out_begin_prologue.
36243 (final_scan_insn): Update override_columnnum. Pass last_columnnum
36244 to source_line debug hook.
36245 (notice_source_line): Compute last_columnnum and for debug_column_info
36246 return true on column changes.
36247 * debug.h (struct gcc_debug_hooks): Add column argument to
36248 source_line and begin_prologue hooks.
36249 (debug_nothing_int_charstar_int_bool): Remove prototype.
36250 (debug_nothing_int_int_charstar,
36251 debug_nothing_int_int_charstar_int_bool): New prototypes.
36252 (dwarf2out_begin_prologue): Add column argument.
36253 * debug.c (do_nothing_debug_hooks): Adjust source_line and
36254 begin_prologue hooks.
36255 (debug_nothing_int_charstar_int_bool): Remove.
36256 (debug_nothing_int_int_charstar,
36257 debug_nothing_int_int_charstar_int_bool): New functions.
36258 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
36259 through to dwarf2out_source_line.
36260 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
36261 (dwarf2out_source_line): Add column argument, emit it if requested.
36262 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
36264 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
36265 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
36266 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
36267 through to dwarf2out_begin_prologue.
36268 (vmsdbgout_source_line): Add column argument, pass it through to
36269 dwarf2out_source_line.
36270 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
36271 dbxout_source_line caller.
36272 (dbxout_source_line): Add column argument.
36274 * common.opt (gno-column-info, gcolumn-info): New options.
36275 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
36276 (check_die): Also test for multiple DW_AT_decl_column attributes.
36277 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
36278 DW_AT_decl_column if requested.
36279 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
36281 (gen_variable_die): Likewise.
36282 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
36283 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
36286 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
36287 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
36288 (ix86_handle_option): Handle OPT_m3dnowa.
36289 * doc/invoke.texi (-m3dnowa): Document.
36290 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
36291 -m3dnowa instead of -m3dnow -march=athlon.
36294 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
36295 instead of gcc_assert for K, r and R code checks. Formatting fixes.
36297 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36300 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
36301 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
36302 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
36303 generator for vsx_xxpermdi_<mode>_be.
36304 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
36305 force big-endian semantics.
36306 (vsx_xxpermdi_<mode>_be): New define_expand with same
36307 implementation as previous version of vsx_xxpermdi_<mode>.
36309 2017-02-17 Jakub Jelinek <jakub@redhat.com>
36311 PR tree-optimization/79327
36312 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
36313 variable, its initialization and use.
36315 2017-02-17 Julia Koval <julia.koval@intel.com>
36317 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
36318 (OPTION_MASK_ISA_PKU_UNSET): New.
36319 (ix86_handle_option): Handle -mrdpid.
36320 * config/i386/cpuid.h (bit_RDPID): New.
36321 * config/i386/driver-i386.c (host_detect_local_cpu):
36322 Detect RDPID feature.
36323 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
36324 * config/i386/i386-c.c (ix86_target_macros_internal):
36326 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
36327 (ix86_valid_target_attribute_inner_p): Add "rdpid".
36328 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
36329 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
36330 * config/i386/i386.md (define_insn "rdpid"): New.
36331 * config/i386/i386.opt Add -mrdpid.
36332 * config/i386/immintrin.h (_rdpid_u32): New.
36334 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
36336 PR rtl-optimization/79541
36337 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
36338 instead of transforming it into USE.
36340 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
36342 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
36343 If HONOR_SNANS (SFmode) force the input to a register.
36344 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
36345 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
36346 an frsp or similar insn.
36348 2017-02-17 Martin Liska <mliska@suse.cz>
36350 PR rtl-optimization/79577
36351 * params.def (selsched-max-sched-times): Increase minimum to 1.
36353 2017-02-17 Martin Liska <mliska@suse.cz>
36355 PR rtl-optimization/79574
36356 * gcse.c (want_to_gcse_p): Prevent integer overflow.
36358 2017-02-17 Martin Liska <mliska@suse.cz>
36360 PR tree-optimization/79529
36361 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
36362 ssa_defined_default_def_p to handle cases which are implicitly
36364 * tree-ssa.c (ssa_defined_default_def_p): New function.
36365 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
36366 which are implicitly defined.
36367 * tree-ssa.h (ssa_defined_default_def_p): Declare.
36369 2017-02-17 Richard Biener <rguenther@suse.de>
36371 PR middle-end/79576
36372 * params.def (max-ssa-name-query-depth): Limit to 10.
36374 2017-02-17 Richard Biener <rguenther@suse.de>
36376 PR tree-optimization/79552
36377 * tree-ssa-structalias.c (visit_loadstore): Properly verify
36380 2017-02-17 Richard Biener <rguenther@suse.de>
36383 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
36385 2017-02-17 Marek Polacek <polacek@redhat.com>
36387 PR middle-end/79536
36388 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
36389 (fold_negate_expr): New wrapper.
36391 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
36393 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
36394 Correct terminology and de-emphasize pre-standard behavior.
36396 2017-02-16 Alan Modra <amodra@gmail.com>
36398 PR rtl-optimization/79286
36399 * ira.c (def_dominates_uses): New function.
36400 (update_equiv_regs): Don't create an equivalence for insns that
36401 may trap where the register def does not dominate the use.
36403 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
36405 PR rtl-optimization/78127
36406 * lra.c (lra): Call lra_eliminate before finish the loop after
36409 2017-02-16 Richard Biener <rguenther@suse.de>
36411 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
36413 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
36414 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
36415 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
36416 (isl_val_int_from_wi): New function.
36417 (extract_affine_gmp): Rename to ...
36418 (extract_affine_wi): ... this, take a widest_int.
36419 (extract_affine_int): Just wrap extract_affine_wi.
36420 (add_param_constraints): Use isl_val_int_from_wi.
36421 (add_loop_constraints): Likewise, and extract_affine_wi.
36423 2017-02-15 Jeff Law <law@redhat.com>
36425 PR middle-end/79521
36426 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
36427 ira_init_register_move_cost_if_necessary.
36429 2017-02-15 Martin Sebor <msebor@redhat.com>
36431 PR middle-end/32003
36432 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
36433 removed in a prior commit.
36435 2017-02-15 Bin Cheng <bin.cheng@arm.com>
36437 PR tree-optimization/79347
36438 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
36439 counters during peeling.
36441 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
36443 * Makefile.in (site.exp): Remove "set ISLVER".
36445 2017-02-15 Jakub Jelinek <jakub@redhat.com>
36448 * real.c (real_from_integer): Call real_convert even for decimal.
36450 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
36453 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
36455 2017-02-14 Andrew Pinski <apinski@cavium.com>
36457 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
36458 cores and change the partno/implementer to be correct.
36459 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
36460 the 'B" as the implementer.
36461 * config/aarch64/aarch64-tune.md: Regenerate.
36463 2017-02-14 Carl Love <cel@us.ibm.com>
36465 * config/rs6000/rs6000.c: Add case statement entry to make the
36466 xvcvuxdsp built-in argument unsigned.
36467 * config/rs6000/vsx.md: Fix the source and return operand types so they
36468 match the instruction definitions from the ISA document. Fix typo
36469 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
36472 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
36475 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
36476 member early_clobber_alts.
36477 * lra-lives.c (reg_early_clobber_p): New.
36478 (process_bb_lives): Use it.
36479 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
36480 (debug_operand_data): Initialize early_clobber_alts.
36481 (setup_operand_alternative): Set up early_clobber_alts.
36482 (collect_non_operand_hard_regs): Ditto. Pass early clobber
36483 alternatives to new_insn_reg.
36484 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
36486 (lra_update_insn_regno_info): Pass the new arg.
36488 2017-02-14 Jakub Jelinek <jakub@redhat.com>
36490 PR middle-end/79505
36491 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
36492 (new_oacc_loop_raw): Don't clear already cleared fields.
36495 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
36496 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
36497 _mm512_prefetch_i64gather_ps): New inline functions and macros.
36499 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
36502 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
36504 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
36507 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
36508 the extra instruction to the right place to store 128-bit constant
36511 2017-02-14 Martin Sebor <msebor@redhat.com>
36513 PR middle-end/79448
36514 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
36515 warning for strings of unknown length.
36517 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
36519 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
36521 2017-02-14 Jeff Law <law@redhat.com>
36524 * ira-costs.c (scan_one_insn): Initialize register move costs
36525 for pseudos seen in USE/CLOBBER insns.
36527 PR tree-optimization/79095
36528 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
36529 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
36530 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
36531 if the operands are known to be not equal, then the resulting range
36533 (intersect_ranges): If the new range is ~[0,0] and the old range is
36534 wide, then prefer ~[0,0].
36535 * tree-vrp.c (overflow_comparison_p_1): New function.
36536 (overflow_comparison_p): New function.
36537 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
36538 if NAME is used in an overflow test.
36539 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
36540 overflow check that can be expressed as an equality test, then adjust
36541 ops to be that equality test.
36543 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36545 * config/s390/s390-builtin-types.def: Remove flags argument.
36546 * config/s390/s390.c (s390_init_builtins): Likewise.
36548 2017-02-14 Martin Liska <mliska@suse.cz>
36550 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
36551 vector. Fix trailing white spaces.
36553 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
36555 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
36558 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36560 PR rtl-optimization/68664
36561 * config/arm/arm.c (arm_sched_can_speculate_insn):
36562 New function. Declare prototype.
36563 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
36565 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36567 PR rtl-optimization/68664
36568 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
36570 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
36572 2017-02-14 Amit Pawar <amit.pawar@amd.com>
36574 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
36575 max skip bytes for function, loop and jump.
36577 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
36579 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
36580 ABS_EXPR for gimple dump.
36582 2017-02-14 Jakub Jelinek <jakub@redhat.com>
36585 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
36587 PR tree-optimization/79408
36588 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
36589 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
36590 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
36591 also if rhs1 is INTEGER_CST.
36593 2017-02-14 Richard Biener <rguenther@suse.de>
36595 PR middle-end/79432
36596 * tree-into-ssa.c (insert_phi_nodes): When the function can
36597 have abnormal edges rewrite SSA names with broken use-def
36598 dominance out of SSA and register them for PHI insertion.
36600 2017-02-13 Martin Sebor <msebor@redhat.com>
36602 PR middle-end/79496
36603 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
36604 clearing info.nowrite flag when snprintf size argument is a range.
36606 2017-02-13 Jakub Jelinek <jakub@redhat.com>
36608 * cprop.c (cprop_jump): Add missing space in string literal.
36609 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
36610 (get_constraint_for_component_ref): Likewise.
36611 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
36612 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
36613 * lra-constraints.c (process_alt_operands): Likewise.
36614 * ipa-inline.c (inline_small_functions): Likewise.
36615 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
36616 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
36617 * trans-mem.c (diagnose_tm_1_op): Likewise.
36618 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
36619 (grid_parallel_clauses_gridifiable): Likewise.
36621 * config/nvptx/mkoffload.c (process): Add space in between
36624 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
36625 "MOD4_SSE_REGS" and "ALL_REGS".
36627 * spellcheck.c (test_data): Add , in between "foo" and "food".
36629 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36632 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
36633 boundary crossing check and subsequent code generation agree.
36635 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36637 * config/aarch64/aarch64.c (has_memory_op): Delete.
36638 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
36641 2017-02-13 Jakub Jelinek <jakub@redhat.com>
36643 PR rtl-optimization/79388
36644 PR rtl-optimization/79450
36645 * combine.c (distribute_notes): When removing TEM_INSN for which
36646 corresponding dest has last value recorded, invalidate that last
36649 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36651 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
36652 of explicit '@'. Add missing assembly comment marker on branch costs
36655 2017-02-13 Nathan Sidwell <nathan@acm.org>
36657 * gengtype-lex.l (<in_struct>): Add '/'.
36659 2017-02-13 Martin Liska <mliska@suse.cz>
36662 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
36664 2017-02-13 Richard Biener <rguenther@suse.de>
36666 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
36668 * configure: Re-generate.
36669 * config.in: Likewise.
36670 * graphite-dependences.c: Simplify as if
36671 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
36672 * graphite-isl-ast-to-gimple.c: Likewise.
36673 * graphite-optimize-isl.c: Likewise.
36674 * graphite-poly.c: Likewise.
36675 * graphite-sese-to-poly.c: Likewise.
36676 * graphite.h: Likewise.
36677 * toplev.c: Include isl/version.h and use isl_version () for
36678 printing the ISL version.
36679 * doc/install.texi: Update ISL requirement.
36681 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
36683 * doc/standards.texi (Standards): Update reference to
36686 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
36688 * doc/extend.texi (Named Address Spaces): sourceware.org now
36690 * doc/install.texi (Binaries): Ditto.
36693 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
36695 * doc/cpp.texi: Replace "stringify"/"stringification" with C
36696 standard terminology "stringize"/"stringizing" throughout.
36697 * doc/cppinternals.texi: Likewise.
36699 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
36701 * doc/extend.texi: Fix some spelling mistakes and typos.
36702 * doc/invoke.texi: Likewise.
36704 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
36707 * params.def (inline-min-speedup) Change from 10 to 8.
36709 2017-02-11 Jakub Jelinek <jakub@redhat.com>
36711 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
36714 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
36717 * ipa-inline-analysis.c (get_minimal_bb): New function.
36718 (record_modified): Use it.
36719 (remap_edge_change_prob): Handle also ancestor functions.
36721 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
36723 * doc/contrib.texi (Contributors): Remove broken link into
36724 the Mauve CVS repository.
36726 2017-02-11 Jakub Jelinek <jakub@redhat.com>
36728 PR middle-end/79454
36729 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
36730 result computation whenever lhs doesn't have vector mode, not
36731 just when it has BLKmode.
36733 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
36735 * doc/makefile.texi (profiledbootstrap): Refer to the
36736 installation instructions only in textual form.
36738 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36741 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
36743 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
36745 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
36746 (Specific): Update mingw-w64 reference.
36748 (Specific): Remove broken link to Renesas RX processor.
36750 2017-02-10 Richard Biener <rguenther@suse.de>
36752 * toplev.c (process_options): Do not mention obsolete graphite
36753 options when printing sorry message about missing graphite support.
36754 Mention -floop-nest-optimize.
36756 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
36758 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
36759 (vtst_p16): Likewise.
36760 (vtstq_p8): Likewise.
36761 (vtstq_p16): Likewise.
36763 (vtstq_p64): Likewise.
36764 * config/arm/arm_neon.h (vgetq_lane_p64): New.
36765 (vset_lane_p64): New.
36766 (vsetq_lane_p64): New.
36768 2017-02-10 Jakub Jelinek <jakub@redhat.com>
36770 PR tree-optimization/79411
36771 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
36772 stmt operands are SSA_NAMEs used in abnormal phis.
36773 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
36776 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
36779 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
36782 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
36784 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
36786 2017-02-09 Jakub Jelinek <jakub@redhat.com>
36788 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
36792 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
36793 not arbitrary TREE_CONSTANT.
36796 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
36797 "omp declare target link" attribute unless is_global_var.
36798 * omp-offload.c (find_link_var_op): Likewise.
36800 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
36801 Chung-Lin Tang <cltang@codesourcery.com>
36803 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
36805 (gimplify_adjust_omp_clauses): Don't delete TILE.
36806 (gimplify_omp_for): Deal with TILE.
36807 * internal-fn.c (expand_GOACC_TILE): New function.
36808 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
36810 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
36811 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
36813 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
36814 avoid DIV for outermost collapse var.
36815 (expand_oacc_for): Insert tile element loop as needed. Adjust.
36816 Remove out of date comments, fix whitespace.
36817 * omp-general.c (omp_extract_for_data): Deal with tiling.
36818 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
36819 adjust OLF_DIM_BASE value.
36820 (struct omp_for_data): Add tiling field.
36821 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
36822 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
36823 for auto loops. Remove default auto determining, moved to
36824 oacc_loop_fixed_partitions.
36825 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
36826 stmts, add e_mask field.
36827 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
36828 (oacc_thread_numbers): Use oacc_dim_call.
36829 (oacc_xform_tile): New.
36830 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
36831 (finish_oacc_loop): Adjust for ifns vector.
36832 (oacc_loop_discover_walk): Append loop abstraction sites to list,
36833 add case for GOACC_TILE fns.
36834 (oacc_loop_xform_loop): Delete.
36835 (oacc_loop_process): Iterate over call list directly, and add
36836 handling for GOACC_TILE fns.
36837 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
36839 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
36840 vector partitioning to outer loops. Assign 2 partitions to loops
36841 when available. Add TILE handling.
36842 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
36843 (execite_oacc_device_lower): Process GOACC_TILE fns,
36844 ignore unknown specs.
36845 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
36846 * tree.c (omp_clause_num_ops): Adjust TILE ops.
36847 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
36849 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
36851 * configure.ac (ACX_BUGURL): Update.
36852 * configure: Regenerate.
36854 2017-02-09 Richard Biener <rguenther@suse.de>
36856 PR tree-optimization/69823
36857 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
36858 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
36860 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
36862 * config/arc/arc-c.def: Add __NPS400__ definition.
36863 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
36864 (TARGET_NPS400): Define.
36866 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
36868 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
36870 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
36871 pointer, arch_info.
36872 (arc_cpu_types): Fill the arch_info field with a pointer into the
36873 arc_arch_types table.
36874 (arc_selected_cpu): Declare.
36875 * config/arc/arc.c (arc_selected_cpu): Make global.
36876 (arc_selected_arch): Delete.
36877 (arc_base_cpu): Delete.
36878 (arc_override_options): Remove references to deleted variables,
36879 update access to arch information.
36880 (ARC_OPT): Update access to arch information.
36881 (ARC_OPTX): Likewise.
36882 * config/arc/arc.h (arc_base_cpu): Remove declaration.
36883 (TARGET_ARC600): Update access to arch information.
36884 (TARGET_ARC601): Likewise.
36885 (TARGET_ARC700): Likewise.
36886 (TARGET_EM): Likewise.
36887 (TARGET_HS): Likewise.
36888 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
36891 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
36894 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
36895 condition/operands for integer GE/LE/GEU/LEU operations.
36897 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
36899 PR translation/79397
36900 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
36903 2017-02-08 Martin Jambor <mjambor@suse.cz>
36906 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
36907 whether allocation happened.
36908 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
36909 nothing was allocated.
36911 2017-02-08 Jakub Jelinek <jakub@redhat.com>
36913 PR tree-optimization/79408
36914 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
36915 constant, but SSA_NAME with a known integer range, use the minimum
36916 of that range instead of op1 to determine if modulo can be replaced
36917 with its first operand.
36919 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36921 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
36923 2017-02-08 Richard Biener <rguenther@suse.de>
36925 PR tree-optimization/71824
36926 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
36927 Check all loops contained in the merged region.
36929 2017-02-07 Andrew Pinski <apinski@cavium.com>
36931 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
36933 2017-02-07 Andrew Pinski <apinski@cavium.com>
36935 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
36936 (thunderxt88): Likewise.
36937 (thunderxt81): Disable LSE and change v8.1 to v8.
36938 (thunderxt83): Likewise.
36940 2017-02-07 Jakub Jelinek <jakub@redhat.com>
36941 Richard Biener <rguenther@suse.de>
36943 PR middle-end/79399
36944 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
36945 type from int to size_t.
36946 * ira-costs.c (struct_costs_size): Change type from int to size_t.
36948 2017-02-07 Jakub Jelinek <jakub@redhat.com>
36950 PR rtl-optimization/79386
36951 * cprop.c (bypass_conditional_jumps): Initialize
36952 bypass_last_basic_block already before splitting bbs after
36953 unconditional traps...
36954 (bypass_conditional_jumps): ... rather than here.
36957 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
36958 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
36959 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
36960 fix -masm=intel patterns.
36962 2017-02-07 Richard Biener <rguenther@suse.de>
36964 PR tree-optimization/79256
36965 PR middle-end/79278
36966 * builtins.c (get_object_alignment_2): Use min_align_of_type
36967 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
36968 and ADJUST_FIELD_ALIGN.
36970 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
36972 * doc/tm.texi: Regenerate.
36973 * stor-layout.c (layout_decl): Adjust.
36974 (update_alignment_for_field): Likewise.
36975 (place_field): Likewise.
36976 (min_align_of_type): Likewise.
36977 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
36978 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
36979 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
36980 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
36981 * config/frv/frv.c (frv_adjust_field_align): Likewise.
36982 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
36983 * config/i386/i386.c (x86_field_alignment): Likewise.
36984 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
36985 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
36986 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
36987 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
36988 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
36989 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
36993 2017-01-30 Richard Biener <rguenther@suse.de>
36995 PR tree-optimization/79256
36996 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
36997 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
37000 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
37002 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
37003 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
37004 builtins to SImode and emit a zero-extend, if necessary.
37006 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
37008 * docs/invoke.texi (RISC-V Options): Alphabetize.
37010 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
37012 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
37015 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
37017 * config/riscv/riscv.c: New file.
37018 * common/config/riscv/riscv-common.c: Likewise.
37019 * config.gcc: Likewise.
37020 * config/riscv/constraints.md: Likewise.
37021 * config/riscv/elf.h: Likewise.
37022 * config/riscv/generic.md: Likewise.
37023 * config/riscv/linux.h: Likewise.
37024 * config/riscv/multilib-generator: Likewise.
37025 * config/riscv/peephole.md: Likewise.
37026 * config/riscv/pic.md: Likewise.
37027 * config/riscv/predicates.md: Likewise.
37028 * config/riscv/riscv-builtins.c: Likewise.
37029 * config/riscv/riscv-c.c: Likewise.
37030 * config/riscv/riscv-ftypes.def: Likewise.
37031 * config/riscv/riscv-modes.def: Likewise.
37032 * config/riscv/riscv-opts.h: Likewise.
37033 * config/riscv/riscv-protos.h: Likewise.
37034 * config/riscv/riscv.h: Likewise.
37035 * config/riscv/riscv.md: Likewise.
37036 * config/riscv/riscv.opt: Likewise.
37037 * config/riscv/sync.md: Likewise.
37038 * config/riscv/t-elf-multilib: Likewise.
37039 * config/riscv/t-linux: Likewise.
37040 * config/riscv/t-linux-multilib: Likewise.
37041 * config/riscv/t-riscv: Likewise.
37042 * configure.ac: Likewise.
37043 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
37044 Waterman as RISC-V maintainers.
37045 * doc/install.texi: Add RISC-V entries.
37046 * doc/invoke.texi: Add RISC-V options section.
37047 * doc/md.texi: Add RISC-V constraints section.
37048 * configure: Regenerated.
37050 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
37053 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
37054 false values to be constant vectors with all 0 or all 1 bits set.
37055 (vcondu<mode><mode>): Likewise.
37056 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
37058 (fpmask_comparison_operator): Update comment.
37059 (vecint_comparison_operator): New predicate.
37060 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
37061 vector conditionals when the true and false values are constant
37062 vectors with all 0 bits or all 1 bits set.
37064 2017-02-06 Martin Sebor <msebor@redhat.com>
37066 PR tree-optimization/79376
37067 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
37069 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
37071 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
37072 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
37073 to simplify split condition.
37075 2017-02-06 Jakub Jelinek <jakub@redhat.com>
37077 * omp-expand.c (oxpand_omp_atomic_fetch_op,
37078 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
37081 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
37083 PR rtl-optimization/68664
37084 * target.def (can_speculate_insn): New hook.
37085 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
37086 * doc/tm.texi: Regenerate.
37087 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
37088 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
37089 (rs6000_sched_can_speculate_insn): New function.
37091 2017-02-06 Jakub Jelinek <jakub@redhat.com>
37093 PR tree-optimization/79284
37094 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
37095 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
37096 vectorizable_mask_load_store, vectorizable_operation,
37097 vect_is_simple_cond, get_same_sized_vectype): Use it instead
37098 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
37099 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
37100 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
37101 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
37102 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
37103 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
37104 is_gimple_assign (stmt). Replace another such test with
37105 is_gimple_assign (stmt).
37107 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
37110 * config/avr/avr.c (rtl-iter.h): Include it.
37111 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
37112 (avr_legitimate_combined_insn): ...and implementation.
37114 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
37116 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
37117 * config/s390/s390.c (s390_const_operand_ok)
37118 (s390_canonicalize_comparison, s390_extract_part)
37119 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
37120 (s390_contiguous_bitmask_p, s390_rtx_costs)
37121 (legitimize_pic_address): Likewise.
37122 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
37123 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
37124 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
37125 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
37126 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
37128 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
37130 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
37131 REGNO($0) == REGNO($1).
37133 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37135 * config/s390/linux.h(SIZE_TYPE): Add comment.
37137 2017-02-06 Julian Brown <julian@codesourcery.com>
37138 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
37139 Virendra Pathak <virendra.pathak@broadcom.com>
37141 * config/aarch64/aarch64-cores.def: Change the scheduler
37143 * config/aarch64/aarch64.md: Include thunderx2t99.md.
37144 * config/aarch64/thunderx2t99.md: New file.
37146 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
37148 * doc/standards.texi (Go Language): Update link to language
37151 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
37153 * tree-eh.c (lower_resx): Sanitize profile.
37154 (cleanup_empty_eh_move_lp): Likewise.
37156 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
37159 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
37161 * cfgloopmanip.h (loop_version): Update prototype.
37162 * modulo-sched.c (sms_schedule): Update call of loop_version.
37163 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
37164 * tree-parloops.c (gen_parallel_loop): Likewise.
37165 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
37166 * tree-ssa-loop-split.c (split_loop): Likewise.
37167 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
37168 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
37170 2017-02-05 Martin Liska <mliska@suse.cz>
37173 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
37175 (print_operand_address): Initialize a struct to zero.
37177 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
37179 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
37180 garbage collector only in textual form.
37182 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
37184 * doc/extend.texi (x86 specific memory model extensions for
37185 transactional memory): Simplify a phrase.
37187 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
37190 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
37191 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
37192 (atomic_storedi_1): Likewise.
37194 2017-02-04 Jakub Jelinek <jakub@redhat.com>
37196 PR tree-optimization/79338
37197 * tree-parloops.c (gather_scalar_reductions): Don't call
37198 vect_analyze_loop_form for loop->inner before destroying loop's
37201 2017-02-03 Martin Sebor <msebor@redhat.com>
37203 PR tree-optimization/79327
37204 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
37205 when precision has resulted in leading zeros.
37206 (format_integer): Adjust the likely counter to assume an unknown
37207 argument that may be zero is non-zero.
37209 2017-02-03 Jason Merrill <jason@redhat.com>
37212 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
37213 avoid copying non-taken branch.
37215 2017-02-03 Jakub Jelinek <jakub@redhat.com>
37217 PR tree-optimization/79340
37218 * tree-vect-loop.c (vectorizable_reduction): Release
37219 vec_defs elements after safe_splicing them into other vectors.
37222 PR tree-optimization/79327
37223 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
37224 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
37226 (format_integer): Use wide_int_to_tree instead of build_int_cst
37227 + to_?hwi. If argmin is NULL, just set argmin and argmax to
37228 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
37229 of shortest and longest sequence.
37231 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
37233 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
37234 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
37236 2017-02-03 Walter Lee <walt@tilera.com>
37239 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
37240 after initial stackframe link reg save.
37241 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
37243 2017-02-03 Jakub Jelinek <jakub@redhat.com>
37246 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
37247 wu for stxssp alternative.
37249 2017-02-03 Martin Sebor <msebor@redhat.com>
37251 PR tree-optimization/79352
37252 * gimple-fold.c (get_range_strlen): Add argument.
37253 (get_range_strlen): Change return type to bool.
37254 (get_maxval_strlen): Pass in a dummy argument.
37255 * gimple-fold.h (get_range_strlen): Change return type to bool.
37256 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
37257 * tree.h (array_at_struct_end_p): Add argument.
37258 * tree.c (array_at_struct_end_p): Handle it.
37260 2017-02-03 Martin Liska <mliska@suse.cz>
37263 * multiple_target.c (create_dispatcher_calls): Redirect edge
37264 from a caller of a dispatcher.
37265 (expand_target_clones): Make the clones local.
37266 (ipa_target_clone): Do both target clones and resolvers.
37267 (ipa_dispatcher_calls): Remove the pass.
37268 (pass_dispatcher_calls::gate): Likewise.
37269 (make_pass_dispatcher_calls): Likewise.
37270 * passes.def (pass_target_clone): Put as very first IPA early
37273 2017-02-03 Martin Liska <mliska@suse.cz>
37275 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
37276 in case of a function with ifunc attribute.
37278 2017-02-03 Martin Liska <mliska@suse.cz>
37280 * cgraph.c (cgraph_node::dump): Dump function version info.
37281 * symtab.c (symtab_node::dump_base): Add missing new line.
37283 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
37285 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
37286 (ifcombine_ifandif): Use it.
37288 2017-02-03 Martin Liska <mliska@suse.cz>
37290 * doc/invoke.texi: Document default value for
37291 use-after-scope-direct-emission-threshold.
37293 2017-02-03 Martin Liska <mliska@suse.cz>
37295 PR tree-optimization/79339
37296 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
37297 (format_floating): Likewise.
37299 2017-02-03 Martin Liska <mliska@suse.cz>
37302 * ipa-prop.c (ipa_node_params_t::insert): Remove current
37304 (ipa_node_params_t::remove): Likewise.
37305 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
37306 initialization from removed ipa_node_params_t::insert.
37307 (ipa_node_params::~ipa_node_params): Move from removed
37308 ipa_node_params_t::release.
37309 * symbol-summary.h (symbol_summary::m_released): New member.
37310 Do not release a summary twice. Do not allow to call finalizer
37311 for types of a summary that live in GGC memory.
37313 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
37315 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
37318 2017-02-02 Martin Sebor <msebor@redhat.com>
37320 PR middle-end/79275
37321 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
37322 (format_string): Tighten up the range of output for non-constant
37323 strings and correct the expected range for wide non-constant strings.
37325 2017-02-02 Martin Sebor <msebor@redhat.com>
37327 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
37329 PR middle-end/32003
37330 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
37332 (-fdump-tree-@var): Add to index and document how to come up
37333 with pass-specific option and dump file names.
37334 (-fdump-passes): Clarify where to look for output.
37336 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
37338 PR middle-end/77445
37339 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
37340 statistics of the analyzed path; allow threading for speed when
37341 any of BBs along the path are optimized for speed.
37343 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
37345 PR middle-end/78468
37346 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
37347 settings of the virtual registers.
37350 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
37352 * explow.c (get_dynamic_stack_size): Take known alignment of stack
37353 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
37356 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37358 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
37359 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
37361 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37363 * config/s390/s390.md: Add missing comments with the expanded
37365 * config/s390/vector.md: Likewise.
37366 * config/s390/vx-builtins.md: Likewise.
37368 2017-02-02 Jakub Jelinek <jakub@redhat.com>
37371 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
37372 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
37373 conditions on a single line.
37375 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37377 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
37378 __S390_VX__ to __VX__.
37380 2017-02-01 Andrew Pinski <apinski@cavium.com>
37382 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
37383 stmt_info to record_stmt_cost.
37384 (vect_get_known_peeling_cost): Pass stmt_info if known to
37386 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
37387 cpu_vector_cost field into
37388 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
37389 field into vec_int_stmt_cost and vec_fp_stmt_cost.
37390 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
37391 splitting of scalar_stmt_cost and vec_stmt_cost.
37392 (thunderx_vector_cost): Likewise.
37393 (cortexa57_vector_cost): LIkewise.
37394 (exynosm1_vector_cost): Likewise.
37395 (xgene1_vector_cost): Likewise.
37396 (thunderx2t99_vector_cost): Improve after the splitting of the two
37398 (aarch64_builtin_vectorization_cost): Update for the splitting of
37399 scalar_stmt_cost and vec_stmt_cost.
37401 2017-02-01 Torvald Riegel <triegel@redhat.com>
37402 Richard Henderson <rth@redhat.com>
37404 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
37405 conditional on existance of a fast atomic load.
37406 * optabs-query.c (can_atomic_load_p): New function.
37407 * optabs-query.h (can_atomic_load_p): Declare it.
37408 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
37409 no fast atomic load is available for the particular size of access.
37410 (expand_atomic_compare_and_swap): Likewise.
37411 (expand_atomic_load): Likewise.
37412 (expand_atomic_store): Likewise.
37413 (expand_atomic_fetch_op): Likewise.
37414 * testsuite/lib/target-supports.exp
37415 (check_effective_target_sync_int_128): Remove x86 because it provides
37416 no fast atomic load.
37417 (check_effective_target_sync_int_128_runtime): Likewise.
37419 2017-02-01 Richard Biener <rguenther@suse.de>
37421 * graphite.c: Include tree-vectorizer.h for find_loop_location.
37422 (graphite_transform_loops): Provide opt-info for optimized nests.
37423 * tree-parloop.c (parallelize_loops): Provide opt-info for
37424 parallelized loops.
37426 2017-02-01 Richard Biener <rguenther@suse.de>
37428 PR middle-end/79315
37429 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
37430 was not set before.
37432 2017-02-01 Richard Biener <rguenther@suse.de>
37434 PR tree-optimization/71824
37435 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
37436 Verify the loops are valid in the merged SESE region.
37437 (scop_detection::can_represent_loop_1): Check analyzing the
37438 evolution of the number of iterations in the region succeeds.
37440 2017-01-31 Ian Lance Taylor <iant@golang.org>
37442 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
37443 REG_ARGS_SIZE note to 32-bit push insns and call insn.
37445 2017-01-31 David Malcolm <dmalcolm@redhat.com>
37447 PR preprocessor/79210
37448 * input.c (get_substring_ranges_for_loc): Replace line_width
37449 assertion with error-handling.
37451 2017-01-31 Richard Biener <rguenther@suse.de>
37453 PR tree-optimization/77318
37454 * graphite-sese-to-poly.c (extract_affine): Fix assert.
37455 (create_pw_aff_from_tree): Take loop parameter.
37456 (add_condition_to_pbb): Pass loop of the condition to
37457 create_pw_aff_from_tree.
37459 2017-01-31 Jakub Jelinek <jakub@redhat.com>
37461 * config/s390/s390.c (s390_asan_shadow_offset): New function.
37462 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
37464 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
37468 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
37470 (convert_int_to_float128): Likewise.
37471 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
37472 (convert_int_to_float128): Likewise.
37473 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
37474 (UNSPEC_IEEE128_CONVERT): Likewise.
37475 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
37476 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
37477 Use local variables for IBM extended format.
37478 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
37479 (fix_trunc<mode>si2_fprs): Likewise.
37480 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
37481 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
37482 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
37483 to know that we can now have integers of all sizes in vector
37485 (fix<uns>_<mode>di2_hw): Likewise.
37486 (float<uns>_<mode>si2_hw): Likewise.
37487 (fix_<mode>si2_hw): Likewise.
37488 (fixuns_<mode>si2_hw): Likewise.
37489 (float<uns>_<mode>di2_hw): Likewise.
37490 (float_<mode>di2_hw): Likewise.
37491 (float_<mode>si2_hw): Likewise.
37492 (floatuns_<mode>di2_hw): Likewise.
37493 (floatuns_<mode>si2_hw): Likewise.
37494 (xscvqp<su>wz_<mode>): Delete, no longer used.
37495 (xscvqp<su>dz_<mode>): Likewise.
37496 (xscv<su>dqp_<mode>): Likewise.
37497 (ieee128_mfvsrd_64bit): Likewise.
37498 (ieee128_mfvsrd_32bit): Likewise.
37499 (ieee128_mfvsrwz): Likewise.
37500 (ieee128_mtvsrw): Likewise.
37501 (ieee128_mtvsrd_64bit): Likewise.
37502 (ieee128_mtvsrd_32bit): Likewise.
37504 2017-01-31 Martin Liska <mliska@suse.cz>
37507 * ipa-prop.c (ipa_free_all_node_params): Call release method
37508 instead of ~sumbol_summary to not to trigger double times
37511 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
37513 PR tree-optimization/71691
37514 * bitmap.h (class auto_bitmap): New.
37515 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
37516 is_maybe_undefined instead of ssa_undefined_value_p.
37518 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37520 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
37521 __S390_ARCH_LEVEL__ to __ARCH__.
37523 2017-01-31 Jakub Jelinek <jakub@redhat.com>
37525 PR tree-optimization/79267
37526 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
37527 if should_remove_lhs_p is true.
37529 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
37532 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
37533 (add_alignment_attribute): New.
37534 (base_type_die): Add alignment attribute.
37535 (subrange_type_die): Likewise.
37536 (modified_type_die): Likewise.
37537 (gen_array_type_die): Likewise.
37538 (gen_descr_array_type_die: Likewise.
37539 (gen_enumeration_type_die): Likewise.
37540 (gen_subprogram_die): Likewise.
37541 (gen_variable_die): Likewise.
37542 (gen_field_die): Likewise.
37543 (gen_ptr_to_mbr_type_die): Likewise.
37544 (gen_struct_or_union_type_die): Likewise.
37545 (gen_subroutine_type_die): Likewise.
37546 (gen_typedef_die): Likewise.
37547 (base_type_cmp): Compare alignment attribute.
37549 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37552 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
37553 (setb_unsigned) New pattern for setb with CCUNS.
37554 * config/rs6000/rs6000.c (expand_block_compare): Use a different
37555 subfc./subfe sequence to avoid overflow problems. Generate a
37556 shorter sequence with cmpld/setb for power9.
37557 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
37558 for generating subfc. instruction.
37559 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
37560 now uses this instruction.
37562 2017-01-30 Ian Lance Taylor <iant@google.com>
37565 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
37566 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
37568 2017-01-30 Martin Sebor <msebor@redhat.com>
37570 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
37571 Move constant to the right of a relational operator.
37572 (get_mpfr_format_length, format_character, format_string): Ditto.
37573 (should_warn_p, maybe_warn): Same.
37575 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
37577 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
37580 * asan.c (get_translation_unit_decl): Remove function.
37581 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
37583 2017-01-30 Martin Liska <mliska@suse.cz>
37585 PR gcov-profile/79259
37586 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
37587 -fprofile-generate.
37589 2017-01-30 Martin Liska <mliska@suse.cz>
37592 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
37593 Initialize variables with NULL value.
37595 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
37598 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
37600 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
37602 2017-01-30 Richard Biener <rguenther@suse.de>
37604 PR tree-optimization/79276
37605 * tree-vrp.c (process_assert_insertions): Properly adjust common
37606 when removing a duplicate.
37608 2017-01-30 Richard Biener <rguenther@suse.de>
37610 PR tree-optimization/79256
37611 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
37612 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
37614 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
37616 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
37619 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
37620 ("*r<noxa>sbg_<mode>_sll_bitmask")
37621 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
37622 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
37623 Use contiguous_bitmask_nowrap_operand.
37625 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37628 * config/rs6000/altivec.h (vec_xl): Revise #define.
37629 (vec_xst): Likewise.
37631 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
37633 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
37635 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
37637 PR rtl-optimization/79194
37638 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
37639 traps before call to bypass_conditional_jumps.
37641 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
37643 PR tree-optimization/71374
37644 * lra-constraints.c (check_conflict_input_operands): New.
37645 (match_reload): Use it.
37647 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
37650 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
37651 account to calculate conflict_set.
37653 2017-01-27 Bin Cheng <bin.cheng@arm.com>
37655 PR rtl-optimization/78559
37656 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
37657 other_insn in combine.
37659 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
37661 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
37662 uint16_type_node for BT_UINT16.
37664 2017-01-27 David Malcolm <dmalcolm@redhat.com>
37666 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
37667 "RTL Tests" to menu.
37668 (GIMPLE Tests): New node.
37669 (RTL Tests): New node.
37671 2017-01-27 Richard Biener <rguenther@suse.de>
37673 PR tree-optimization/79245
37674 * tree-loop-distribution.c (distribute_loop): Apply cost
37675 modeling also to detected patterns.
37677 2017-01-27 Richard Biener <rguenther@suse.de>
37679 PR tree-optimization/71433
37680 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
37681 (compare_assert_loc): New function.
37682 (process_assert_insertions): Sort and optimize assert locations
37683 to remove duplicates and push down identical assertions on
37684 edges to their destination block.
37686 2017-01-27 Richard Biener <rguenther@suse.de>
37688 PR tree-optimization/79244
37689 * tree-vrp.c (remove_range_assertions): Forcefully propagate
37690 out SSA names even if abnormal.
37692 2017-01-27 Jakub Jelinek <jakub@redhat.com>
37694 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
37695 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
37696 instead of MPFR_RNDN.
37698 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
37701 * arm.c (arm_option_override): Don't call build_target_option_node
37702 until after doing all option overrides.
37703 (arm_valid_target_attribute_tree): Likewise.
37705 2017-01-27 Martin Liska <mliska@suse.cz>
37707 * doc/invoke.texi (-fprofile-arcs): Document profiling support
37708 for {cd}tors and C++ {cd}tors.
37710 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
37712 * config/s390/s390.md ("*setmem_long_and")
37713 ("*setmem_long_and_31z"): Use zero_extend instead of and.
37715 2017-01-26 Martin Sebor <msebor@redhat.com>
37717 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
37720 2017-01-26 Martin Sebor <msebor@redhat.com>
37722 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
37723 HAVE_DFmode before using XFmode or DFmode.
37724 (parse_directive): Avoid using the z length modifier to avoid
37725 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
37727 PR middle-end/78703
37728 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
37729 to accept adjustment as an array.
37730 (get_int_range): New function.
37731 (struct directive): Make width and prec arrays.
37732 (directive::set_width, directive::set_precision): Call get_int_range.
37733 (format_integer, format_floating): Handle width and precision ranges.
37734 (format_string, parse_directive): Same.
37736 2017-01-26 Jakub Jelinek <jakub@redhat.com>
37739 * dwarf2out.c (generate_skeleton_bottom_up): For children with
37740 comdat_type_p set, just clone them, but keep the children in the
37744 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
37745 which have direct callers with -fvar-tracking-assignments enabled
37747 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
37748 inside of type units.
37750 2017-01-26 Martin Sebor <msebor@redhat.com>
37752 PR middle-end/78703
37753 * gimple-ssa-sprintf.c (struct result_range): Add likely and
37755 (struct format_result): Replace number_chars, number_chars_min,
37756 and number_chars_max with a single member of struct result_range.
37758 (format_result::operator+=): Adjust.
37759 (struct fmtresult): Remove bounded. Handle likely and unlikely
37761 (fmtresult::adjust_for_width_or_precision): New function.
37762 (fmtresult:type_max_digits): New function.
37763 (bytes_remaining): Handle likely and unlikely counters.
37764 (min_bytes_remaining): Remove.
37765 (format_percent): Simplify.
37766 (format_integer, format_floating): Set likely and unlikely counters.
37767 (get_string_length, format_character, format_string): Same.
37768 (format_plain, should_warn_p): New function.
37769 (maybe_warn): Call should_warn_p. Update diagnostic messages
37770 and handle those for all directives, including plain strings.
37771 (format_directive): Handle likely and unlikely counters.
37772 Remove unnecessary quoting from diagnostics. Add an informational
37774 (add_bytes): Remove.
37775 (pass_sprintf_length::compute_format_length): Simplify.
37776 (try_substitute_return_value): Handle likely and unlikely counters.
37778 2017-01-26 Carl Love <cel@us.ibm.com>
37780 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
37781 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
37783 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
37786 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
37787 endianess for subregs into account.
37788 * lra-constraints.c (lra_constraints): Do risky transformations
37789 always on the first iteration.
37790 * lra-lives.c (check_pseudos_live_through_calls): Add arg
37791 last_call_used_reg_set.
37792 (process_bb_lives): Define and use last_call_used_reg_set.
37793 * lra.c (lra): Always continue after lra_constraints on the first
37796 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
37798 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
37800 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
37802 2017-01-26 Jakub Jelinek <jakub@redhat.com>
37804 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
37805 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
37806 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
37807 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
37808 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
37809 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
37810 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
37811 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
37812 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
37814 2017-01-26 Marek Polacek <polacek@redhat.com>
37817 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
37818 for the third operand.
37820 2017-01-26 Jakub Jelinek <jakub@redhat.com>
37822 PR middle-end/79236
37823 * omp-low.c (struct omp_context): Add simt_stmt field.
37824 (scan_omp_for): Return omp_context *.
37825 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
37826 context to the _simt_ SIMD stmt.
37827 (lower_omp_for): For combined SIMD with sibling _simt_
37828 SIMD, make sure to use the same decls in _looptemp_
37829 clauses as in the sibling.
37831 2017-01-26 David Sherwood <david.sherwood@arm.com>
37833 PR middle-end/79212
37834 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
37837 2017-01-26 Jakub Jelinek <jakub@redhat.com>
37840 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
37841 emit fld b; fld a; if possible.
37843 * brig-builtins.def: Update copyright years.
37844 * config/arm/arm_acle_builtins.def: Update copyright years.
37846 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
37849 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
37850 constraint instead of o for the stxsd instruction.
37852 2017-01-25 Carl Love <cel@us.ibm.com>
37854 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
37855 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
37857 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
37859 * doc/invoke.texi (C++ Dialect Options): Fix typo.
37861 2017-01-25 Richard Biener <rguenther@suse.de>
37863 PR tree-optimization/69264
37864 * target.def (vector_alignment_reachable): Improve documentation.
37865 * doc/tm.texi: Regenerate.
37866 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
37868 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
37869 earlier changes with respect to TYPE_USER_ALIGN.
37870 (vector_alignment_reachable_p): Likewise. Improve dumping.
37872 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37875 * config/arm/arm.md (xordi3): Force constant operand into a register
37878 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37880 * doc/invoke.texi (-fstore-merging): Correct default optimization
37881 levels at which it is enabled.
37882 (-O): Move -fstore-merging from list to...
37885 2017-01-25 Richard Biener <rguenther@suse.de>
37888 * omp-expand.c: Include debug.h.
37889 (expand_omp_taskreg): Make sure to generate early debug before
37890 outlining anything from a function.
37891 (expand_omp_target): Likewise.
37892 (grid_expand_target_grid_body): Likewise.
37894 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
37897 * asan.c (get_translation_unit_decl): New function.
37898 (asan_add_global): Extract modules file name from globals
37899 TRANSLATION_UNIT_DECL name.
37901 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
37904 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
37905 for long calls with APCS frame and VFP.
37907 2017-01-24 David Malcolm <dmalcolm@redhat.com>
37909 * cfg.c (original_copy_tables_initialized_p): New function.
37910 * cfg.h (original_copy_tables_initialized_p): New decl.
37911 * cfgrtl.c (relink_block_chain): Guard the call to
37912 free_original_copy_tables with a call to
37913 original_copy_tables_initialized_p.
37914 * cgraph.h (symtab_node::native_rtl_p): New decl.
37915 * cgraphunit.c (symtab_node::native_rtl_p): New function.
37916 (symtab_node::needed_p): Don't assert for early assembly output
37917 for __RTL functions.
37918 (cgraph_node::finalize_function): Set "force_output" for __RTL
37920 (cgraph_node::analyze): Bail out early for __RTL functions.
37921 (analyze_functions): Update assertion to support __RTL functions.
37922 (cgraph_node::expand): Bail out early for __RTL functions.
37923 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
37925 * function.h (struct function): Update comment for field
37927 * gimple-expr.c: Include "tree-pass.h".
37928 (gimple_has_body_p): Return false for __RTL functions.
37929 * Makefile.in (OBJS): Add run-rtl-passes.o.
37930 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
37932 (gcc::pass_manager::get_clean_slate): New accessor.
37933 * passes.c: Include "insn-addr.h".
37934 (should_skip_pass_p): Add logging. Update logic for running
37935 "expand" to be compatible with both __GIMPLE and __RTL. Guard
37936 property-provider override so it is only done for gimple passes.
37938 (skip_pass): New function.
37939 (execute_one_pass): Call skip_pass when skipping passes.
37940 * read-md.c (md_reader::read_char): Support filtering
37941 the input to a subset of line numbers.
37942 (md_reader::md_reader): Initialize fields
37943 m_first_line and m_last_line.
37944 (md_reader::read_file_fragment): New function.
37945 * read-md.h (md_reader::read_file_fragment): New decl.
37946 (md_reader::m_first_line): New field.
37947 (md_reader::m_last_line): New field.
37948 * read-rtl-function.c (function_reader::create_function): Only
37949 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
37950 curr_properties. Set DECL_INITIAL to a dummy block.
37951 (read_rtl_function_body_from_file_range): New function.
37952 * read-rtl-function.h (read_rtl_function_body_from_file_range):
37954 * run-rtl-passes.c: New file.
37955 * run-rtl-passes.h: New file.
37957 2017-01-24 Jeff Law <law@redhat.com>
37959 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
37962 2017-01-24 Bin Cheng <bin.cheng@arm.com>
37964 PR tree-optimization/79159
37965 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
37966 (record_nonwrapping_iv): Improve boundary using above function if no
37967 value range information.
37969 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
37970 Martin Jambor <mjambor@suse.cz>
37972 * brig-builtins.def: New file.
37973 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
37974 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
37975 (DEF_HSAIL_SAT_BUILTIN): Likewise.
37976 (DEF_HSAIL_INTR_BUILTIN): Likewise.
37977 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
37978 * builtin-types.def (BT_INT8): New.
37979 (BT_INT16): Likewise.
37980 (BT_UINT8): Likewise.
37981 (BT_UINT16): Likewise.
37982 (BT_FN_ULONG): Likewise.
37983 (BT_FN_UINT_INT): Likewise.
37984 (BT_FN_UINT_ULONG): Likewise.
37985 (BT_FN_UINT_LONG): Likewise.
37986 (BT_FN_UINT_PTR): Likewise.
37987 (BT_FN_ULONG_PTR): Likewise.
37988 (BT_FN_INT8_FLOAT): Likewise.
37989 (BT_FN_INT16_FLOAT): Likewise.
37990 (BT_FN_UINT32_FLOAT): Likewise.
37991 (BT_FN_UINT16_FLOAT): Likewise.
37992 (BT_FN_UINT8_FLOAT): Likewise.
37993 (BT_FN_UINT64_FLOAT): Likewise.
37994 (BT_FN_UINT16_UINT32): Likewise.
37995 (BT_FN_UINT32_UINT16): Likewise.
37996 (BT_FN_UINT16_UINT16_UINT16): Likewise.
37997 (BT_FN_INT_PTR_INT): Likewise.
37998 (BT_FN_UINT_PTR_UINT): Likewise.
37999 (BT_FN_LONG_PTR_LONG): Likewise.
38000 (BT_FN_ULONG_PTR_ULONG): Likewise.
38001 (BT_FN_VOID_UINT64_UINT64): Likewise.
38002 (BT_FN_UINT8_UINT8_UINT8): Likewise.
38003 (BT_FN_INT8_INT8_INT8): Likewise.
38004 (BT_FN_INT16_INT16_INT16): Likewise.
38005 (BT_FN_INT_INT_INT): Likewise.
38006 (BT_FN_UINT_FLOAT_UINT): Likewise.
38007 (BT_FN_FLOAT_UINT_UINT): Likewise.
38008 (BT_FN_ULONG_UINT_UINT): Likewise.
38009 (BT_FN_ULONG_UINT_PTR): Likewise.
38010 (BT_FN_ULONG_ULONG_ULONG): Likewise.
38011 (BT_FN_UINT_UINT_UINT): Likewise.
38012 (BT_FN_VOID_UINT_PTR): Likewise.
38013 (BT_FN_UINT_UINT_PTR: Likewise.
38014 (BT_FN_UINT32_UINT64_PTR): Likewise.
38015 (BT_FN_INT_INT_UINT_UINT): Likewise.
38016 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
38017 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
38018 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
38019 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
38020 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
38021 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
38022 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
38023 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
38024 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
38025 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
38026 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
38027 * doc/frontends.texi: List BRIG FE.
38028 * doc/install.texi (Testing): Add BRIG tesring requirements.
38029 * doc/invoke.texi (Overall Options): Mention BRIG.
38030 * doc/standards.texi (Standards): Doucment BRIG HSA version.
38032 2017-01-24 Richard Biener <rguenther@suse.de>
38034 PR translation/79208
38035 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
38037 2017-01-24 Martin Jambor <mjambor@suse.cz>
38040 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
38041 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
38042 and known_contexts.
38044 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
38046 PR middle-end/79123
38047 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
38048 casts from signed to unsigned really don't have a range.
38050 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
38052 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
38053 GMP_RNDx for compatiblity.
38055 2017-01-24 Martin Liska <mliska@suse.cz>
38058 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
38059 that would prevent us to call alloca with -1 as argument.
38061 2017-01-24 Jakub Jelinek <jakub@redhat.com>
38063 * dwarf2out.c (output_compilation_unit_header, output_file_names):
38064 Avoid -Wformat-security warning.
38066 2017-01-23 Andrew Pinski <apinski@cavium.com>
38068 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
38071 2017-01-23 Martin Sebor <msebor@redhat.com>
38073 PR middle-end/78703
38074 * gimple-ssa-sprintf.c (warn_level): New global.
38075 (format_integer): Use it here and throughout the rest of the file.
38076 Use the same switch to compute sign as base.
38077 (maybe_warn): New function.
38078 (format_directive): Factor out warnings into maybe_warn.
38079 Add debugging output. Use warn_level.
38080 (add_bytes): Use warn_level.
38081 (pass_sprintf_length::compute_format_length): Add debugging output.
38082 (try_substitute_return_value): Same.
38083 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
38085 PR middle-end/78703
38086 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
38087 (struct fmtresult, format_integer, format_floating): Adjust.
38088 (fmtresult::fmtresult): Set max correctly in two argument ctor.
38089 (get_string_length, format_string,format_directive): Same.
38090 (pass_sprintf_length::compute_format_length): Same.
38091 (try_substitute_return_value): Simplify slightly.
38093 PR middle-end/78703
38094 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
38095 (fmtresult::operator+=): Outlined.
38096 (struct fmtresult): Add ctors.
38097 (struct conversion_spec): Rename...
38098 (struct directive): ...to this. Add and remove data members.
38099 (directive::set_width, directive::set_precision): New functions.
38100 (format_percent): Use fmtresult ctor.
38101 (get_width_and_precision): Remove.
38102 (format_integer): Make naming changes. Avoid computing width and
38104 (format_floating): Same. Adjust indentation.
38105 (format_character, format_none): New functions.
38106 (format_string): Moved character handling to format_character.
38107 (format_directive): Remove arguments, change return type.
38108 (parse_directive): New function.
38109 (pass_sprintf_length::compute_format_length): Move directive
38110 parsing to parse_directive.
38112 2017-01-23 Jakub Jelinek <jakub@redhat.com>
38114 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
38115 (assign_assembler_name_if_needed): ... this.
38116 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
38117 (assign_assembler_name_if_needed): ... this.
38118 (free_lang_data_in_cgraph): Adjust callers.
38119 * cgraphunit.c (cgraph_node::analyze): Likewise.
38120 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
38123 2017-01-23 Richard Biener <rguenther@suse.de>
38125 PR tree-optimization/79088
38126 PR tree-optimization/79188
38127 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
38128 resetting loop bounds after last path deletion. Reset loop
38129 bounds of the target loop, make code match the comments.
38130 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
38131 Make sure loops need no fixups.
38133 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
38135 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
38136 exponent support with double type for first argument.
38137 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
38138 type returned by __builtin_vec_extract_sig,
38139 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
38140 functions from "vector int" to "vector unsigned int" or from
38141 "vector long long int" to "vector unsigned long long int".
38142 Changed type returned by __builtin_vec_extract_exp,
38143 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
38144 functions from "vector int" to "vector unsigned int" or from
38145 "vector long long int" to "vector unsigned long long int".
38146 Changed return type of __builtin_vec_test_data_class,
38147 __builtin_vec_test_data_class_sp, and
38148 __builtin_vec_test_data_class_dp from "vector int" to
38149 "vector bool int" or from "vector long long int" to "vector bool
38150 long long int" and changed second argument type from "unsigned
38151 int" to "int". Added new overloaded function forms "vector float
38152 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
38153 "vector float __builtin_vec_insert_exp_sp (vector float, vector
38154 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
38155 double, vector unsigned long long int)" and "vector double
38156 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
38157 long int)". Changed return type of
38158 __builtin_scalar_test_data_class and
38159 __builtin_scalar_test_data_class_sp and
38160 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
38161 int" and changed second argument from "unsigned int" to "int".
38162 Changed type returned by __builtin_scalar_test_neg,
38163 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
38164 from "int" to "bool int". Added new overloaded function form
38165 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
38166 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
38167 exponent double-precision with floating point first argument.
38168 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
38169 documentation of scalar_test_data_class, scalar_test_neg,
38170 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
38171 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
38172 vec_test_data_class built-in functions to reflect refinements in
38173 their type signatures.
38175 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
38177 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
38179 (aarch64_elf_asm_destructor): Likewise.
38181 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
38183 PR rtl-optimization/78634
38184 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
38185 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
38186 * ifcvt.c (noce_try_cmove): Add missing cost check.
38188 PR rtl-optimization/71724
38189 * combine.c (if_then_else_cond): Look for situations where it is
38190 beneficial to undo the work of one of the recursive calls.
38192 2017-01-23 Bin Cheng <bin.cheng@arm.com>
38194 PR tree-optimization/70754
38195 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
38196 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
38197 combined stmt before it if not NULL.
38198 (combine_chains): Process refs reversely and compute dominance point
38201 2017-01-23 Martin Liska <mliska@suse.cz>
38203 PR tree-optimization/79196
38204 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
38205 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
38207 (strlen_optimize_stmt): Call the renamed function.
38209 2017-01-23 Michael Matz <matz@suse.de>
38211 PR tree-optimization/78384
38212 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
38214 2017-01-23 Richard Biener <rguenther@suse.de>
38216 PR tree-optimization/79186
38217 * tree-vrp.c (register_new_assert_for): Make sure we've seen
38218 both incoming edges before moving an assert.
38220 2017-01-23 Martin Jambor <mjambor@suse.cz>
38222 * ipa-prop.c (load_from_param_1): Removed.
38223 (load_from_unmodified_param): Bits from load_from_param_1 put back
38225 (load_from_param): Removed.
38226 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
38227 with stmt. Reverted back to use of load_from_unmodified_param.
38229 2017-01-23 Martin Jambor <mjambor@suse.cz>
38232 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
38233 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
38234 field a pointer to garbage collected vector, mark lattices and
38235 ipcp_orig_node with GTY((skip)).
38236 (ipa_get_param_count): Adjust to descriptors being a pointer.
38237 (ipa_get_param): Likewise.
38238 (ipa_get_type): Likewise.
38239 (ipa_get_param_move_cost): Likewise.
38240 (ipa_set_param_used): Likewise.
38241 (ipa_get_controlled_uses): Likewise.
38242 (ipa_set_controlled_uses): Likewise.
38243 (ipa_is_param_used): Likewise.
38244 (ipa_node_params_t): Move into garbage collector. New methods insert
38246 (ipa_node_params_sum): Annotate wth GTY(()).
38247 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
38249 (ipa_load_from_parm_agg): Adjust declaration.
38250 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
38251 * ipa-profile.c (ipa_profile): Likewise.
38252 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
38253 (ipa_populate_param_decls): Make descriptors parameter garbage
38255 (ipa_dump_param): Adjust to descriptors being a pointer.
38256 (ipa_alloc_node_params): Likewise.
38257 (ipa_initialize_node_params): Likewise.
38258 (load_from_param_1): Make descriptors parameter garbage collected.
38259 (load_from_unmodified_param): Likewise.
38260 (load_from_param): Likewise.
38261 (ipa_load_from_parm_agg): Likewise.
38262 (ipa_node_params::~ipa_node_params): Removed.
38263 (ipa_free_all_node_params): Remove call to delete operator.
38264 (ipa_node_params_t::insert): New.
38265 (ipa_node_params_t::remove): Likewise.
38266 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
38267 copy known_csts and known_contexts vectors.
38268 (ipa_read_node_info): Adjust to descriptors being a pointer.
38269 (ipcp_modif_dom_walker): Make m_descriptors field garbage
38271 (ipcp_transform_function): Make descriptors variable garbage
38274 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
38276 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
38277 * config/i386/avx512dqintrin.h: Ditto.
38278 * config/i386/avx512fintrin.h: Ditto.
38279 * config/i386/i386.c: Handle new builtins.
38280 * config/i386/i386-builtin.def: Add new builtins.
38281 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
38282 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
38284 2017-01-23 Jakub Jelinek <jakub@redhat.com>
38285 Martin Liska <mliska@suse.cz>
38287 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
38288 * asan.c (asan_expand_poison_ifn): Support stores and use
38289 appropriate ASAN report function.
38290 * internal-fn.c (expand_ASAN_POISON_USE): New function.
38291 * internal-fn.def (ASAN_POISON_USE): Declare.
38292 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
38293 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
38294 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
38295 ASAN_POISON calls w/o LHS.
38296 * tree-ssa.c (execute_update_addresses_taken): Create clobber
38297 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
38298 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
38299 * gimplify.c (asan_poison_variables): Add attribute
38300 use_after_scope_memory to variables that really needs to live
38302 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
38303 having the attribute.
38305 2017-01-23 Martin Liska <mliska@suse.cz>
38307 * asan.c (create_asan_shadow_var): New function.
38308 (asan_expand_poison_ifn): Likewise.
38309 * asan.h (asan_expand_poison_ifn): New declaration.
38310 * internal-fn.c (expand_ASAN_POISON): Likewise.
38311 * internal-fn.def (ASAN_POISON): New builtin.
38312 * sanopt.c (pass_sanopt::execute): Expand
38313 asan_expand_poison_ifn.
38314 * tree-inline.c (copy_decl_for_dup_finish): Make function
38316 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
38317 * tree-ssa.c (is_asan_mark_p): New function.
38318 (execute_update_addresses_taken): Rewrite local variables
38319 (identified just by use-after-scope as addressable) into SSA.
38321 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
38323 * doc/install.texi (Specific): opensource.apple.com uses https
38324 now. Remove trailing slash.
38326 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
38328 * README.Portability: Remove note on an Irix compatibility issue.
38330 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
38332 * gcov.c (INCLUDE_ALGORITHM): Define.
38333 (INCLUDE_VECTOR): Define.
38334 No longer include <vector> and <algorithm> directly.
38336 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
38338 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
38340 * doc/invoke.texi (Code Gen Options): Ditto.
38342 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
38345 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
38347 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
38349 rtl-optimization/79125
38350 * cprop.c (local_cprop_pass): Handle cases where we make an
38351 unconditional trap.
38353 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
38357 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
38358 read from, for big endian.
38360 2017-01-20 Jiong Wang <jiong.wang@arm.com>
38362 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
38363 register pauth builtins for LP64 only.
38365 2017-01-20 Marek Polacek <polacek@redhat.com>
38368 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
38371 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
38373 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
38375 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
38376 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
38377 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
38379 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38382 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
38383 in big-endian mode when they are not a single duplicated value.
38385 2017-01-20 Richard Biener <rguenther@suse.de>
38387 * BASE-VER: Bump to 7.0.1.
38389 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
38391 * omp-low.c (omplow_simd_context): New struct. Use it...
38392 (lower_rec_simd_input_clauses): ...here and...
38393 (lower_rec_input_clauses): ...here to hold common data. Adjust all
38394 references to idx, lane, max_vf, is_simt.
38396 2017-01-20 Graham Markall <graham.markall@embecosm.com>
38398 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
38401 2017-01-20 Martin Jambor <mjambor@suse.cz>
38403 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
38404 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
38406 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
38407 (GTFILES): Rename hsa.c to hsa-common.c.
38408 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
38409 * hsa-dump.c: Likewise.
38410 * hsa-gen.c: Likewise.
38411 * hsa-regalloc.c: Likewise.
38412 * ipa-hsa.c: Likewise.
38413 * omp-expand.c: Likewise.
38414 * omp-low.c: Likewise.
38415 * toplev.c: Likewise.
38417 2017-01-20 Marek Polacek <polacek@redhat.com>
38420 * doc/invoke.texi: Document -Wduplicated-branches.
38421 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
38422 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
38423 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
38424 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
38425 return 0 only when not OEP_LEXICOGRAPHIC.
38426 (fold_build_cleanup_point_expr): Use the expression
38427 location when building CLEANUP_POINT_EXPR.
38428 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
38429 * tree.c (add_expr): Handle error_mark_node.
38431 2017-01-20 Martin Liska <mliska@suse.cz>
38434 * tree-profile.c (init_ic_make_global_vars): Do not call
38436 (gimple_init_gcov_profiler): Likewise.
38438 2017-01-20 Martin Liska <mliska@suse.cz>
38441 * cgraph.h (maybe_create_reference): Remove argument and
38443 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
38445 * ipa-cp.c (create_specialized_node): Likewise.
38446 * symtab.c (symtab_node::maybe_create_reference): Handle
38447 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
38449 2017-01-20 Martin Liska <mliska@suse.cz>
38451 * read-rtl-function.c (function_reader::create_function): Use
38452 build_decl instread of build_decl_stat.
38454 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
38456 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
38457 * config/i386/avx512dqintrin.h: Ditto.
38458 * config/i386/avx512fintrin.h: Ditto.
38459 * config/i386/i386-builtin-types.def: Add new types.
38460 * config/i386/i386.c: Handle new types.
38461 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
38462 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
38463 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
38464 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
38465 (__builtin_ia32_kshiftridi): New.
38466 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
38468 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
38472 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
38473 define to rs6000_init_stack_protect_guard.
38474 (rs6000_init_stack_protect_guard): New function.
38476 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
38477 Yunqiang Su <yunqiang.su@imgtec.com>
38479 * config.gcc (supported_defaults): Add madd4.
38480 (with_madd4): Add validation.
38481 (all_defaults): Add madd4.
38482 * config/mips/mips.opt (mmadd4): New option.
38483 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
38485 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
38487 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
38488 (ISA_HAS_FUSED_MADD4): Likewise.
38489 * doc/invoke.texi (-mmadd4): Document the new option.
38490 * doc/install.texi (--with-madd4): Document the new option.
38492 2017-01-19 Jiong Wang <jiong.wang@arm.com>
38494 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
38495 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
38496 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
38497 (aarch64_init_pauth_hint_builtins): New.
38498 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
38499 (aarch64_expand_builtin): Expand new builtins.
38501 2017-01-19 Jiong Wang <jiong.wang@arm.com>
38503 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
38504 * combine-stack-adj.c (no_unhandled_cfa): Handle
38505 REG_CFA_TOGGLE_RA_MANGLE.
38506 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
38507 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
38508 info for return address signing.
38509 (aarch64_expand_epilogue): Likewise.
38511 2017-01-19 Jiong Wang <jiong.wang@arm.com>
38513 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
38514 * config/aarch64/aarch64-protos.h
38515 (aarch64_return_address_signing_enabled): New declaration.
38516 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
38518 (aarch64_expand_prologue): Sign return address before it's pushed onto
38520 (aarch64_expand_epilogue): Authenticate return address fetched from
38522 (aarch64_override_options): Sanity check for ILP32 and ISA level.
38523 (aarch64_attributes): New function attributes for "sign-return-address".
38524 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
38525 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
38526 ("*do_return"): Generate combined instructions according to key index.
38527 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
38528 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
38530 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
38531 * config/aarch64/aarch64.opt (msign-return-address=): New.
38532 * doc/extend.texi (AArch64 Function Attributes): Documents
38533 "sign-return-address=".
38534 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
38536 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
38538 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
38539 overall option summary.
38541 2017-01-19 Jiong Wang <jiong.wang@arm.com>
38543 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
38544 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
38545 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
38546 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
38548 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
38550 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
38551 -mpower9-minmax by default for -mcpu=power9.
38552 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
38553 128-bit floating point.
38555 2017-01-20 Alan Modra <amodra@gmail.com>
38557 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
38558 optimizing for size.
38560 2017-01-20 Alan Modra <amodra@gmail.com>
38563 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
38564 for strcmp and strncmp from corresponding builtin decl.
38566 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
38568 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
38569 instead of i386/rtems-64.h.
38570 * config/i386/rtems-64.h: Remove.
38572 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
38576 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
38578 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
38580 2017-01-19 Tamar Christina <tamar.christina@arm.com>
38582 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
38583 Change int to HOST_WIDE_INT.
38584 * config/aarch64/aarch64-protos.h
38585 (aarch64_simd_gen_const_vector_dup): Likewise.
38586 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
38588 2017-01-19 David Malcolm <dmalcolm@redhat.com>
38590 * langhooks-def.h (lhd_type_for_size): New decl.
38591 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
38592 * langhooks.c (lhd_type_for_size): New function, taken from
38595 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
38597 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
38598 define_bypass for CR latency.
38599 (power9-cracked-alu): Update bypass latency and remove power9-branch.
38600 (power9-alu2): Add define_bypass for CR latency.
38602 (power9-mul): Update insn latency.
38603 (power9-mul-compare): Update insn latency, bypass latency and remove
38606 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38608 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
38610 * config/aarch64/aarch64.md
38611 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
38612 aarch64_nopcrelative_literal_loads.
38613 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
38615 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
38617 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
38618 TARGET_LOONGSON_3A.
38619 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
38621 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
38624 * config.gcc (supported_defaults): Add lxc1-sxc1.
38625 (with_lxc1_sxc1): Add validation.
38626 (all_defaults): Add lxc1-sxc1.
38627 * config/mips/mips.opt (mlxc1-sxc1): New option.
38628 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
38630 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
38631 __mips_no_lxc1_sxc1.
38632 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
38633 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
38634 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
38636 2017-01-19 Richard Biener <rguenther@suse.de>
38638 PR tree-optimization/72488
38639 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
38640 sure to restore SSA info.
38641 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
38643 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
38645 PR rtl-optimization/79121
38646 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
38647 of the inner type when shifting an extended value.
38649 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
38652 * symtab.c (symtab_node::equal_address_to): Fix comparing of
38653 interposable aliases.
38655 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
38658 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
38659 Use the evmergelohi instruction.
38660 (mov_si<mode>_e500_subreg4_2_le): Likewise.
38661 (mov_sitf_e500_subreg8_2_be): Likewise.
38662 (mov_sitf_e500_subreg12_2_le): Likewise.
38663 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
38664 (mov_si<mode>_e500_subreg4_2_be): Likewise.
38665 (mov_sitf_e500_subreg8_2_le): Likewise.
38666 (mov_sitf_e500_subreg12_2_be): Likewise.
38668 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38670 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
38671 attribute from vecsimple to vecperm.
38672 (altivec_vbpermq2): Likewise.
38674 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38677 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
38679 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38680 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
38681 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
38682 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
38683 case where N arg is SIZE_MAX.
38684 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
38685 (cmpstrsi): Add pattern.
38687 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
38689 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
38690 __builtin_vec_revb builtins.
38691 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
38692 built-in functions to support generation of the ISA 3.0 XXBR<x>
38693 vector byte reverse instructions.
38694 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
38695 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
38696 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
38697 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
38698 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
38699 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
38700 (P9V_BUILTIN_VEC_REVB): Likewise.
38701 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
38702 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
38703 (p9_xxbrq_v16qi): Likewise.
38704 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
38705 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
38706 (p9_xxbrh_v8hi): Likewise.
38707 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
38708 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
38709 vec_revb built-in functions.
38711 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
38713 PR rtl-optimization/78952
38714 * config/i386/i386.md (any_extract): New code iterator.
38715 (*insvqi_2): Use any_extract for source operand.
38716 (*insvqi_3): Use any_shiftrt for source operand.
38718 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
38720 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
38722 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
38724 2017-01-18 Matthias Klose <doko@ubuntu.com>
38726 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
38728 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38730 * config/rs6000/altivec.h (vec_bperm): Change #define.
38731 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
38732 (altivec_vbpermq2): New define_insn.
38733 (altivec_vbpermd): Likewise.
38734 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
38735 function interface.
38736 (VBPERMD): Likewise.
38737 (VBPERM): New polymorphic function interface.
38738 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
38739 Add entries for P9V_BUILTIN_VEC_VBPERM.
38740 * doc/extend.texi: Add interfaces for vec_bperm.
38742 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38744 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
38745 first letter of error messages.
38746 (s390_resolve_overloaded_builtin): Likewise.
38747 * config/s390/s390.c (s390_expand_builtin): Likewise.
38748 (s390_invalid_arg_for_unprototyped_fn): Likewise.
38749 (s390_valid_target_attribute_inner_p): Likewise.
38750 * config/s390/s390.md ("tabort"): Likewise.
38752 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
38754 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
38755 (ISA_AVOID_DIV_HILO): New macro.
38756 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
38757 (ISA_HAS_DDIV): Likewise.
38759 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
38761 * doc/invoke.texi (fabi-version): Correct number of occurrences.
38763 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
38765 * doc/invoke.texi (fabi-version): Spelling fix.
38767 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
38770 * doc/invoke.texi (fabi-version): Mention mangling fix for
38773 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
38776 * doc/invoke.texi (fabi-version): Document discriminator mangling.
38778 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
38781 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
38782 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
38784 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
38786 (stack_protect_test): Ditto.
38787 * config/rs6000/rs6000.opt (mstack-protector-guard=,
38788 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
38790 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
38791 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
38792 -mstack-protector-guard-offset=.
38793 (RS/6000 and PowerPC Options): Ditto.
38795 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
38797 * config/i386/i386.h (MASK_CLASS_P): New define.
38798 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
38799 there are no registers from different register sets also when
38800 mask registers are used. Update function comment.
38801 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
38802 to (*k/*r) and (*k/*km) alternatives.
38804 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
38806 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
38807 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
38808 (EH_RETURN_HANDLER_RTX): New define.
38809 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
38810 Force frame pointer in EH return functions.
38811 (aarch64_expand_epilogue): Add barrier for eh_return.
38812 (aarch64_final_eh_return_addr): Remove.
38813 (aarch64_eh_return_handler_rtx): New function.
38814 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
38816 (aarch64_eh_return_handler_rtx): New prototype.
38818 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38820 * config/rs6000/altivec.h (vec_rlmi): New #define.
38821 (vec_vrlnm): Likewise.
38822 (vec_rlnm): Likewise.
38823 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
38824 (UNSPEC_VRLNM): Likewise.
38825 (VIlong): New mode iterator.
38826 (altivec_vrl<VI_char>mi): New define_insn.
38827 (altivec_vrl<VI_char>nm): Likewise.
38828 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
38830 (VRLDNM): Likewise.
38831 (RLNM): New polymorphic function entry.
38832 (VRLWMI): New monomorphic function entry.
38833 (VRLDMI): Likewise.
38834 (RLMI): New polymorphic function entry.
38835 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
38836 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
38837 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
38840 2017-01-17 Jakub Jelinek <jakub@redhat.com>
38843 * dwarf2out.c (field_byte_offset): Restore the
38844 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
38845 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
38846 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
38849 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
38852 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
38854 2017-01-17 Jakub Jelinek <jakub@redhat.com>
38857 * dwarf2out.c (add_data_member_location_attribute): For constant
38858 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
38859 instead of DW_AT_data_member_location, DW_AT_bit_offset and
38860 DW_AT_byte_size attributes.
38862 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
38864 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
38865 after forcing to constant memory when the code model is medium.
38867 2017-01-17 Julia Koval <julia.koval@intel.com>
38870 * config/i386/avx512fintrin.h
38871 (_mm512_i32gather_ps): Change __addr type to void const*.
38872 (_mm512_mask_i32gather_ps): Ditto.
38873 (_mm512_i32gather_pd): Ditto.
38874 (_mm512_mask_i32gather_pd): Ditto.
38875 (_mm512_i64gather_ps): Ditto.
38876 (_mm512_mask_i64gather_ps): Ditto.
38877 (_mm512_i64gather_pd): Ditto.
38878 (_mm512_mask_i64gather_pd): Ditto.
38879 (_mm512_i32gather_epi32): Ditto.
38880 (_mm512_mask_i32gather_epi32): Ditto.
38881 (_mm512_i32gather_epi64): Ditto.
38882 (_mm512_mask_i32gather_epi64): Ditto.
38883 (_mm512_i64gather_epi32): Ditto.
38884 (_mm512_mask_i64gather_epi32): Ditto.
38885 (_mm512_i64gather_epi64): Ditto.
38886 (_mm512_mask_i64gather_epi64): Ditto.
38887 (_mm512_i32scatter_ps): Change __addr type to void*.
38888 (_mm512_mask_i32scatter_ps): Ditto.
38889 (_mm512_i32scatter_pd): Ditto.
38890 (_mm512_mask_i32scatter_pd): Ditto.
38891 (_mm512_i64scatter_ps): Ditto.
38892 (_mm512_mask_i64scatter_ps): Ditto.
38893 (_mm512_i64scatter_pd): Ditto.
38894 (_mm512_mask_i64scatter_pd): Ditto.
38895 (_mm512_i32scatter_epi32): Ditto.
38896 (_mm512_mask_i32scatter_epi32): Ditto.
38897 (_mm512_i32scatter_epi64): Ditto.
38898 (_mm512_mask_i32scatter_epi64): Ditto.
38899 (_mm512_i64scatter_epi32): Ditto.
38900 (_mm512_mask_i64scatter_epi32): Ditto.
38901 (_mm512_i64scatter_epi64): Ditto.
38902 (_mm512_mask_i64scatter_epi64): Ditto.
38903 * config/i386/avx512pfintrin.h
38904 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
38905 (_mm512_mask_prefetch_i32gather_ps): Ditto.
38906 (_mm512_mask_prefetch_i64gather_pd): Ditto.
38907 (_mm512_mask_prefetch_i64gather_ps): Ditto.
38908 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
38909 (_mm512_prefetch_i32scatter_ps): Ditto.
38910 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
38911 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
38912 (_mm512_prefetch_i64scatter_pd): Ditto.
38913 (_mm512_prefetch_i64scatter_ps): Ditto.
38914 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
38915 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
38916 * config/i386/avx512vlintrin.h
38917 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
38918 (_mm_mmask_i32gather_ps): Ditto.
38919 (_mm256_mmask_i32gather_pd): Ditto.
38920 (_mm_mmask_i32gather_pd): Ditto.
38921 (_mm256_mmask_i64gather_ps): Ditto.
38922 (_mm_mmask_i64gather_ps): Ditto.
38923 (_mm256_mmask_i64gather_pd): Ditto.
38924 (_mm_mmask_i64gather_pd): Ditto.
38925 (_mm256_mmask_i32gather_epi32): Ditto.
38926 (_mm_mmask_i32gather_epi32): Ditto.
38927 (_mm256_mmask_i32gather_epi64): Ditto.
38928 (_mm_mmask_i32gather_epi64): Ditto.
38929 (_mm256_mmask_i64gather_epi32): Ditto.
38930 (_mm_mmask_i64gather_epi32): Ditto.
38931 (_mm256_mmask_i64gather_epi64): Ditto.
38932 (_mm_mmask_i64gather_epi64): Ditto.
38933 (_mm256_i32scatter_ps): Change __addr type to void*.
38934 (_mm256_mask_i32scatter_ps): Ditto.
38935 (_mm_i32scatter_ps): Ditto.
38936 (_mm_mask_i32scatter_ps): Ditto.
38937 (_mm256_i32scatter_pd): Ditto.
38938 (_mm256_mask_i32scatter_pd): Ditto.
38939 (_mm_i32scatter_pd): Ditto.
38940 (_mm_mask_i32scatter_pd): Ditto.
38941 (_mm256_i64scatter_ps): Ditto.
38942 (_mm256_mask_i64scatter_ps): Ditto.
38943 (_mm_i64scatter_ps): Ditto.
38944 (_mm_mask_i64scatter_ps): Ditto.
38945 (_mm256_i64scatter_pd): Ditto.
38946 (_mm256_mask_i64scatter_pd): Ditto.
38947 (_mm_i64scatter_pd): Ditto.
38948 (_mm_mask_i64scatter_pd): Ditto.
38949 (_mm256_i32scatter_epi32): Ditto.
38950 (_mm256_mask_i32scatter_epi32): Ditto.
38951 (_mm_i32scatter_epi32): Ditto.
38952 (_mm_mask_i32scatter_epi32): Ditto.
38953 (_mm256_i32scatter_epi64): Ditto.
38954 (_mm256_mask_i32scatter_epi64): Ditto.
38955 (_mm_i32scatter_epi64): Ditto.
38956 (_mm_mask_i32scatter_epi64): Ditto.
38957 (_mm256_i64scatter_epi32): Ditto.
38958 (_mm256_mask_i64scatter_epi32): Ditto.
38959 (_mm_i64scatter_epi32): Ditto.
38960 (_mm_mask_i64scatter_epi32): Ditto.
38961 (_mm256_i64scatter_epi64): Ditto.
38962 (_mm256_mask_i64scatter_epi64): Ditto.
38963 (_mm_i64scatter_epi64): Ditto.
38964 (_mm_mask_i64scatter_epi64): Ditto.
38965 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
38966 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
38967 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
38968 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
38969 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
38970 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
38971 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
38972 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
38973 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
38974 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
38975 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
38976 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
38977 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
38978 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
38979 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
38980 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
38981 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
38982 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
38983 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
38984 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
38985 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
38986 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
38987 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
38988 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
38989 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
38990 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
38991 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
38992 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
38993 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
38994 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
38995 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
38996 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
38997 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
38998 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
38999 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
39000 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
39001 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
39002 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
39003 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
39004 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
39005 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
39006 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
39007 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
39008 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
39009 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
39010 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
39011 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
39012 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
39013 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
39014 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
39015 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
39016 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
39017 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
39018 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
39019 definitions accordingly.
39021 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
39022 Kuan-Lin Chen <kuanlinchentw@gmail.com>
39025 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
39028 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
39031 * ira-conflicts.c (ira_build_conflicts): Update total conflict
39032 hard regs for inner regno.
39034 2017-01-17 Martin Liska <mliska@suse.cz>
39037 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
39038 assumption and add comment.
39040 2017-01-17 Nathan Sidwell <nathan@acm.org>
39042 * ipa-visibility.c (localize_node): New function, broken out of ...
39043 (function_and_variable_visibility): ... here. Call it.
39045 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
39047 PR middle-end/77445
39048 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
39049 correctly set frequency of oudgoing edge.
39050 (duplicate_thread_path): Fix profile updating.
39052 2017-01-17 Jakub Jelinek <jakub@redhat.com>
39055 * configure.ac: Add GCC_BASE_VER.
39056 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
39057 version from BASE-VER file.
39058 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
39059 (gcc.o): Depend on $(BASEVER).
39060 * common.opt (dumpfullversion): New option.
39061 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
39062 * doc/invoke.texi: Document -dumpfullversion.
39063 * doc/install.texi: Document --with-gcc-major-version-only.
39064 * configure: Regenerated.
39066 2017-01-17 Richard Biener <rguenther@suse.de>
39068 PR tree-optimization/71433
39069 * tree-vrp.c (register_new_assert_for): Merge same asserts
39070 on all incoming edges.
39071 (process_assert_insertions_for): Handle insertions at the
39074 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
39076 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
39077 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
39079 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
39082 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
39085 2017-01-17 Alan Modra <amodra@gmail.com>
39088 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
39089 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
39090 symbolic stack limit when pic.
39092 2017-01-16 Martin Sebor <msebor@redhat.com>
39094 PR tree-optimization/78608
39095 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
39097 2017-01-16 Jeff Law <law@redhat.com>
39100 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
39101 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
39102 for several include directories that may be relative to sysroot.
39103 * config/i386/x-mingw32 (gplus_includedir): Define.
39104 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
39105 (native_system_includedir): Likewise.
39106 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
39107 override if TARGET_SYSTEM_ROOT is defined.
39108 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
39110 PR tree-optimization/79090
39111 PR tree-optimization/33562
39112 PR tree-optimization/61912
39113 PR tree-optimization/77485
39114 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
39115 and computed trims into the dump file.
39117 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
39119 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
39121 2017-01-16 Jakub Jelinek <jakub@redhat.com>
39124 * gimplify.c (gimplify_init_constructor): If want_value and
39125 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
39129 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
39130 sequence. Formatting fixes.
39131 (doloop_optimize): Formatting fixes.
39134 * gcc.c (debug_level_greater_than_spec_func): New function.
39135 (static_spec_functions): Add debug-level-gt spec function.
39136 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
39138 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
39139 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
39140 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
39141 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
39142 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
39143 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
39145 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
39147 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
39148 QImode fixups to general and mask registers only.
39150 2017-01-16 Carl Love <cel@us.ibm.com>
39152 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
39153 for built-in functions
39154 vector signed char vec_nabs (vector signed char)
39155 vector signed short vec_nabs (vector signed short)
39156 vector signed int vec_nabs (vector signed int)
39157 vector signed long long vec_nabs (vector signed long long)
39158 vector float vec_nabs (vector float)
39159 vector double vec_nabs (vector double)
39160 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
39162 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
39163 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
39164 * doc/extend.texi: Update the documentation file for the new built-in
39167 2017-01-16 Martin Sebor <msebor@redhat.com>
39169 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
39172 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39174 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
39175 UNSPEC_VSX__XXSPLTD to require special splat handling.
39177 2017-01-16 David Malcolm <dmalcolm@redhat.com>
39180 * system.h: Poison strndup.
39182 2017-01-16 Alan Modra <amodra@gmail.com>
39185 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
39188 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
39190 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
39192 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
39194 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
39195 call recog here. Assert that INSN_CODE (insn) is non-negative.
39197 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
39200 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
39202 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
39203 in the currently scheduled RTL fragment.
39205 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
39207 PR rtl-optimization/78751
39208 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
39211 2017-01-14 Jeff Law <law@redhat.com>
39213 PR tree-optimization/79090
39214 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
39215 variable length stores.
39216 (compute_trims): Delete dead assignment to *trim_tail.
39217 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
39220 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
39222 PR rtl-optimization/78626
39223 PR rtl-optimization/78727
39224 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
39225 of a block, and split such blocks after everything else is finished.
39227 2017-01-14 Alan Modra <amodra@gmail.com>
39230 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
39231 target legitimate_combined_insn.
39232 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
39233 (rs6000_legitimate_combined_insn): New function.
39234 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
39236 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
39237 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
39238 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
39240 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
39242 * doc/frontends.texi (G++ and GCC): Remove references to Java.
39244 2017-01-13 Jeff Law <law@redhat.com>
39246 PR tree-optimization/33562
39247 PR tree-optimization/61912
39248 PR tree-optimization/77485
39249 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
39251 (delete_dead_assignment): Likewise.
39252 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
39253 statement to delete_dead_call and delete_dead_assignment.
39255 2017-01-13 David Malcolm <dmalcolm@redhat.com>
39258 * substring-locations.c (format_warning_va): Strengthen case 1 so
39259 that both endpoints of the substring must be within the format
39260 range for just the substring to be printed.
39262 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
39264 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
39265 * config/i386/i386.c (ix86_target_string): Add missing options
39266 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
39267 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
39268 flags_other and ix86_target_other to flags2_other. Display unknown
39270 (ix86_valid_target_attribute_inner_p): Add missing options and
39271 reorder options by implied ISAs, as in ix86_target_string.
39273 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
39275 * hash-table.h (hash_table::too_empty_p): New function.
39276 (hash_table::expand): Use it.
39277 (hash_table::traverse): Likewise.
39278 (hash_table::empty_slot): Use sizeof (value_type) instead of
39279 sizeof (PTR) to convert bytes to elements. Shrink the table
39280 if the current size is excessive for the current number of
39283 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
39285 * ira-costs.c (record_reg_classes): Break from the inner loop
39286 early once alt_fail is known to be true. Update outer loop
39287 handling accordingly.
39289 2017-01-13 Jeff Law <law@redhat.com>
39291 * tree-ssa-dse.c (decrement_count): New function.
39292 (increment_start_addr, maybe_trim_memstar_call): Likewise.
39293 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
39294 when we know the partially dead statement is a mem* function.
39296 PR tree-optimization/61912
39297 PR tree-optimization/77485
39298 * tree-ssa-dse.c: Include expr.h.
39299 (maybe_trim_constructor_store): New function.
39300 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
39302 PR tree-optimization/33562
39303 PR tree-optimization/61912
39304 PR tree-optimization/77485
39305 * doc/invoke.texi: Document new dse-max-object-size param.
39306 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
39307 * tree-ssa-dse.c: Include params.h.
39308 (dse_store_status): New enum.
39309 (initialize_ao_ref_for_dse): New, partially extracted from
39311 (valid_ao_ref_for_dse, normalize_ref): New.
39312 (setup_live_bytes_from_ref, compute_trims): Likewise.
39313 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
39314 (maybe_trim_partially_dead_store): Likewise.
39315 (maybe_trim_complex_store): Likewise.
39316 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
39317 Track what bytes live from the original store. Return tri-state
39318 for dead, partially dead or live.
39319 (dse_dom_walker): Add constructor, destructor and new private members.
39320 (delete_dead_call, delete_dead_assignment): New extracted from
39322 (dse_optimize_stmt): Make a member of dse_dom_walker.
39323 Use initialize_ao_ref_for_dse.
39325 PR tree-optimization/33562
39326 PR tree-optimization/61912
39327 PR tree-optimization/77485
39328 * sbitmap.h (bitmap_count_bits): Prototype.
39329 (bitmap_clear_range, bitmap_set_range): Likewise.
39330 * sbitmap.c (bitmap_clear_range): New function.
39331 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
39333 2017-01-13 Martin Liska <mliska@suse.cz>
39336 * function.c (set_cfun): Add new argument force.
39337 * function.h (set_cfun): Likewise.
39338 * ipa-inline-transform.c (inline_call): Use the function when
39339 strict alising from is dropped for function we inline to.
39341 2017-01-13 Richard Biener <rguenther@suse.de>
39343 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
39344 for dumping GIMPLE INTEGER_CSTs.
39346 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
39348 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
39349 to 201112L since C++17.
39351 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
39354 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
39355 if -fsanitize=kernel-address is present.
39357 2017-01-13 Richard Biener <rguenther@suse.de>
39359 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
39360 as _Literal ( type ) number in case usual suffixes do not
39361 preserve all information.
39363 2017-01-13 Richard Biener <rguenther@suse.de>
39365 PR tree-optimization/77283
39366 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
39367 and ssa-iterators.h.
39368 (is_feasible_trace): Implement a cost model based on joiner
39371 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
39374 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
39375 char or short to __float128/_Float128 directly.
39377 2017-01-12 Martin Sebor <msebor@redhat.com>
39379 to -Wformat-overflow.
39380 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
39381 (min_bytes_remaining): Same.
39382 (get_string_length): Same.
39383 (format_string): Same.
39384 (format_directive): Same.
39386 (pass_sprintf_length::handle_gimple_call): Same.
39388 2017-01-12 Jakub Jelinek <jakub@redhat.com>
39390 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
39391 info.nowrite calls with no lhs that can't throw. Return bool
39392 whether gsi_remove has been called or not.
39393 (pass_sprintf_length::handle_gimple_call): Return bool whether
39394 try_substitute_return_value called gsi_remove. Formatting fix.
39395 (pass_sprintf_length::execute): Don't use gsi_remove if
39396 handle_gimple_call returned true.
39399 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
39400 be removed due to side-effects, don't remove following barrier nor
39401 turn the successor edge into fallthru edge.
39403 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39406 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
39407 element-reversing loads and stores as not swappable.
39409 2017-01-12 Nathan Sidwell <nathan@acm.org>
39410 Nicolai Stange <nicstange@gmail.com>
39412 * combine.c (try_combine): Don't ignore result of overlap checking
39413 loop. Combine overlap & asm check into single loop.
39415 2017-01-12 Richard Biener <rguenther@suse.de>
39417 * tree-pretty-print.c (dump_generic_node): Provide -gimple
39418 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
39420 2017-01-12 Richard Biener <rguenther@suse.de>
39422 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
39423 and TS_TARGET_OPTION directly derive from TS_BASE.
39424 * tree-core.h (tree_optimization_option): Derive from tree_base.
39425 (tree_target_option): Likewise.
39427 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
39429 * config/i386/i386.c (memory_address_length): Increase len
39430 only when rip_relative_addr_p returns false.
39432 2017-01-11 Julia Koval <julia.koval@intel.com>
39434 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
39435 (OPTION_MASK_ISA_SGX_SET): New.
39436 (ix86_handle_option): Handle OPT_msgx.
39437 * config.gcc: Added sgxintrin.h.
39438 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
39439 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
39440 * config/i386/i386.c (ix86_target_string): Add -msgx.
39442 (ix86_option_override_internal): Handle new options.
39443 (ix86_valid_target_attribute_inner_p): Add sgx.
39444 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
39445 * config/i386/i386.opt: Add msgx.
39446 * config/i386/sgxintrin.h: New file.
39447 * config/i386/x86intrin.h: Add sgxintrin.h.
39449 2017-01-11 Jakub Jelinek <jakub@redhat.com>
39452 * fold-const.c (maybe_nonzero_address): Return 1 for function
39454 (tree_single_nonzero_warnv_p): Don't handle function local objects
39458 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
39461 2017-01-11 David Malcolm <dmalcolm@redhat.com>
39464 * opts.c: Include "spellcheck.h"
39465 (struct string_fragment): New struct.
39466 (struct edit_distance_traits<const string_fragment &>): New
39468 (get_closest_sanitizer_option): New function.
39469 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
39471 2017-01-11 Jakub Jelinek <jakub@redhat.com>
39473 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
39475 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
39476 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
39477 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
39478 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
39479 for initial die_offset if dwarf_split_debug_info.
39480 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
39481 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
39483 (output_skeleton_debug_sections): Formatting fix. Use
39484 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
39485 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
39487 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
39489 * config/arm/cortex-a53.md: Add bypasses for
39490 cortex_a53_r2f_cvt.
39491 (cortex_a53_r2f): Only use for transfers.
39492 (cortex_a53_f2r): Likewise.
39493 (cortex_a53_r2f_cvt): Add reservation for conversions.
39494 (cortex_a53_f2r_cvt): Likewise.
39496 2017-01-11 Tamar Christina <tamar.christina@arm.com>
39498 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
39499 to all inlined functions, change static to extern.
39501 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
39504 * config/arm/arm.c (legitimize_pic_address): Handle reference to
39506 (arm_assemble_integer): Likewise.
39508 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
39510 * config.gcc: Use new awk script to check CPU, FPU and architecture
39511 parameters for --with-... options.
39512 * config/arm/parsecpu.awk: New file
39513 * config/arm/arm-cpus.in: New file.
39514 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
39516 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
39518 * config/arm/t-arm: Update dependency rules.
39519 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
39520 of processing .def files.
39521 * config/arm/genopt.sh: Deleted.
39522 * config/arm/gentune.sh: Deleted.
39523 * config/arm/arm-cores.def: Deleted.
39524 * config/arm/arm-arches.def: Deleted.
39525 * config/arm/arm-fpus.def: Deleted.
39526 * config/arm/arm-tune.md: Regenerated.
39527 * config/arm/arm-tables.opt: Regenerated.
39528 * config/arm/arm-cpu.h: New generated file.
39529 * config/arm/arm-cpu-data.h: New generated file.
39530 * config/arm/arm-cpu-cdata.h: New generated file.
39532 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
39535 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
39537 (input_varpool_node): Unpack dynamically_initialized bit.
39539 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
39541 PR rtl-optimization/79032
39542 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
39543 the alignment of the adjusted memory reference against that of MODE,
39544 instead of the alignment of the original memory reference.
39546 2017-01-11 Martin Jambor <mjambor@suse.cz>
39548 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
39550 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
39551 decorated functions.
39553 2017-01-11 Richard Biener <rguenther@suse.de>
39555 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
39556 set range/nonnull info for PHI results. Do not set it on
39557 stmts marked for removal.
39559 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
39561 * expr.c (store_field): In the bitfield case, fetch the return value
39562 from the registers before applying a single big-endian adjustment.
39563 Always do a final load for a BLKmode value not larger than a word.
39565 2017-01-10 David Malcolm <dmalcolm@redhat.com>
39568 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
39569 that we correctly handle column numbers greater than
39570 LINE_MAP_MAX_COLUMN_NUMBER.
39572 2017-01-10 Martin Sebor <msebor@redhat.com>
39574 PR middle-end/78245
39575 * gimple-ssa-sprintf.c (get_destination_size): Call
39576 {init,fini}object_sizes.
39577 * tree-object-size.c (addr_object_size): Adjust.
39578 (pass_through_call): Adjust.
39579 (pass_object_sizes::execute): Adjust.
39580 * tree-object-size.h (fini_object_sizes): Declare.
39582 2017-01-10 Martin Sebor <msebor@redhat.com>
39584 PR tree-optimization/78775
39585 * builtins.c (get_size_range): Move...
39586 * calls.c: ...to here.
39587 (alloc_max_size): Accept zero argument.
39588 (operand_signed_p): Remove.
39589 (maybe_warn_alloc_args_overflow): Call get_size_range.
39590 * calls.h (get_size_range): Declare.
39592 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
39594 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
39595 from TI's devices.csv file as of September 2016.
39596 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
39598 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
39600 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
39601 * doc/invoke.texi: Likewise.
39602 * doc/md.texi: Likewise.
39603 * doc/objc.texi: Likewise.
39605 2017-01-10 Joshua Conner <joshconner@google.com>
39607 * config/arm/fuchsia-elf.h: New file.
39608 * config/fuchsia.h: New file.
39609 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
39610 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
39612 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
39614 2016-01-10 Richard Biener <rguenther@suse.de>
39616 PR tree-optimization/79034
39617 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
39618 Propagate out degenerate PHIs in the joiner.
39620 2017-01-10 Martin Liska <mliska@suse.cz>
39622 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
39623 (sort_congruence_classes_by_decl_uid): Likewise.
39624 (sort_congruence_class_groups_by_decl_uid): Likewise.
39625 (sem_item_optimizer::merge_classes): Sort class, groups in these
39626 classes and members in the groups by DECL_UID of declarations.
39627 This would make merge operations stable.
39629 2017-01-10 Martin Liska <mliska@suse.cz>
39631 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
39632 usage of m_classes_vec.
39633 (sem_item_optimizer::~sem_item_optimizer): Likewise.
39634 (sem_item_optimizer::get_group_by_hash): Likewise.
39635 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
39636 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
39637 (sem_item_optimizer::verify_classes): Likewise.
39638 (sem_item_optimizer::process_cong_reduction): Likewise.
39639 (sem_item_optimizer::dump_cong_classes): Likewise.
39640 (sem_item_optimizer::merge_classes): Likewise.
39641 * ipa-icf.h (congruence_class_hash): Rename from
39642 congruence_class_group_hash. Remove declaration of m_classes_vec.
39644 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
39646 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
39647 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
39648 * config.gcc: Add avx512vpopcntdqintrin.h.
39649 * config/i386/avx512vpopcntdqintrin.h: New.
39650 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
39651 * config/i386/i386-builtin-types.def: Add new types.
39652 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
39653 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
39654 __builtin_ia32_vpopcountq_v8di_mask): New.
39655 * config/i386/i386-c.c (ix86_target_macros_internal): Define
39656 __AVX512VPOPCNTDQ__.
39657 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
39658 (PTA_AVX512VPOPCNTDQ): Define.
39659 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
39660 TARGET_AVX512VPOPCNTDQ_P): Define.
39661 * config/i386/i386.opt: Add mavx512vpopcntdq.
39662 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
39663 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
39665 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
39667 PR middle-end/77484
39668 * predict.def (PRED_CALL): Set to 67.
39670 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
39672 * expr.c (store_field): In the bitfield case, if the value comes from
39673 a function call and is of an aggregate type returned in registers, do
39674 not modify the field mode; extract the value in all cases if the mode
39675 is BLKmode and the size is not larger than a word.
39677 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
39680 * config/i386/cpuid.h: Fix undefined behavior.
39682 2017-01-04 Jeff Law <law@redhat.com>
39684 PR tree-optimization/79007
39685 PR tree-optimization/67955
39686 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
39687 conservative for pt.null when flag_non_call_exceptions is on.
39689 2017-01-09 Jakub Jelinek <jakub@redhat.com>
39691 PR translation/79019
39692 PR translation/79020
39693 * params.def (PARAM_INLINE_MIN_SPEEDUP,
39694 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
39695 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
39697 * config/avr/avr.opt (maccumulate-args): Likewise.
39698 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
39699 * common.opt (freport-bug): Likewise.
39700 * cif-code.def (CIF_FINAL_ERROR): Likewise.
39701 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
39702 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
39703 translatable string.
39704 * config/i386/i386.c (function_value_32): Likewise.
39705 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
39706 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
39708 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
39709 * common/config/msp430/msp430-common.c (msp430_handle_option):
39711 * symtab.c (symtab_node::verify_base): Likewise.
39712 * opts.c (set_debug_level): Likewise.
39713 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
39714 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
39715 missing whitespace to translatable strings.
39716 * config/avr/avr.md (bswapsi2): Fix typo in comment.
39717 * config/sh/superh.h: Likewise.
39718 * config/i386/xopintrin.h: Likewise.
39719 * config/i386/znver1.md: Likewise.
39720 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
39721 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
39722 * double-int.h (struct double_int): Likewise.
39723 * double-int.c (div_and_round_double): Likewise.
39724 * wide-int.cc: Likewise.
39725 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
39726 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
39727 * cfgcleanup.c (crossjumps_occured): Renamed to ...
39728 (crossjumps_occurred): ... this.
39729 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
39732 PR tree-optimization/78899
39733 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
39734 returning bool return struct loop *, NULL for failure and the new
39736 (versionable_outer_loop_p): Don't version outer loop if it has
39737 dont_vectorized bit set.
39738 (tree_if_conversion): When versioning outer loop, ensure
39739 tree_if_conversion is performed also on the inner loop of the
39740 non-vectorizable outer loop copy.
39741 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
39742 LOOP_VECTORIZED in inner loop of the scalar outer loop and
39743 prevent vectorization of it.
39744 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
39745 the outer loop vectorization of the non-scalar version is attempted
39746 before vectorization of the inner loop in scalar version. If
39747 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
39748 vectorization of its inner loop.
39749 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
39750 has 2 inner loops, rename also on edges from bb whose single pred
39751 is outer_loop->header. Fix typo in function comment.
39753 2017-01-09 Martin Sebor <msebor@redhat.com>
39756 * asan.c (asan_emit_stack_protection): Increase local buffer size
39757 to avoid snprintf truncation warning.
39759 2017-01-09 Andrew Pinski <apinski@cavium.com>
39761 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
39762 to reference thunderx2t99 for the tuning structure
39763 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
39765 (thunderx2t99_extra_costs): This.
39766 * config/aarch64/aarch64-tune.md: Regenerate.
39767 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
39768 (vulcan_addrcost_table): This.
39769 (vulcan_regmove_cost): Rename to ...
39770 (thunderx2t99_regmove_cost): This.
39771 (vulcan_vector_cost): Rename to ...
39772 (thunderx2t99_vector_cost): this.
39773 (vulcan_branch_cost): Rename to ...
39774 (thunderx2t99_branch_cost): This.
39775 (vulcan_tunings): Rename to ...
39776 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
39777 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
39779 2017-01-09 Martin Jambor <mjambor@suse.cz>
39783 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
39784 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
39785 (propagate_vr_accross_jump_function): Use the above function for all
39786 value range computations for pass-through jump functions and type
39787 converasion from explicit value range values.
39788 (ipcp_propagate_stage): Do not attempt to deduce types of formal
39789 parameters from TYPE_ARG_TYPES.
39790 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
39791 (ipa_write_node_info): Stream type of the actual argument.
39792 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
39794 2017-01-09 Martin Liska <mliska@suse.cz>
39797 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
39798 (lookup_compiler): Do not show error message with have_E.
39800 2017-01-09 Jakub Jelinek <jakub@redhat.com>
39802 PR tree-optimization/78938
39803 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
39804 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
39805 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
39806 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
39809 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39811 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
39814 2017-01-09 Richard Biener <rguenther@suse.de>
39816 PR tree-optimization/78997
39817 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
39818 name condition properly.
39820 2017-01-09 Richard Biener <rguenther@suse.de>
39823 * dwarf2out.c (is_cxx): New overload with context.
39824 (is_naming_typedef_decl): Use it.
39826 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
39828 * invoke.texi (Option Summary): Correct spacing in option lists
39829 and add line breaks to fix over-long lines.
39831 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
39833 PR middle-end/17660
39835 * extend.texi (Common Variable Attributes): Add xref to GCC
39836 Internals manual to explain mode attribute keywords.
39838 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
39841 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
39842 and Preprocessor Options.
39843 (Options for Linking): Document -pthread here....
39844 (RS/6000 and PowerPC Options): ...not here.
39845 (Solaris 2 Options): ...or here.
39846 * doc/cppopts.texi: Document -pthread.
39848 2017-01-08 Martin Sebor <msebor@redhat.com>
39850 PR middle-end/77708
39851 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
39852 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
39853 New member functions.
39854 (format_directive): Used them.
39856 (pass_sprintf_length::handle_gimple_call): Same.
39857 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
39858 to avoid truncation for any argument.
39859 (extract_affine_mul): Same.
39860 * tree.c (get_file_function_name): Same.
39862 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
39864 PR middle-end/77484
39865 * predict.def (PRED_INDIR_CALL): Set to 86.
39867 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
39869 PR preprocessor/54124
39870 * doc/cppopts.texi: Reformat -d subtable to list the full name
39871 of the options. Add cross-reference to the docs for the general
39872 compiler -d options.
39873 * doc/invoke.texi (Developer Options): Add cross-reference to the
39874 preprocessor-specific -d option documentation.
39876 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
39878 PR preprocessor/13498
39879 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
39880 redudant material, and reflect new command-line options.
39881 (System Headers): Likewise.
39883 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
39885 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
39886 -isystem, and -idirafter. Copy-edit.
39887 * doc/cppopts.texi: Copy-edit. Remove contradiction about
39888 default for -ftrack-macro-expansion. Delete obsolete and
39889 badly-formatted implementation details about -fdebug-cpp output.
39890 * doc/cppwarnopts.texi: Copy-edit.
39892 2017-01-07 David Malcolm <dmalcolm@redhat.com>
39895 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
39896 that the transition from a max line width >= 1<<10 to narrower
39897 lines works correctly.
39899 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
39901 * doc/options.texi (PerFunction): New.
39902 * opt-functions.awk (switch_flags): Map both Optimization and
39903 PerFunction to CL_OPTIMIZATION.
39904 * opth-gen.awk: Test for PerFunction flag along with
39906 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
39907 it only when the latter is present. Skip those that don't in
39908 the hash function generator.
39909 * common.opt (fvar-tracking): Mark as PerFunction instead of
39911 (fvar-tracking-assignments): Likewise.
39912 (fvar-tracking-assignments-toggle): Likewise.
39913 (fvar-tracking-uninit): Likewise.
39915 2017-01-07 Jakub Jelinek <jakub@redhat.com>
39917 PR translation/79018
39918 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
39921 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
39924 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
39925 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
39926 TARGET_LONG_JUMP_TABLE_OFFSETS.
39927 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
39928 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
39929 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
39930 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
39931 * config/m68k/m68k.md (tablejump expander): Likewise.
39932 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
39933 TARGET_LONG_JUMP_TABLE_OFFSETS.
39934 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
39935 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
39937 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
39938 David Holsgrove <david.holsgrove@xilinx.com>
39940 * common/config/microblaze/microblaze-common.c
39941 (TARGET_EXCEPT_UNWIND_INFO): Remove.
39942 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
39944 * config/microblaze/microblaze.c (microblaze_must_save_register)
39945 (microblaze_expand_epilogue, microblaze_return_addr): Handle
39947 (microblaze_eh_return): New function.
39948 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
39949 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
39950 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
39951 * config/microblaze/microblaze.md (eh_return): New pattern.
39953 2017-01-06 Jakub Jelinek <jakub@redhat.com>
39955 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
39956 GCC_DIAGNOSTIC_STRINGIFY): Define.
39958 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
39960 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
39962 * config/arm/arm.md (<mcrr>): New.
39964 * config/arm/arm.c (arm_arch5te): New.
39965 (arm_option_override): Set arm_arch5te.
39966 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
39968 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
39969 (arm_mcrr_qualifiers): ... this. New.
39970 (MRRC_QUALIFIERS): Define to...
39971 (arm_mrrc_qualifiers): ... this. New.
39972 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
39974 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
39975 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
39976 (MRRCI, mrrc, MRRC): New.
39977 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
39978 VUNSPEC_MRRC2): New.
39980 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
39982 * config/arm/arm.md (<mcr>): New.
39984 * config/arm/arm.c (arm_coproc_builtin_available): Add
39985 support for mcr, mrc, mcr2 and mrc2.
39986 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
39987 (arm_mcr_qualifiers): ... this. New.
39988 (MRC_QUALIFIERS): Define to ...
39989 (arm_mrc_qualifiers): ... this. New.
39990 (MCR_QUALIFIERS): Define to ...
39991 (arm_mcr_qualifiers): ... this. New.
39992 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
39994 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
39995 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
39996 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
39997 VUNSPEC_MRC2): New.
39999 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40001 * config/arm/arm.md (*ldc): New.
40005 * config/arm/arm.c (arm_coproc_builtin_available): Add
40006 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
40007 (arm_coproc_ldc_stc_legitimate_address): New.
40008 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
40009 'qualifier_const_pointer'.
40010 (LDC_QUALIFIERS): Define to...
40011 (arm_ldc_qualifiers): ... this. New.
40012 (STC_QUALIFIERS): Define to...
40013 (arm_stc_qualifiers): ... this. New.
40014 * config/arm/arm-protos.h
40015 (arm_coproc_ldc_stc_legitimate_address): New.
40016 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
40017 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
40018 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
40019 stc2, stcl, stc2l): New.
40020 * config/arm/constraints.md (Uz): New.
40021 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
40022 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
40023 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
40024 VUNSPEC_STC2L): New.
40026 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40028 * config/arm/arm.md (<cdp>): New.
40029 * config/arm/arm.c (neon_const_bounds): Rename this ...
40030 (arm_const_bounds): ... this.
40031 (arm_coproc_builtin_available): New.
40032 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
40033 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
40034 (CDP_QUALIFIERS): Define to...
40035 (arm_cdp_qualifiers): ... this. New.
40037 (arm_expand_builtin_args): Add case for 6 arguments.
40038 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
40039 (arm_const_bounds): ... this.
40040 (arm_coproc_builtin_available): New.
40041 * config/arm/arm_acle.h (__arm_cdp): New.
40043 * config/arm/arm_acle_builtins.def (cdp): New.
40045 * config/arm/iterators.md (CDPI,CDP,cdp): New.
40046 * config/arm/neon.md: Rename all 'neon_const_bounds' to
40047 'arm_const_bounds'.
40048 * config/arm/types.md (coproc): New.
40049 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
40050 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
40051 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
40052 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
40054 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40056 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
40057 (UBINOP_QUALIFIERS): New.
40059 (acle_builtin_data): New. Change comment.
40060 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
40061 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
40062 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
40063 arm_acle_builtins.def.
40064 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
40065 (arm_init_acle_builtins): New.
40066 (CRC32_BUILTIN): Remove.
40067 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
40068 crc32cb, crc32ch and crc32cw.
40069 (arm_init_crc32_builtins): Remove.
40070 (arm_init_builtins): Use arm_init_acle_builtins rather
40071 than arm_init_crc32_builtins.
40072 (arm_expand_acle_builtin): New.
40073 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
40074 * config/arm/arm_acle_builtins.def: New.
40076 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40078 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
40079 (arm_builtin_datum): ... this.
40080 (arm_init_neon_builtin): Rename to ...
40081 (arm_init_builtin): ... this. Add a new parameters PREFIX
40082 and USE_SIG_IN_NAME.
40083 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
40084 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
40085 'arm_builtin_datum'.
40086 (arm_init_vfp_builtins): Likewise.
40087 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
40088 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
40089 (arm_expand_neon_args): Rename to ...
40090 (arm_expand_builtin_args): ... this. Rename builtin_arg
40091 enum values and differentiate between ARG_BUILTIN_MEMORY
40092 and ARG_BUILTIN_NEON_MEMORY.
40093 (arm_expand_neon_builtin_1): Rename to ...
40094 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
40095 values, arm_expand_builtin_args and add bool parameter NEON.
40096 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
40097 (arm_expand_vfp_builtin): Likewise.
40098 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
40100 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40102 PR middle-end/77484
40103 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
40104 * predict.c (tree_estimate_probability_bb): Reverse direction of
40105 polymorphic call predictor.
40107 2017-01-06 David Malcolm <dmalcolm@redhat.com>
40109 * passes.c (execute_one_pass): Split out pass-skipping logic into...
40110 (determine_pass_name_match): ...this new function and...
40111 (should_skip_pass_p): ...this new function.
40113 2017-01-06 Nathan Sidwell <nathan@acm.org>
40115 * ipa-visibility.c (function_and_variable_visibility): Reformat
40116 comments and long lines. Remove extrneous if.
40117 * symtab.c (symtab_node::make_decl_local): Fix code format.
40118 (symtab_node::set_section_for_node): Fix comment typo.
40120 2017-01-06 Martin Liska <mliska@suse.cz>
40123 * lra-constraints.c: Rename invariant to lra_invariant.
40124 * predict.c (set_even_probabilities): Initialize e to NULL.
40126 2017-01-05 Martin Sebor <msebor@redhat.com>
40128 PR tree-optimization/78910
40129 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
40130 (format_integer): Correct off-by-one error in the handling
40131 of precision with negative numbers in signed conversions..
40133 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
40135 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
40137 2017-01-05 Jakub Jelinek <jakub@redhat.com>
40139 PR tree-optimization/71016
40140 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
40141 factor_out_conditional_conversion. Formatting fix.
40142 (factor_out_conditional_conversion): Add cond_stmt argument.
40143 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
40144 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
40147 2017-01-05 David Malcolm <dmalcolm@redhat.com>
40149 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
40150 read-rtl-function.o, and selftest-rtl.o.
40151 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
40152 (selftest::aarch64_test_loading_full_dump): New function.
40153 (selftest::aarch64_run_selftests): New function.
40154 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
40155 selftest::aarch64_run_selftests.
40156 * config/i386/i386.c
40157 (selftest::ix86_test_loading_dump_fragment_1): New function.
40158 (selftest::ix86_test_loading_call_insn): New function.
40159 (selftest::ix86_test_loading_full_dump): New function.
40160 (selftest::ix86_test_loading_unspec): New function.
40161 (selftest::ix86_run_selftests): Call the new functions.
40162 * emit-rtl.c (maybe_set_max_label_num): New function.
40163 * emit-rtl.h (maybe_set_max_label_num): New decl.
40164 * function.c (instantiate_decls): Guard call to
40165 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
40166 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
40168 * gensupport.c (gen_reader::gen_reader): Pass "false"
40169 for new "compact" param of rtx_reader.
40170 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
40171 rather than an empty string for NULL strings.
40172 * read-md.c: Potentially include config.h rather than bconfig.h.
40173 Wrap include of errors.h with #ifdef GENERATOR_FILE.
40174 (have_error): New global, copied from errors.c.
40175 (md_reader::read_name): Rename to...
40176 (md_reader::read_name_1): ...this, adding "out_loc" param,
40177 and converting "missing name or number" to returning false, rather
40179 (md_reader::read_name): Reimplement in terms of read_name_1.
40180 (md_reader::read_name_or_nil): New function.
40181 (md_reader::read_string): Handle "(nil)" by returning NULL.
40182 (md_reader::md_reader): Add new param "compact".
40183 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
40184 (md_reader::read_file): New method.
40185 * read-md.h (md_reader::md_reader): Add new param "compact".
40186 (md_reader::read_file): New method.
40187 (md_reader::is_compact): New accessor.
40188 (md_reader::read_name): Convert return type from void to file_location.
40189 (md_reader::read_name_or_nil): New decl.
40190 (md_reader::read_name_1): New decl.
40191 (md_reader::m_compact): New field.
40192 (noop_reader::noop_reader): Pass "false" for new "compact" param
40194 (rtx_reader::rtx_reader): Add new "compact" param.
40195 (rtx_reader::read_rtx_operand): Make virtual and convert return
40196 type from void to rtx.
40197 (rtx_reader::read_until): New decl.
40198 (rtx_reader::handle_any_trailing_information): New virtual function.
40199 (rtx_reader::postprocess): New virtual function.
40200 (rtx_reader::finalize_string): New virtual function.
40201 (rtx_reader::m_in_call_function_usage): New field.
40202 (rtx_reader::m_reuse_rtx_by_id): New field.
40203 * read-rtl-function.c: New file.
40204 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
40205 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
40206 (selftest::verify_three_block_rtl_cfg): New decl.
40207 * read-rtl-function.h: New file.
40208 * read-rtl.c: Potentially include config.h rather than bconfig.h.
40209 For host, include function.h, memmodel.h, and emit-rtl.h.
40210 (one_time_initialization): New function.
40211 (struct compact_insn_name): New struct.
40212 (compact_insn_names): New array.
40213 (find_code): Handle insn codes in compact dumps.
40214 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
40215 (bind_subst_iter_and_attr): Likewise.
40216 (add_condition_to_string): Likewise.
40217 (add_condition_to_rtx): Likewise.
40218 (apply_attribute_uses): Likewise.
40219 (add_current_iterators): Likewise.
40220 (apply_iterators): Likewise.
40221 (initialize_iterators): Guard usage of apply_subst_iterator with
40222 #ifdef GENERATOR_FILE.
40223 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
40224 (md_reader::read_mapping): Likewise.
40225 (add_define_attr_for_define_subst): Likewise.
40226 (add_define_subst_attr): Likewise.
40227 (read_subst_mapping): Likewise.
40228 (check_code_iterator): Likewise.
40229 (rtx_reader::read_rtx): Likewise. Move one-time initialization
40231 (one_time_initialization): New function.
40232 (rtx_reader::read_until): New method.
40233 (read_flags): New function.
40234 (parse_reg_note_name): New function.
40235 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
40236 Handle reuse_rtx ids.
40237 Wrap iterator lookup within #ifdef GENERATOR_FILE.
40238 Add parsing support for RTL dumps, mirroring the special-cases in
40239 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
40240 values, and calling handle_any_trailing_information.
40241 (rtx_reader::read_rtx_operand): Convert return type from void
40242 to rtx, returning return_rtx. Handle case 'e'. Call
40243 finalize_string on XSTR and XTMPL fields.
40244 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
40245 "(nil)" values were omitted. Call the postprocess vfunc on the
40247 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
40248 class ctor. Initialize m_in_call_function_usage. Call
40249 one_time_initialization.
40250 * rtl-tests.c (selftest::test_uncond_jump): Call
40251 set_new_first_and_last_insn.
40252 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
40253 * selftest-rtl.c: New file.
40254 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
40255 (selftest::get_insn_by_uid): New decl.
40256 * selftest-run-tests.c (selftest::run_tests): Call
40257 read_rtl_function_c_tests.
40258 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
40259 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
40262 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
40264 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
40265 operands in a special way. Assert that pos+len <= mode precision.
40267 2017-01-05 Jakub Jelinek <jakub@redhat.com>
40269 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
40270 3 argument Alias with unlimited for the negative form.
40271 (fno-vect-cost-model): Removed.
40273 2017-01-05 Martin Liska <mliska@suse.cz>
40275 * hsa-gen.c (gen_hsa_divmod): New function.
40276 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
40278 2017-01-05 Martin Liska <mliska@suse.cz>
40281 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
40284 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
40286 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
40287 small constant length operands.
40289 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
40291 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
40292 between loop iterations.
40294 2017-01-05 Martin Liska <mliska@suse.cz>
40297 * gimplify.c (gimplify_decl_expr): Compare to
40298 asan_poisoned_variables instread of checking flags.
40299 (gimplify_target_expr): Likewise.
40300 (gimplify_expr): Likewise.
40301 (gimplify_function_tree): Conditionally initialize
40302 asan_poisoned_variables.
40304 2017-01-04 Jeff Law <law@redhat.com>
40306 PR tree-optimizatin/78812
40307 * rtl.h (contains_mem_rtx_p): Prototype.
40308 * ifcvt.c (containts_mem_rtx_p): Move from here to...
40309 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
40310 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
40311 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
40312 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
40314 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
40316 * input.c (assert_char_at_range): Default-initialize actual_range.
40318 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
40320 * df-scan.c (df_ref_create_structure): Make regno unsigned,
40321 to match the caller.
40323 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
40325 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
40326 insns after final jump in test to emit dummy move.
40328 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
40330 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
40331 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
40333 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
40335 * multiple_target.c (create_dispatcher_calls): Init e_next.
40336 * tree-ssa-loop-split.c (split_loop): Init border.
40337 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
40340 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
40345 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
40346 (altivec_register_operand): Do not return true if the operand
40347 contains a SUBREG mixing SImode and SFmode.
40348 (vsx_register_operand): Likewise.
40349 (vsx_reg_sfsubreg_ok): New predicate.
40350 (vfloat_operand): Do not return true if the operand contains a
40351 SUBREG mixing SImode and SFmode.
40352 (vint_operand): Likewise.
40353 (vlogical_operand): Likewise.
40354 (gpc_reg_operand): Likewise.
40355 (int_reg_operand): Likewise.
40356 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
40357 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
40358 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
40360 (rs6000_emit_move_si_sf_subreg): New helper function.
40361 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
40362 fixup SUBREGs involving SImode and SFmode.
40363 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
40364 numbers for the new peephole2 optimization.
40365 (peephole2 for SFmode unions): New peephole2 to optimize cases in
40366 the GLIBC math library that do AND/IOR/XOR operations on single
40367 precision floating point.
40368 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
40369 target macros to say whether we need to avoid SUBREGs mixing
40371 (TARGET_ALLOW_SF_SUBREG): Likewise.
40372 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
40373 (UNSPEC_SI_FROM_SF): Likewise.
40374 (iorxor): Change spacing.
40375 (and_ior_xor): New iterator for AND, IOR, and XOR.
40376 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
40377 (movdi_from_sf_zero_ext): Likewise.
40378 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
40379 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
40380 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
40381 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
40382 (fms<mode>4): Likewise.
40383 (fnma<mode>4): Likewise.
40384 (fnms<mode>4): Likewise.
40385 (nfma<mode>4): Likewise.
40386 (nfms<mode>4): Likewise.
40388 2017-01-04 Marek Polacek <polacek@redhat.com>
40391 * doc/invoke.texi: Document -Wpointer-compare.
40393 2017-01-04 Jakub Jelinek <jakub@redhat.com>
40395 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
40398 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
40399 descriptions for -gdwarf-5 and emit them as uleb128 instead of
40402 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
40405 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
40406 documentation of the powerpc_popcntb_ok attribute.
40407 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
40408 code to issue warning messages if a requested CPU configuration is
40409 not supported by the binary (assembler and loader) toolchain.
40410 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
40411 made to define a built-in function that has been disabled.
40412 (paired_init_builtins): Add assertion to prevent ICE if attempt is
40413 made to define a built-in function that has been disabled.
40414 (altivec_init_builtins): Add comment explaining why definition
40415 of the DST built-in functions is not preceded by an assertion
40416 check. Add assertions to prevent ICE if attempts are made to
40417 define an altivec predicate or an abs* built-in function that has
40419 (htm_init_builtins): Add comment explaining why definition of the
40420 htm built-in functions is not preceded by an assertion check.
40422 2017-01-04 Jeff Law <law@redhat.com>
40424 PR tree-optimizatin/67955
40425 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
40426 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
40427 the points-to solution does not include pt_null. Use DECL_PT_UID
40430 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
40432 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
40433 Use gen_int_mode instead of gen_lopwart for const_int operands.
40435 2017-01-04 Jakub Jelinek <jakub@redhat.com>
40437 PR tree-optimization/71563
40438 * match.pd: Simplify X << Y into X if Y is known to be 0 or
40439 out of range value - has low bits known to be zero.
40441 2017-01-04 Alan Modra <amodra@gmail.com>
40443 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
40444 * configure: Regenerate.
40445 * config.in: Regenerate.
40447 2017-01-04 Jakub Jelinek <jakub@redhat.com>
40450 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
40451 a substring of the message, but strcmp with the whole message. Ifdef
40452 ENABLE_NLS, translate the message first using dgettext.
40454 2017-01-03 Jeff Law <law@redhat.com>
40456 PR tree-optimizatin/78856
40457 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
40458 (mark_threaded_blocks): Remove code to truncate thread paths that
40459 cross multiple loop headers. Instead invalidate the cached loop
40460 iteration information and handle case of a thread path walking
40461 into an irreducible region.
40463 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
40466 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
40467 assertions. Add support for doing the signbit if the IEEE 128-bit
40468 floating point value is in a GPR.
40469 * config/rs6000/rs6000.md (Fsignbit): Delete.
40470 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
40471 Update the length attribute if the value is in a GPR.
40472 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
40473 the sign or zero extension instruction, since the value is always 0/1.
40474 (signbit<mode>2_dm2): Delete using <Fsignbit>.
40477 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
40478 extracting SImode to a GPR register so that we can generate a
40479 store, limit the vector to be in a traditional Altivec register
40480 for the vextuwrx instruction.
40482 2017-01-03 Ian Lance Taylor <iant@google.com>
40484 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
40486 2017-01-03 Martin Sebor <msebor@redhat.com>
40488 PR tree-optimization/78696
40489 * gimple-ssa-sprintf.c (format_floating): Correct handling of
40490 precision. Use MPFR for %f for greater fidelity. Correct handling
40492 (pass_sprintf_length::compute_format_length): Set width and precision
40493 specified by asrerisk to void_node for vararg functions.
40494 (try_substitute_return_value): Adjust dump output.
40496 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
40498 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
40500 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
40502 * doc/invoke.texi (SPARC options): Document -mlra as the default.
40503 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
40504 -mlra/-mno-lra was passed to the compiler.
40506 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
40508 PR rtl-optimization/65618
40509 * emit-rtl.c (try_split): Move initialization of "before" and
40510 "after" to just before the call to emit_insn_after_setloc.
40512 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
40514 * doc/md.texi (Standard Names): Remove reference to Java frontend.
40516 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
40518 * dwarf2out.c (gen_enumeration_type_die): When
40519 -gno-strict-dwarf, add a DW_AT_encoding attribute.
40521 2017-01-03 Jakub Jelinek <jakub@redhat.com>
40523 PR tree-optimization/78965
40524 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
40525 Change first argument from const call_info & to call_info &. For %n
40526 set info.nowrite to false.
40528 PR middle-end/78901
40529 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
40530 possibly throwing calls.
40532 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
40533 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
40534 and fns handling, rather than in a separate case SSA_NAME.
40536 2017-01-02 Jeff Law <law@redhat.com>
40538 * config/darwin-driver.c (darwin_driver_init): Const-correctness
40539 fixes for first_period and second_period variables.
40541 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
40544 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
40545 (*insvqi_1): New insn pattern.
40546 (*insvqi_1_mem_rex64): Ditto.
40547 (*insvqi_2): Ditto.
40548 (*insvqi_3): Rename from *insvqi.
40550 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
40552 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
40554 * doc/cfg.texi (Edges): Remove reference to Java.
40555 (Maintaining the CFG): Ditto.
40557 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40559 PR middle-end/77674
40560 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
40561 transparent aliases.
40563 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40565 PR middle-end/77484
40566 * predict.def (PRED_CALL): Update hitrate.
40567 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
40568 * predict.c (tree_estimate_probability_bb): Split CALL predictor
40569 into direct/indirect/polymorphic variants.
40571 2017-01-01 Jakub Jelinek <jakub@redhat.com>
40573 Update copyright years.
40575 * gcc.c (process_command): Update copyright notice dates.
40576 * gcov-dump.c (print_version): Ditto.
40577 * gcov.c (print_version): Ditto.
40578 * gcov-tool.c (print_version): Ditto.
40579 * gengtype.c (create_file): Ditto.
40580 * doc/cpp.texi: Bump @copying's copyright year.
40581 * doc/cppinternals.texi: Ditto.
40582 * doc/gcc.texi: Ditto.
40583 * doc/gccint.texi: Ditto.
40584 * doc/gcov.texi: Ditto.
40585 * doc/install.texi: Ditto.
40586 * doc/invoke.texi: Ditto.
40588 Copyright (C) 2017 Free Software Foundation, Inc.
40590 Copying and distribution of this file, with or without modification,
40591 are permitted in any medium without royalty provided the copyright
40592 notice and this notice are preserved.